Part I. File Systems
The File Systems section provides information on the file system structure and maintenance, the Btrfs Technology Preview, and file systems that Red Hat fully supports: ext3, ext4, GFS2, XFS, NFS, and FS-Cache.
Btrfs is available as a Technology Preview feature in Red Hat Enterprise Linux 7 but has been deprecated since the Red Hat Enterprise Linux 7.4 release. It will be removed in a future major release of Red Hat Enterprise Linux.
For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.4 Release Notes.
XFS is the default file system in Red Hat Enterprise Linux 7 and Red Hat, and Red Hat recommends you to use XFS unless you have a strong reason to use another file system. For general information on common file systems and their properties, see the following Red Hat Knowledgebase article: How to Choose your Red Hat Enterprise Linux File System.
Chapter 2. File System Structure and Maintenance
The file system structure is the most basic level of organization in an operating system. The way an operating system interacts with its users, applications, and security model nearly always depends on how the operating system organizes files on storage devices. Providing a common file system structure ensures users and programs can access and write files.
File systems break files down into two logical categories:
Shareable and unsharable files
Shareable files can be accessed locally and by remote hosts. Unsharable files are only available locally.
Variable and static filesVariable files, such as documents, can be changed at any time. Static files, such as binaries, do not change without an action from the system administrator.
Categorizing files in this manner helps correlate the function of each file with the permissions assigned to the directories which hold them. How the operating system and its users interact with a file determines the directory in which it is placed, whether that directory is mounted with read-only or read and write permissions, and the level of access each user has to that file. The top level of this organization is crucial; access to the underlying directories can be restricted, otherwise security problems could arise if, from the top level down, access rules do not adhere to a rigid structure.
2.1. Overview of Filesystem Hierarchy Standard (FHS)
Red Hat Enterprise Linux uses the Filesystem Hierarchy Standard (FHS) file system structure, which defines the names, locations, and permissions for many file types and directories.
The FHS document is the authoritative reference to any FHS-compliant file system, but the standard leaves many areas undefined or extensible. This section is an overview of the standard and a description of the parts of the file system not covered by the standard.
The two most important elements of FHS compliance are:
Compatibility with other FHS-compliant systems
The ability to mount a /usr/ partition as read-only. This is crucial, since /usr/ contains common executables and should not be changed by users. In addition, since /usr/ is mounted as read-only, it should be mountable from the CD-ROM drive or from another machine via a read-only NFS mount.
2.1.1. FHS Organization
The directories and files noted here are a small subset of those specified by the FHS document. For the most complete information, see the latest FHS documentation at //refspecs.linuxfoundation.org/FHS_3.0/fhs-3.0.pdf; the file-hierarchy(7) man page also provides an overview.
The directories that are available depend on what is installed on any given system. The following lists are only an example of what may be found.
2.1.1.1. Gathering File System Information
df Command
The df command reports the system's disk space usage. Its output looks similar to the following:
Example 2.1. df Command Output
Filesystem 1K-blocks Used Available Use% Mounted on /dev/mapper/VolGroup00-LogVol00 11675568 6272120 4810348 57% / /dev/sda1 100691 9281 86211 10% /boot none 322856 0 322856 0% /dev/shm
By default, df shows the partition size in 1 kilobyte blocks and the amount of used and available disk space in kilobytes. To view the information in megabytes and gigabytes, use the command df -h. The -h argument stands for "human-readable" format. The output for df -h looks similar to the following:
Example 2.2. df -h Command Output
Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 12G 6.0G 4.6G 57% / /dev/sda1 99M 9.1M 85M 10% /boot none 316M 0 316M 0% /dev/shm
In the given examples, the mounted partition /dev/shm represents the system's virtual memory file system.
du CommandThe du command displays the estimated amount of space being used by files in a directory, displaying the disk usage of each subdirectory. The last line in the output of du shows the total disk usage of the directory. To see only the total disk usage of a directory in human-readable format, use du -hs. For more options, see man du.
Gnome System MonitorTo view the system's partitions and disk space usage in a graphical format, use the Gnome System Monitor by clicking on → → or using the command gnome-system-monitor. Select the File Systems tab to view the system's partitions. The following figure illustrates the File Systems tab.
Figure 2.1. File Systems Tab in GNOME System Monitor
2.1.1.2. The /boot/ Directory
The /boot/ directory contains static files required to boot the system, for example, the Linux kernel. These files are essential for the system to boot properly.
Do not remove the /boot/ directory. Doing so renders the system unbootable.
2.1.1.3. The /dev/ Directory
The /dev/ directory contains device nodes that represent the following device types:
devices attached to the system;
virtual devices provided by the kernel.
These device nodes are essential for the system to function properly. The udevd daemon creates and removes device nodes in /dev/ as needed.
Devices in the /dev/ directory and subdirectories are defined as either character (providing only a serial stream of input and output, for example, mouse or keyboard) or block (accessible randomly, such as a hard drive or a floppy drive). If GNOME or KDE is installed, some storage devices are automatically detected when connected (such as with USB) or inserted (such as a CD or DVD drive), and a pop-up window displaying the contents appears.
Table 2.1. Examples of Common Files in the /dev Directory
/dev/hda | The master device on the primary IDE channel. |
/dev/hdb | The slave device on the primary IDE channel. |
/dev/tty0 | The first virtual console. |
/dev/tty1 | The second virtual console. |
/dev/sda | The first device on the primary SCSI or SATA channel. |
/dev/lp0 | The first parallel port. |
A valid block device can be one of two types of entries:
Mapped device
A logical volume in a volume group, for example, /dev/mapper/VolGroup00-LogVol02.
Static deviceA traditional storage volume, for example, /dev/sdbX, where sdb is a storage device name and X is the partition number. /dev/sdbX can also be /dev/disk/by-id/WWID, or /dev/disk/by-uuid/UUID. For more information, see Section 25.8, “Persistent Naming”.
2.1.1.4. The /etc/ Directory
The /etc/ directory is reserved for configuration files that are local to the machine. It should not contain any binaries; if there are any binaries, move them to /usr/bin/ or /usr/sbin/.
For example, the /etc/skel/ directory stores "skeleton" user files, which are used to populate a home directory when a user is first created. Applications also store their configuration files in this directory and may reference them when executed. The /etc/exports file controls which file systems export to remote hosts.
2.1.1.5. The /mnt/ Directory
The /mnt/ directory is reserved for temporarily mounted file systems, such as NFS file system mounts. For all removable storage media, use the /media/ directory. Automatically detected removable media is mounted in the /media directory.
The /mnt directory must not be used by installation programs.
2.1.1.6. The /opt/ Directory
The /opt/ directory is normally reserved for software and add-on packages that are not part of the default installation. A package that installs to /opt/ creates a directory bearing its name, for example, /opt/packagename/. In most cases, such packages follow a predictable subdirectory structure; most store their binaries in /opt/packagename/bin/ and their man pages in /opt/packagename/man/.
2.1.1.7. The /proc/ Directory
The /proc/ directory contains special files that either extract information from the kernel or send information to it. Examples of such information include system memory, CPU information, and hardware configuration. For more information about /proc/, see Section 2.3, “The /proc Virtual File System”.
2.1.1.8. The /srv/ Directory
The /srv/ directory contains site-specific data served by a Red Hat Enterprise Linux system. This directory gives users the location of data files for a particular service, such as FTP, WWW, or CVS. Data that only pertains to a specific user should go in the /home/ directory.
2.1.1.9. The /sys/ Directory
The /sys/ directory utilizes the new sysfs virtual file system specific to the kernel. With the increased support for hot plug hardware devices in the kernel, the /sys/ directory contains information similar to that held by /proc/, but displays a hierarchical view of device information specific to hot plug devices.
2.1.1.10. The /usr/ Directory
The /usr/ directory is for files that can be shared across multiple machines. The /usr/ directory is often on its own partition and is mounted read-only. At a minimum, /usr/ should contain the following subdirectories:
/usr/bin
This directory is used for binaries.
/usr/etcThis directory is used for system-wide configuration files.
/usr/gamesThis directory stores games.
/usr/includeThis directory is used for C header files.
/usr/kerberosThis directory is used for Kerberos-related binaries and files.
/usr/libThis directory is used for object files and libraries that are not designed to be directly utilized by shell scripts or users.
As of Red Hat Enterprise Linux 7.0, the /lib/ directory has been merged with /usr/lib. Now it also contains libraries needed to execute the binaries in /usr/bin/ and /usr/sbin/. These shared library images are used to boot the system or execute commands within the root file system.
/usr/libexecThis directory contains small helper programs called by other programs.
/usr/sbinAs of Red Hat Enterprise Linux 7.0, /sbin has been moved to /usr/sbin. This means that it contains all system administration binaries, including those essential for booting, restoring, recovering, or repairing the system. The binaries in /usr/sbin/ require root privileges to use.
/usr/shareThis directory stores files that are not architecture-specific.
/usr/srcThis directory stores source code.
/usr/tmp linked to /var/tmpThis directory stores temporary files.
The /usr/ directory should also contain a /local/ subdirectory. As per the FHS, this subdirectory is used by the system administrator when installing software locally, and should be safe from being overwritten during system updates. The /usr/local directory has a structure similar to /usr/, and contains the following subdirectories:
/usr/local/bin
/usr/local/etc
/usr/local/games
/usr/local/include
/usr/local/lib
/usr/local/libexec
/usr/local/sbin
/usr/local/share
/usr/local/src
Red Hat Enterprise Linux's usage of /usr/local/ differs slightly from the FHS. The FHS states that /usr/local/ should be used to store software that should remain safe from system software upgrades. Since the RPM Package Manager can perform software upgrades safely, it is not necessary to protect files by storing them in /usr/local/.
Instead, Red Hat Enterprise Linux uses /usr/local/ for software local to the machine. For instance, if the /usr/ directory is mounted as a read-only NFS share from a remote host, it is still possible to install a package or program under the /usr/local/ directory.
2.1.1.11. The /var/ Directory
Since the FHS requires Linux to mount /usr/ as read-only, any programs that write log files or need spool/ or lock/ directories should write them to the /var/ directory. The FHS states /var/ is for variable data, which includes spool directories and files, logging data, transient and temporary files.
Following are some of the directories found within the /var/ directory:
/var/account/
/var/arpwatch/
/var/cache/
/var/crash/
/var/db/
/var/empty/
/var/ftp/
/var/gdm/
/var/kerberos/
/var/lib/
/var/local/
/var/lock/
/var/log/
/var/mail linked to /var/spool/mail/
/var/mailman/
/var/named/
/var/nis/
/var/opt/
/var/preserve/
/var/run/
/var/spool/
/var/tmp/
/var/tux/
/var/www/
/var/yp/
The /var/run/media/user directory contains subdirectories used as mount points for removable media such as USB storage media, DVDs, CD-ROMs, and Zip disks. Note that previously, the /media/ directory was used for this purpose.
System log files, such as messages and lastlog, go in the /var/log/ directory. The /var/lib/rpm/ directory contains RPM system databases. Lock files go in the /var/lock/ directory, usually in directories for the program using the file. The /var/spool/ directory has subdirectories that store data files for some programs. These subdirectories include:
/var/spool/at/
/var/spool/clientmqueue/
/var/spool/cron/
/var/spool/cups/
/var/spool/exim/
/var/spool/lpd/
/var/spool/mail/
/var/spool/mailman/
/var/spool/mqueue/
/var/spool/news/
/var/spool/postfix/
/var/spool/repackage/
/var/spool/rwho/
/var/spool/samba/
/var/spool/squid/
/var/spool/squirrelmail/
/var/spool/up2date/
/var/spool/uucp/
/var/spool/uucppublic/
/var/spool/vbox/
2.2. Special Red Hat Enterprise Linux File Locations
Red Hat Enterprise Linux extends the FHS structure slightly to accommodate special files.
Most files pertaining to RPM are kept in the /var/lib/rpm/ directory. For more information on RPM, see man rpm.
The /var/cache/yum/ directory contains files used by the Package Updater, including RPM header information for the system. This location may also be used to temporarily store RPMs downloaded while updating the system. For more information about the Red Hat Network, see //rhn.redhat.com/.
Another location specific to Red Hat Enterprise Linux is the /etc/sysconfig/ directory. This directory stores a variety of configuration information. Many scripts that run at boot time use the files in this directory.
2.3. The /proc Virtual File System
Unlike most file systems, /proc contains neither text nor binary files. Because it houses virtual files, the /proc is referred to as a virtual file system. These virtual files are typically zero bytes in size, even if they contain a large amount of information.
The /proc file system is not used for storage. Its main purpose is to provide a file-based interface to hardware, memory, running processes, and other system components. Real-time information can be retrieved on many system components by viewing the corresponding /proc file. Some of the files within /proc can also be manipulated (by both users and applications) to configure the kernel.
The following /proc files are relevant in managing and monitoring system storage:
/proc/devices
Displays various character and block devices that are currently configured.
/proc/filesystemsLists all file system types currently supported by the kernel.
/proc/mdstatContains current information on multiple-disk or RAID configurations on the system, if they exist.
/proc/mountsLists all mounts currently used by the system.
/proc/partitionsContains partition block allocation information.
For more information about the /proc file system, see the Red Hat Enterprise Linux 7 Deployment Guide.
2.4. Discard Unused Blocks
Batch discard and online discard operations are features of mounted file systems that discard blocks not in use by the file system. They are useful for both solid-state drives and thinly-provisioned storage.
Batch discard operations are run explicitly by the user with the fstrim command. This command discards all unused blocks in a file system that match the user's criteria.
Online discard operations are specified at mount time, either with the -o discard option as part of a mount command or with the discard option in the /etc/fstab file. They run in real time without user intervention. Online discard operations only discard blocks that are transitioning from used to free.
Both operation types are supported for use with ext4 file systems as of Red Hat Enterprise Linux 6.2 and later and with XFS file systems since Red Hat Enterprise Linux 6.4. Also, the block device underlying the file system must support physical discard operations. Physical discard operations are supported if the value stored in the /sys/block/device/queue/discard_max_bytes file is not zero.
If you are executing the fstrim command on:
a device that does not support discard operations, or
a logical device (LVM or MD) comprised of multiple devices, where any one of the device does not support discard operations
the following message will be displayed:
fstrim -v /mnt/non_discard fstrim: /mnt/non_discard: the discard operation is not supportedThe mount command allows you to mount a device that does not support discard operations with the -o discard option.
Red Hat recommends batch discard operations unless the system's workload is such that batch discard is not feasible, or online discard operations are necessary to maintain performance.
For more information, see the fstrim(8) and mount(8) man pages.
Chapter 3. The XFS File System
XFS is a highly scalable, high-performance file system which was originally designed at Silicon Graphics, Inc. XFS is the default file system for Red Hat Enterprise Linux 7.
Main Features of XFS
XFS supports metadata journaling, which facilitates quicker crash recovery.
The XFS file system can be defragmented and enlarged while mounted and active.
In addition, Red Hat Enterprise Linux 7 supports backup and restore utilities specific to XFS.
XFS features the following allocation schemes:
Extent-based allocation
-
Stripe-aware allocation policies
Delayed allocation
Space pre-allocation
Delayed allocation and other performance optimizations affect XFS the same way that they do ext4. Namely, a program's writes to an XFS file system are not guaranteed to be on-disk unless the program issues an fsync() call afterwards.
Creating or expanding files occasionally fails with an unexpected ENOSPC write failure even though the disk space appears to be sufficient. This is due to XFS's performance-oriented design. In practice, it does not become a problem since it only occurs if remaining space is only a few blocks.
Other XFS FeaturesThe XFS file system also supports the following:
Extended attributes (xattr)
This allows the system to associate several additional name/value pairs per file. It is enabled by default.
Quota journalingThis avoids the need for lengthy quota consistency checks after a crash.
Project/directory quotasThis allows quota restrictions over a directory tree.
Subsecond timestampsThis allows timestamps to go to the subsecond.
Default atime behavior is relatimeRelatime is on by default for XFS. It has almost no overhead compared to noatime while still maintaining sane atime values.
3.1. Creating an XFS File System
To create an XFS file system, use the following command:
# mkfs.xfs block_deviceReplace block_device with the path to a block device. For example, /dev/sdb1, /dev/disk/by-uuid/05e99ec8-def1-4a5e-8a9d-5945339ceb2a, or /dev/my-volgroup/my-lv.
In general, the default options are optimal for common use.
When using mkfs.xfs on a block device containing an existing file system, add the -f option to overwrite that file system.
Example 3.1. mkfs.xfs Command Output
Following is a sample output of the mkfs.xfs command:
meta-data=/dev/device isize=256 agcount=4, agsize=3277258 blks = sectsz=512 attr=2 data = bsize=4096 blocks=13109032, imaxpct=25 = sunit=0 swidth=0 blks naming =version 2 bsize=4096 ascii-ci=0 log =internal log bsize=4096 blocks=6400, version=2 = sectsz=512 sunit=0 blks, lazy-count=1 realtime =none extsz=4096 blocks=0, rtextents=0Striped Block Devices
For striped block devices (for example, RAID5 arrays), the stripe geometry can be specified at the time of file system creation. Using proper stripe geometry greatly enhances the performance of an XFS filesystem.
When creating filesystems on LVM or MD volumes, mkfs.xfs chooses an optimal geometry. This may also be true on some hardware RAIDs that export geometry information to the operating system.
If the device exports stripe geometry information, the mkfs utility (for ext3, ext4, and xfs) will automatically use this geometry. If stripe geometry is not detected by the mkfs utility and even though the storage does, in fact, have stripe geometry, it is possible to manually specify it when creating the file system using the following options:
su=value
Specifies a stripe unit or RAID chunk size. The value must be specified in bytes, with an optional k, m, or g suffix.
sw=valueSpecifies the number of data disks in a RAID device, or the number of stripe units in the stripe.
The following example specifies a chunk size of 64k on a RAID device containing 4 stripe units:
# mkfs.xfs -d su=64k,sw=4 /dev/block_deviceAdditional Resources
For more information about creating XFS file systems, see:
The mkfs.xfs(8) man page
3.2. Mounting an XFS File System
An XFS file system can be mounted with no extra options, for example:
# mount /dev/device /mount/pointThe default for Red Hat Enterprise Linux 7 is inode64.
Unlike mke2fs, mkfs.xfs does not utilize a configuration file; they are all specified on the command line.
Write Barriers
By default, XFS uses write barriers to ensure file system integrity even when power is lost to a device with write caches enabled. For devices without write caches, or with battery-backed write caches, disable the barriers by using the nobarrier option:
# mount -o nobarrier /dev/device /mount/pointDirect Access Technology Preview
Since Red Hat Enterprise Linux 7.3, Direct Access (DAX) is available as a Technology Preview on the ext4 and XFS file systems. It is a means for an application to directly map persistent memory into its address space. To use DAX, a system must have some form of persistent memory available, usually in the form of one or more Non-Volatile Dual Inline Memory Modules (NVDIMMs), and a file system that supports DAX must be created on the NVDIMM(s). Also, the file system must be mounted with the dax mount option. Then, an mmap of a file on the dax-mounted file system results in a direct mapping of storage into the application's address space.
3.3. XFS Quota Management
The XFS quota subsystem manages limits on disk space (blocks) and file (inode) usage. XFS quotas control or report on usage of these items on a user, group, or directory or project level. Also, note that while user, group, and directory or project quotas are enabled independently, group and project quotas are mutually exclusive.
When managing on a per-directory or per-project basis, XFS manages the disk usage of directory hierarchies associated with a specific project. In doing so, XFS recognizes cross-organizational "group" boundaries between projects. This provides a level of control that is broader than what is available when managing quotas for users or groups.
XFS quotas are enabled at mount time, with specific mount options. Each mount option can also be specified as noenforce; this allows usage reporting without enforcing any limits. Valid quota mount options are:
uquota/uqnoenforce: User quotas
gquota/gqnoenforce: Group quotas
pquota/pqnoenforce: Project quota
Once quotas are enabled, the xfs_quota tool can be used to set limits and report on disk usage. By default, xfs_quota is run interactively, and in basic mode. Basic mode subcommands simply report usage, and are available to all users. Basic xfs_quota subcommands include:
quota username/userID
Show usage and limits for the given username or numeric userID
dfShows free and used counts for blocks and inodes.
In contrast, xfs_quota also has an expert mode. The subcommands of this mode allow actual configuration of limits, and are available only to users with elevated privileges. To use expert mode subcommands interactively, use the following command:
# xfs_quota -xExpert mode subcommands include:
report /path
Reports quota information for a specific file system.
limitModify quota limits.
For a complete list of subcommands for either basic or expert mode, use the subcommand help.
All subcommands can also be run directly from a command line using the -c option, with -x for expert subcommands.
Example 3.2. Display a Sample Quota Report
For example, to display a sample quota report for /home (on /dev/blockdevice), use the command xfs_quota -x -c 'report -h' /home. This displays output similar to the following:
User quota on /home (/dev/blockdevice) Blocks User ID Used Soft Hard Warn/Grace ---------- --------------------------------- root 0 0 0 00 [------] testuser 103.4G 0 0 00 [------] ...To set a soft and hard inode count limit of 500 and 700 respectively for user john, whose home directory is /home/john, use the following command:
# xfs_quota -x -c 'limit isoft=500 ihard=700 john' /home/In this case, pass mount_point which is the mounted xfs file system.
By default, the limit subcommand recognizes targets as users. When configuring the limits for a group, use the -g option (as in the previous example). Similarly, use -p for projects.
Soft and hard block limits can also be configured using bsoft or bhard instead of isoft or ihard.
Example 3.3. Set a Soft and Hard Block Limit
For example, to set a soft and hard block limit of 1000m and 1200m, respectively, to group accounting on the /target/path file system, use the following command:
# xfs_quota -x -c 'limit -g bsoft=1000m bhard=1200m accounting' /target/pathThe commands bsoft and bhard count by the byte.
While real-time blocks (rtbhard/rtbsoft) are described in man xfs_quota as valid units when setting quotas, the real-time sub-volume is not enabled in this release. As such, the rtbhard and rtbsoft options are not applicable.
Setting Project Limits
With XFS file system, you can set quotas on individual directory hierarchies in the file system that are known as managed trees. Each managed tree is uniquely identified by a project ID and an optional project name.
Add the project-controlled directories to /etc/projects. For example, the following adds the /var/log path with a unique ID of 11 to /etc/projects. Your project ID can be any numerical value mapped to your project.
# echo 11:/var/log >> /etc/projectsAdd project names to /etc/projid to map project IDs to project names. For example, the following associates a project called Logs with the project ID of 11 as defined in the previous step.
# echo Logs:11 >> /etc/projidInitialize the project directory. For example, the following initializes the project directory /var:
# xfs_quota -x -c 'project -s logfiles' /varConfigure quotas for projects with initialized directories:
# xfs_quota -x -c 'limit -p bhard=lg logfiles' /var
Generic quota configuration tools (quota, repquota, and edquota for example) may also be used to manipulate XFS quotas. However, these tools cannot be used with XFS project quotas.
Red Hat recommends the use of xfs_quota over all other available tools.
For more information about setting XFS quotas, see man xfs_quota, man projid(5), and man projects(5).
3.4. Increasing the Size of an XFS File System
An XFS file system may be grown while mounted using the xfs_growfs command:
# xfs_growfs /mount/point -D sizeThe -D size option grows the file system to the specified size (expressed in file system blocks). Without the -D size option, xfs_growfs will grow the file system to the maximum size supported by the device.
Before growing an XFS file system with -D size, ensure that the underlying block device is of an appropriate size to hold the file system later. Use the appropriate resizing methods for the affected block device.
While XFS file systems can be grown while mounted, their size cannot be reduced at all.
For more information about growing a file system, see man xfs_growfs.
3.5. Repairing an XFS File System
To repair an XFS file system, use xfs_repair:
# xfs_repair /dev/deviceThe xfs_repair utility is highly scalable and is designed to repair even very large file systems with many inodes efficiently. Unlike other Linux file systems, xfs_repair does not run at boot time, even when an XFS file system was not cleanly unmounted. In the event of an unclean unmount, xfs_repair simply replays the log at mount time, ensuring a consistent file system.
The xfs_repair utility cannot repair an XFS file system with a dirty log. To clear the log, mount and unmount the XFS file system. If the log is corrupt and cannot be replayed, use the -L option ("force log zeroing") to clear the log, that is, xfs_repair -L /dev/device. Be aware that this may result in further corruption or data loss.
For more information about repairing an XFS file system, see man xfs_repair.
3.6. Suspending an XFS File System
To suspend or resume write activity to a file system, use the following command:
# xfs_freeze mount-pointSuspending write activity allows hardware-based device snapshots to be used to capture the file system in a consistent state.
The xfs_freeze utility is provided by the xfsprogs package, which is only available on x86_64.
To suspend (that is, freeze) an XFS file system, use:
# xfs_freeze -f /mount/pointTo unfreeze an XFS file system, use:
# xfs_freeze -u /mount/pointWhen taking an LVM snapshot, it is not necessary to use xfs_freeze to suspend the file system first. Rather, the LVM management tools will automatically suspend the XFS file system before taking the snapshot.
For more information about freezing and unfreezing an XFS file system, see man xfs_freeze.
3.7. Backing Up and Restoring XFS File Systems
XFS file system backup and restoration involve these utilities:
xfsdump for creating the backup
xfsrestore for restoring from backup
3.7.1. Features of XFS Backup and Restoration
Backup
You can use the xfsdump utility to:
Perform backups to regular file images.
Only one backup can be written to a regular file.
Perform backups to tape drives.
The xfsdump utility also allows you to write multiple backups to the same tape. A backup can span multiple tapes.
To back up multiple file systems to a single tape device, simply write the backup to a tape that already contains an XFS backup. This appends the new backup to the previous one. By default, xfsdump never overwrites existing backups.
Create incremental backups.
The xfsdump utility uses dump levels to determine a base backup to which other backups are relative. Numbers from 0 to 9 refer to increasing dump levels. An incremental backup only backs up files that have changed since the last dump of a lower level:
To perform a full backup, perform a level 0 dump on the file system.
A level 1 dump is the first incremental backup after a full backup. The next incremental backup would be level 2, which only backs up files that have changed since the last level 1 dump; and so on, to a maximum of level 9.
Exclude files from a backup using size, subtree, or inode flags to filter them.
Restoration
The xfsrestore utility restores file systems from backups produced by xfsdump. The xfsrestore utility has two modes:
The simple mode enables users to restore an entire file system from a level 0 dump. This is the default mode.
The cumulative mode enables file system restoration from an incremental backup: that is, level 1 to level 9.
A unique session ID or session label identifies each backup. Restoring a backup from a tape containing multiple backups requires its corresponding session ID or label.
To extract, add, or delete specific files from a backup, enter the xfsrestore interactive mode. The interactive mode provides a set of commands to manipulate the backup files.
3.7.2. Backing Up an XFS File System
This procedure describes how to back up the content of an XFS file system into a file or a tape.
Procedure 3.1. Backing Up an XFS File System
Use the following command to back up an XFS file system:
# xfsdump -l level [-L label] -f backup-destination path-to-xfs-filesystemReplace level with the dump level of your backup. Use 0 to perform a full backup or 1 to 9 to perform consequent incremental backups.
Replace backup-destination with the path where you want to store your backup. The destination can be a regular file, a tape drive, or a remote tape device. For example, /backup-files/Data.xfsdump for a file or /dev/st0 for a tape drive.
Replace path-to-xfs-filesystem with the mount point of the XFS file system you want to back up. For example, /mnt/data/. The file system must be mounted.
When backing up multiple file systems and saving them on a single tape device, add a session label to each backup using the -L label option so that it is easier to identify them when restoring. Replace label with any name for your backup: for example, backup_data.
Example 3.4. Backing up Multiple XFS File Systems
To back up the content of XFS file systems mounted on the /boot/ and /data/ directories and save them as files in the /backup-files/ directory:
# xfsdump -l 0 -f /backup-files/boot.xfsdump /boot # xfsdump -l 0 -f /backup-files/data.xfsdump /dataTo back up multiple file systems on a single tape device, add a session label to each backup using the -L label option:
# xfsdump -l 0 -L "backup_boot" -f /dev/st0 /boot # xfsdump -l 0 -L "backup_data" -f /dev/st0 /data
Additional Resources
For more information about backing up XFS file systems, see the xfsdump(8) man page.
3.7.3. Restoring an XFS File System from Backup
This procedure describes how to restore the content of an XFS file system from a file or tape backup.
Prerequisites
Procedure 3.2. Restoring an XFS File System from Backup
The command to restore the backup varies depending on whether you are restoring from a full backup or an incremental one, or are restoring multiple backups from a single tape device:
# xfsrestore [-r] [-S session-id] [-L session-label] [-i] -f backup-location restoration-pathReplace backup-location with the location of the backup. This can be a regular file, a tape drive, or a remote tape device. For example, /backup-files/Data.xfsdump for a file or /dev/st0 for a tape drive.
Replace restoration-path with the path to the directory where you want to restore the file system. For example, /mnt/data/.
To restore a file system from an incremental (level 1 to level 9) backup, add the -r option.
To restore a backup from a tape device that contains multiple backups, specify the backup using the -S or -L options.
The -S lets you choose a backup by its session ID, while the -L lets you choose by the session label. To obtain the session ID and session labels, use the xfsrestore -I command.
Replace session-id with the session ID of the backup. For example, b74a3586-e52e-4a4a-8775-c3334fa8ea2c. Replace session-label with the session label of the backup. For example, my_backup_session_label.
To use xfsrestore interactively, use the -i option.
The interactive dialog begins after xfsrestore finishes reading the specified device. Available commands in the interactive xfsrestore shell include cd, ls, add, delete, and extract; for a complete list of commands, use the help command.
Example 3.5. Restoring Multiple XFS File Systems
To restore the XFS backup files and save their content into directories under /mnt/:
# xfsrestore -f /backup-files/boot.xfsdump /mnt/boot/ # xfsrestore -f /backup-files/data.xfsdump /mnt/data/To restore from a tape device containing multiple backups, specify each backup by its session label or session ID:
# xfsrestore -f /dev/st0 -L "backup_boot" /mnt/boot/ # xfsrestore -f /dev/st0 -S "45e9af35-efd2-4244-87bc-4762e476cbab" /mnt/data/Informational Messages When Restoring a Backup from a Tape
When restoring a backup from a tape with backups from multiple file systems, the xfsrestore utility might issue messages. The messages inform you whether a match of the requested backup has been found when xfsrestore examines each backup on the tape in sequential order. For example:
xfsrestore: preparing drive xfsrestore: examining media file 0 xfsrestore: inventory session uuid (8590224e-3c93-469c-a311-fc8f23029b2a) does not match the media header's session uuid (7eda9f86-f1e9-4dfd-b1d4-c50467912408) xfsrestore: examining media file 1 xfsrestore: inventory session uuid (8590224e-3c93-469c-a311-fc8f23029b2a) does not match the media header's session uuid (7eda9f86-f1e9-4dfd-b1d4-c50467912408) [...]The informational messages keep appearing until the matching backup is found.
Additional Resources
For more information about restoring XFS file systems, see the xfsrestore(8) man page.
3.8. Configuring Error Behavior
When an error occurs during an I/O operation, the XFS driver responds in one of two ways:
Continue retries until either:
the I/O operation succeeds, or
an I/O operation retry count or time limit is exceeded.
Consider the error permanent and halt the system.
XFS currently recognizes the following error conditions for which you can configure the desired behavior specifically:
EIO: Error while trying to write to the device
ENOSPC: No space left on the device
ENODEV: Device cannot be found
All other possible error conditions, which do not have specific handlers defined, share a single, global configuration.
You can set the conditions under which XFS deems the errors permanent, both in the maximum number of retries and the maximum time in seconds. XFS stops retrying when any one of the conditions is met.
There is also an option to immediately cancel the retries when unmounting the file system, regardless of any other configuration. This allows the unmount operation to succeed even in case of persistent errors.
3.8.1. Configuration Files for Specific and Undefined Conditions
Configuration files controlling error behavior are located in the /sys/fs/xfs/device/error/ directory.
The /sys/fs/xfs/device/error/metadata/ directory contains subdirectories for each specific error condition:
/sys/fs/xfs/device/error/metadata/EIO/ for the EIO error condition
/sys/fs/xfs/device/error/metadata/ENODEV/ for the ENODEV error condition
/sys/fs/xfs/device/error/metadata/ENOSPC/ for the ENOSPC error condition
Each one then contains the following configuration files:
/sys/fs/xfs/device/error/metadata/condition/max_retries: controls the maximum number of times that XFS retries the operation.
/sys/fs/xfs/device/error/metadata/condition/retry_timeout_seconds: the time limit in seconds after which XFS will stop retrying the operation
All other possible error conditions, apart from those described in the previous section, share a common configuration in these files:
/sys/fs/xfs/device/error/metadata/default/max_retries: controls the maximum number of retries
/sys/fs/xfs/device/error/metadata/default/retry_timeout_seconds: controls the time limit for retrying
3.8.2. Setting File System Behavior for Specific and Undefined Conditions
To set the maximum number of retries, write the desired number to the max_retries file.
For specific conditions:
# echo value > /sys/fs/xfs/device/error/metadata/condition/max_retriesFor undefined conditions:
# echo value > /sys/fs/xfs/device/error/metadata/default/max_retries
value is a number between -1 and the maximum possible value of int, the C signed integer type. This is 2147483647 on 64-bit Linux.
To set the time limit, write the desired number of seconds to the retry_timeout_seconds file.
For specific conditions:
# echo value > /sys/fs/xfs/device/error/metadata/condition/retry_timeout_secondsFor undefined conditions:
# echo value > /sys/fs/xfs/device/error/metadata/default/retry_timeout_seconds
value is a number between -1 and 86400, which is the number of seconds in a day.
In both the max_retries and retry_timeout_seconds options, -1 means to retry forever and 0 to stop immediately.
device is the name of the device, as found in the /dev/ directory; for example, sda.
The default behavior for a each error condition is dependent on the error context. Some errors, like ENODEV, are considered to be fatal and unrecoverable, regardless of the retry count, so their default value is 0.
3.8.3. Setting Unmount Behavior
If the fail_at_unmount option is set, the file system overrides all other error configurations during unmount, and immediately umnounts the file system without retrying the I/O operation. This allows the unmount operation to succeed even in case of persistent errors.
To set the unmount behavior:
# echo value > /sys/fs/xfs/device/error/fail_at_unmountvalue is either 1 or 0:
1 means to cancel retrying immediately if an error is found.
0 means to respect the max_retries and retry_timeout_seconds options.
device is the name of the device, as found in the /dev/ directory; for example, sda.
The fail_at_unmount option has to be set as desired before attempting to unmount the file system. After an unmount operation has started, the configuration files and directories may be unavailable.
3.9. Other XFS File System Utilities
Red Hat Enterprise Linux 7 also features other utilities for managing XFS file systems:
xfs_fsr
Used to defragment mounted XFS file systems. When invoked with no arguments, xfs_fsr defragments all regular files in all mounted XFS file systems. This utility also allows users to suspend a defragmentation at a specified time and resume from where it left off later.
In addition, xfs_fsr also allows the defragmentation of only one file, as in xfs_fsr /path/to/file. Red Hat advises not to periodically defrag an entire file system because XFS avoids fragmentation by default. System wide defragmentation could cause the side effect of fragmentation in free space.
xfs_bmapPrints the map of disk blocks used by files in an XFS filesystem. This map lists each extent used by a specified file, as well as regions in the file with no corresponding blocks (that is, holes).
xfs_infoPrints XFS file system information.
xfs_adminChanges the parameters of an XFS file system. The xfs_admin utility can only modify parameters of unmounted devices or file systems.
xfs_copyCopies the contents of an entire XFS file system to one or more targets in parallel.
The following utilities are also useful in debugging and analyzing XFS file systems:
xfs_metadump
Copies XFS file system metadata to a file. Red Hat only supports using the xfs_metadump utility to copy unmounted file systems or read-only mounted file systems; otherwise, generated dumps could be corrupted or inconsistent.
xfs_mdrestoreRestores an XFS metadump image (generated using xfs_metadump) to a file system image.
xfs_dbDebugs an XFS file system.
For more information about these utilities, see their respective man pages.
3.10. Migrating from ext4 to XFS
Starting with Red Hat Enterprise Linux 7.0, XFS is the default file system instead of ext4. This section highlights the differences when using or administering an XFS file system.
The ext4 file system is still fully supported in Red Hat Enterprise Linux 7 and can be selected at installation. While it is possible to migrate from ext4 to XFS, it is not required.
3.10.1. Differences Between Ext3/4 and XFS
File system repair
Ext3/4 runs e2fsck in userspace at boot time to recover the journal as needed. XFS, by comparison, performs journal recovery in kernelspace at mount time. An fsck.xfs shell script is provided but does not perform any useful action as it is only there to satisfy initscript requirements.
When an XFS file system repair or check is requested, use the xfs_repair command. Use the -n option for a read-only check.
The xfs_repair command will not operate on a file system with a dirty log. To repair such a file system mount and unmount must first be performed to replay the log. If the log is corrupt and cannot be replayed, the -L option can be used to zero out in the log.
For more information on file system repair of XFS file systems, see Section 12.2.2, “XFS”
Metadata error behaviorThe ext3/4 file system has configurable behavior when metadata errors are encountered, with the default being to simply continue. When XFS encounters a metadata error that is not recoverable it will shut down the file system and return a EFSCORRUPTED error. The system logs will contain details of the error encountered and will recommend running xfs_repair if necessary.
QuotasXFS quotas are not a remountable option. The -o quota option must be specified on the initial mount for quotas to be in effect.
While the standard tools in the quota package can perform basic quota administrative tasks (tools such as setquota and repquota), the xfs_quota tool can be used for XFS-specific features, such as Project Quota administration.
The quotacheck command has no effect on an XFS file system. The first time quota accounting is turned on XFS does an automatic quotacheck internally. Because XFS quota metadata is a first-class, journaled metadata object, the quota system will always be consistent until quotas are manually turned off.
File system resizeThe XFS file system has no utility to shrink a file system. XFS file systems can be grown online via the xfs_growfs command.
Inode numbersFor file systems larger than 1 TB with 256-byte inodes, or larger than 2 TB with 512-byte inodes, XFS inode numbers might exceed 2^32. Such large inode numbers cause 32-bit stat calls to fail with the EOVERFLOW return value. The described problem might occur when using the default Red Hat Enterprise Linux 7 configuration: non-striped with four allocation groups. A custom configuration, for example file system extension or changing XFS file system parameters, might lead to a different behavior.
Applications usually handle such larger inode numbers correctly. If needed, mount the XFS file system with the -o inode32 parameter to enforce inode numbers below 2^32. Note that using inode32 does not affect inodes that are already allocated with 64-bit numbers.
Do not use the inode32 option unless it is required by a specific environment. The inode32 option changes allocation behavior. As a consequence, the ENOSPC error might occur if no space is available to allocate inodes in the lower disk blocks.
Speculative preallocationXFS uses speculative preallocation to allocate blocks past EOF as files are written. This avoids file fragmentation due to concurrent streaming write workloads on NFS servers. By default, this preallocation increases with the size of the file and will be apparent in "du" output. If a file with speculative preallocation is not dirtied for five minutes the preallocation will be discarded. If the inode is cycled out of cache before that time, then the preallocation will be discarded when the inode is reclaimed.
If premature ENOSPC problems are seen due to speculative preallocation, a fixed preallocation amount may be specified with the -o allocsize=amount mount option.
Fragmentation-related toolsFragmentation is rarely a significant issue on XFS file systems due to heuristics and behaviors, such as delayed allocation and speculative preallocation. However, tools exist for measuring file system fragmentation as well as defragmenting file systems. Their use is not encouraged.
The xfs_db frag command attempts to distill all file system allocations into a single fragmentation number, expressed as a percentage. The output of the command requires significant expertise to understand its meaning. For example, a fragmentation factor of 75% means only an average of 4 extents per file. For this reason the output of xfs_db's frag is not considered useful and more careful analysis of any fragmentation problems is recommended.
The xfs_fsr command may be used to defragment individual files, or all files on a file system. The later is especially not recommended as it may destroy locality of files and may fragment free space.
Commands Used with ext3 and ext4 Compared to XFS
The following table compares common commands used with ext3 and ext4 to their XFS-specific counterparts.
Table 3.1. Common Commands for ext3 and ext4 Compared to XFS
Create a file system | mkfs.ext4 or mkfs.ext3 | mkfs.xfs |
File system check | e2fsck | xfs_repair |
Resizing a file system | resize2fs | xfs_growfs |
Save an image of a file system | e2image | xfs_metadump and xfs_mdrestore |
Label or tune a file system | tune2fs | xfs_admin |
Backup a file system | dump and restore | xfsdump and xfsrestore |
The following table lists generic tools that function on XFS file systems as well, but the XFS versions have more specific functionality and as such are recommended.
Table 3.2. Generic Tools for ext4 and XFS
Quota | quota | xfs_quota |
File mapping | filefrag | xfs_bmap |
More information on many the listed XFS commands is included in Chapter 3, The XFS File System. You can also consult the manual pages of the listed XFS administration tools for more information.
Chapter 4. The ext3 File System
The ext3 file system is essentially an enhanced version of the ext2 file system. These improvements provide the following advantages:
Availability
After an unexpected power failure or system crash (also called an unclean system shutdown), each mounted ext2 file system on the machine must be checked for consistency by the e2fsck program. This is a time-consuming process that can delay system boot time significantly, especially with large volumes containing a large number of files. During this time, any data on the volumes is unreachable.
It is possible to run fsck -n on a live filesystem. However, it will not make any changes and may give misleading results if partially written metadata is encountered.
If LVM is used in the stack, another option is to take an LVM snapshot of the filesystem and run fsck on it instead.
Finally, there is the option to remount the filesystem as read only. All pending metadata updates (and writes) are then forced to the disk prior to the remount. This ensures the filesystem is in a consistent state, provided there is no previous corruption. It is now possible to run fsck -n.
The journaling provided by the ext3 file system means that this sort of file system check is no longer necessary after an unclean system shutdown. The only time a consistency check occurs using ext3 is in certain rare hardware failure cases, such as hard drive failures. The time to recover an ext3 file system after an unclean system shutdown does not depend on the size of the file system or the number of files; rather, it depends on the size of the journal used to maintain consistency. The default journal size takes about a second to recover, depending on the speed of the hardware.
The only journaling mode in ext3 supported by Red Hat is data=ordered (default).
Data IntegrityThe ext3 file system prevents loss of data integrity in the event that an unclean system shutdown occurs. The ext3 file system allows you to choose the type and level of protection that your data receives. With regard to the state of the file system, ext3 volumes are configured to keep a high level of data consistency by default.
SpeedDespite writing some data more than once, ext3 has a higher throughput in most cases than ext2 because ext3's journaling optimizes hard drive head motion. You can choose from three journaling modes to optimize speed, but doing so means trade-offs in regards to data integrity if the system was to fail.
The only journaling mode in ext3 supported by Red Hat is data=ordered (default).
Easy TransitionIt is easy to migrate from ext2 to ext3 and gain the benefits of a robust journaling file system without reformatting. For more information on performing this task, see Section 4.2, “Converting to an ext3 File System” .
Red Hat Enterprise Linux 7 provides a unified extN driver. It does this by disabling the ext2 and ext3 configurations and instead uses ext4.ko for these on-disk formats. This means that kernel messages will always refer to ext4 regardless of the ext file system used.
4.1. Creating an ext3 File System
After installation, it is sometimes necessary to create a new ext3 file system. For example, if a new disk drive is added to the system, you may want to partition the drive and use the ext3 file system.
Format the partition or LVM volume with the ext3 file system using the mkfs.ext3 utility:
# mkfs.ext3 block_deviceReplace block_device with the path to a block device. For example, /dev/sdb1, /dev/disk/by-uuid/05e99ec8-def1-4a5e-8a9d-5945339ceb2a, or /dev/my-volgroup/my-lv.
Label the file system using the e2label utility:
# e2label block_device volume_label
Configuring UUID
It is also possible to set a specific UUID for a file system. To specify a UUID when creating a file system, use the -U option:
# mkfs.ext3 -U UUID deviceReplace UUID with the UUID you want to set: for example, 7cd65de3-e0be-41d9-b66d-96d749c02da7.
Replace device with the path to an ext3 file system to have the UUID added to it: for example, /dev/sda8.
Additional Resources
The mkfs.ext3(8) man page
The e2label(8) man page
4.2. Converting to an ext3 File System
The tune2fs command converts an ext2 file system to ext3.
To convert ext2 to ext3, always use the e2fsck utility to check your file system before and after using tune2fs. Before trying to convert ext2 to ext3, back up all file systems in case any errors occur.
In addition, Red Hat recommends creating a new ext3 file system and migrating data to it, instead of converting from ext2 to ext3 whenever possible.
To convert an ext2 file system to ext3, log in as root and type the following command in a terminal:
# tune2fs -j block_deviceblock_device contains the ext2 file system to be converted.
Issue the df command to display mounted file systems.
4.3. Reverting to an Ext2 File System
In order to revert to an ext2 file system, use the following procedure.
For simplicity, the sample commands in this section use the following value for the block device:
/dev/mapper/VolGroup00-LogVol02
Procedure 4.1. Revert from ext3 to ext2
Unmount the partition by logging in as root and typing:
# umount /dev/mapper/VolGroup00-LogVol02Change the file system type to ext2 by typing the following command:
# tune2fs -O ^has_journal /dev/mapper/VolGroup00-LogVol02Check the partition for errors by typing the following command:
# e2fsck -y /dev/mapper/VolGroup00-LogVol02Then mount the partition again as ext2 file system by typing:
# mount -t ext2 /dev/mapper/VolGroup00-LogVol02 /mount/pointReplace /mount/point with the mount point of the partition.
If a .journal file exists at the root level of the partition, delete it.
To permanently change the partition to ext2, remember to update the /etc/fstab file, otherwise it will revert back after booting.
Chapter 5. The ext4 File System
The ext4 file system is a scalable extension of the ext3 file system. With Red Hat Enterprise Linux 7, it can support a maximum individual file size of 16 terabytes, and file systems to a maximum of 50 terabytes, unlike Red Hat Enterprise Linux 6 which only supported file systems up to 16 terabytes. It also supports an unlimited number of sub-directories (the ext3 file system only supports up to 32,000), though once the link count exceeds 65,000 it resets to 1 and is no longer increased. The bigalloc feature is not currently supported.
Main Features
The ext4 file system uses extents (as opposed to the traditional block mapping scheme used by ext2 and ext3), which improves performance when using large files and reduces metadata overhead for large files. In addition, ext4 also labels unallocated block groups and inode table sections accordingly, which allows them to be skipped during a file system check. This makes for quicker file system checks, which becomes more beneficial as the file system grows in size.
Allocation FeaturesThe ext4 file system features the following allocation schemes:
Persistent pre-allocation
Delayed allocation
Multi-block allocation
Stripe-aware allocation
Because of delayed allocation and other performance optimizations, ext4's behavior of writing files to disk is different from ext3. In ext4, when a program writes to the file system, it is not guaranteed to be on-disk unless the program issues an fsync() call afterwards.
By default, ext3 automatically forces newly created files to disk almost immediately even without fsync(). This behavior hid bugs in programs that did not use fsync() to ensure that written data was on-disk. The ext4 file system, on the other hand, often waits several seconds to write out changes to disk, allowing it to combine and reorder writes for better disk performance than ext3.
Unlike ext3, the ext4 file system does not force data to disk on transaction commit. As such, it takes longer for buffered writes to be flushed to disk. As with any file system, use data integrity calls such as fsync() to ensure that data is written to permanent storage.
Other ext4 FeaturesThe ext4 file system also supports the following:
Extended attributes (xattr) — This allows the system to associate several additional name and value pairs per file.
Quota journaling — This avoids the need for lengthy quota consistency checks after a crash.
The only supported journaling mode in ext4 is data=ordered (default).
Subsecond timestamps — This gives timestamps to the subsecond.
5.1. Creating an ext4 File System
To create an ext4 file system, use the following command:
# mkfs.ext4 block_deviceReplace block_device with the path to a block device. For example, /dev/sdb1, /dev/disk/by-uuid/05e99ec8-def1-4a5e-8a9d-5945339ceb2a, or /dev/my-volgroup/my-lv.
In general, the default options are optimal for most usage scenarios.
Example 5.1. mkfs.ext4 Command Output
Below is a sample output of this command, which displays the resulting file system geometry and features:
~]# mkfs.ext4 /dev/sdb1 mke2fs 1.41.12 (17-May-2010) Filesystem label= OS type: Linux Block size=4096 (log=2) Fragment size=4096 (log=2) Stride=0 blocks, Stripe width=0 blocks 245280 inodes, 979456 blocks 48972 blocks (5.00%) reserved for the super user First data block=0 Maximum filesystem blocks=1006632960 30 block groups 32768 blocks per group, 32768 fragments per group 8176 inodes per group Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736 Writing inode tables: done Creating journal (16384 blocks): done Writing superblocks and filesystem accounting information: doneIt is possible to use tune2fs to enable certain ext4 features on ext3 file systems. However, using tune2fs in this way has not been fully tested and is therefore not supported in Red Hat Enterprise Linux 7. As a result, Red Hat cannot guarantee consistent performance and predictable behavior for ext3 file systems converted or mounted by using tune2fs.
Striped Block Devices
For striped block devices (for example, RAID5 arrays), the stripe geometry can be specified at the time of file system creation. Using proper stripe geometry greatly enhances the performance of an ext4 file system.
When creating file systems on LVM or MD volumes, mkfs.ext4 chooses an optimal geometry. This may also be true on some hardware RAIDs which export geometry information to the operating system.
To specify stripe geometry, use the -E option of mkfs.ext4 (that is, extended file system options) with the following sub-options:
stride=value
Specifies the RAID chunk size.
stripe-width=valueSpecifies the number of data disks in a RAID device, or the number of stripe units in the stripe.
For both sub-options, value must be specified in file system block units. For example, to create a file system with a 64k stride (that is, 16 x 4096) on a 4k-block file system, use the following command:
# mkfs.ext4 -E stride=16,stripe-width=64 /dev/block_deviceConfiguring UUID
It is also possible to set a specific UUID for a file system. To specify a UUID when creating a file system, use the -U option:
# mkfs.ext4 -U UUID deviceReplace UUID with the UUID you want to set: for example, 7cd65de3-e0be-41d9-b66d-96d749c02da7.
Replace device with the path to an ext4 file system to have the UUID added to it: for example, /dev/sda8.
Additional Resources
For more information about creating ext4 file systems, see:
The mkfs.ext4(8) man page
5.2. Mounting an ext4 File System
An ext4 file system can be mounted with no extra options. For example:
# mount /dev/device /mount/pointThe ext4 file system also supports several mount options to influence behavior. For example, the acl parameter enables access control lists, while the user_xattr parameter enables user extended attributes. To enable both options, use their respective parameters with -o, as in:
# mount -o acl,user_xattr /dev/device /mount/pointAs with ext3, the option data_err=abort can be used to abort the journal if an error occurs in file data.
# mount -o data_err=abort /dev/device /mount/pointThe tune2fs utility also allows administrators to set default mount options in the file system superblock. For more information on this, refer to man tune2fs.
Write Barriers
By default, ext4 uses write barriers to ensure file system integrity even when power is lost to a device with write caches enabled. For devices without write caches, or with battery-backed write caches, disable barriers using the nobarrier option, as in:
# mount -o nobarrier /dev/device /mount/pointDirect Access Technology Preview
Starting with Red Hat Enterprise Linux 7.3, Direct Access (DAX) provides, as a Technology Preview on the ext4 and XFS file systems, a means for an application to directly map persistent memory into its address space. To use DAX, a system must have some form of persistent memory available, usually in the form of one or more Non-Volatile Dual In-line Memory Modules (NVDIMMs), and a file system that supports DAX must be created on the NVDIMM(s). Also, the file system must be mounted with the dax mount option. Then, an mmap of a file on the dax-mounted file system results in a direct mapping of storage into the application's address space.
5.3. Resizing an ext4 File System
Before growing an ext4 file system, ensure that the underlying block device is of an appropriate size to hold the file system later. Use the appropriate resizing methods for the affected block device.
An ext4 file system may be grown while mounted using the resize2fs command:
# resize2fs /mount/device sizeThe resize2fs command can also decrease the size of an unmounted ext4 file system:
# resize2fs /dev/device sizeWhen resizing an ext4 file system, the resize2fs utility reads the size in units of file system block size, unless a suffix indicating a specific unit is used. The following suffixes indicate specific units:
s — 512 byte sectors
K — kilobytes
M — megabytes
G — gigabytes
The size parameter is optional (and often redundant) when expanding. The resize2fs automatically expands to fill all available space of the container, usually a logical volume or partition.
For more information about resizing an ext4 file system, refer to man resize2fs.
5.4. Backing up ext2, ext3, or ext4 File Systems
This procedure describes how to back up the content of an ext4, ext3, or ext2 file system into a file.
Prerequisites
If the system has been running for a long time, run the e2fsck utility on the partitions before backup:
# e2fsck /dev/device
Procedure 5.1. Backing up ext2, ext3, or ext4 File Systems
Back up configuration information, including the content of the /etc/fstab file and the output of the fdisk -l command. This is useful for restoring the partitions.
Depending on the role of the partition:
If the partition you are backing up is an operating system partition, boot your system into the rescue mode. See the Booting to Rescue Mode section of the System Administrator's Guide.
When backing up a regular, data partition, unmount it.
Although it is possible to back up a data partition while it is mounted, the results of backing up a mounted data partition can be unpredictable.
If you need to back up a mounted file system using the dump utility, do so when the file system is not under a heavy load. The more activity is happening on the file system when backing up, the higher the risk of backup corruption is.
Use the dump utility to back up the content of the partitions:
# dump -0uf backup-file /dev/deviceReplace backup-file with a path to a file where you want the to store the backup. Replace device with the name of the ext4 partition you want to back up. Make sure that you are saving the backup to a directory mounted on a different partition than the partition you are backing up.
Example 5.2. Backing up Multiple ext4 Partitions
To back up the content of the /dev/sda1, /dev/sda2, and /dev/sda3 partitions into backup files stored in the /backup-files/ directory, use the following commands:
# dump -0uf /backup-files/sda1.dump /dev/sda1 # dump -0uf /backup-files/sda2.dump /dev/sda2 # dump -0uf /backup-files/sda3.dump /dev/sda3To do a remote backup, use the ssh utility or configure a password-less ssh login. For more information on ssh and password-less login, see the Using the ssh Utility and Using Key-based Authentication sections of the System Administrator's Guide.
For example, when using ssh:
Example 5.3. Performing a Remote Backup Using ssh
# dump -0u -f - /dev/device | ssh root@remoteserver.example.com dd of=backup-file
Note that if using standard redirection, you must pass the -f option separately.
Additional Resources
For more information, see the dump(8) man page.
5.5. Restoring ext2, ext3, or ext4 File Systems
This procedure describes how to restore an ext4, ext3, or ext2 file system from a file backup.
Prerequisites
Procedure 5.2. Restoring ext2, ext3, or ext4 File Systems
If you are restoring an operating system partition, boot your system into Rescue Mode. See the Booting to Rescue Mode section of the System Administrator's Guide.
This step is not required for ordinary data partitions.
Rebuild the partitions you want to restore by using the fdisk or parted utilites.
If the partitions no longer exist, recreate them. The new partitions must be large enough to contain the restored data. It is important to get the start and end numbers right; these are the starting and ending sector numbers of the partitions obtained from the fdisk utility when backing up.
Use the mkfs utility to format the destination partition:
# mkfs.ext4 /dev/deviceDo not format the partition that stores your backup files.
If you created new partitions, re-label all the partitions so they match their entries in the /etc/fstab file:
# e2label /dev/device labelCreate temporary mount points and mount the partitions on them:
# mkdir /mnt/device # mount -t ext4 /dev/device /mnt/deviceRestore the data from backup on the mounted partition:
# cd /mnt/device # restore -rf device-backup-fileIf you want to restore on a remote machine or restore from a backup file that is stored on a remote host, you can use the ssh utility. For more information on ssh, see the Using the ssh Utility section of the System Administrator's Guide.
Note that you need to configure a password-less login for the following commands. For more information on setting up a password-less ssh login, see the Using Key-based Authentication section of the System Administrator's Guide.
To restore a partition on a remote machine from a backup file stored on the same machine:
# ssh remote-address "cd /mnt/device && cat backup-file | /usr/sbin/restore -r -f -"To restore a partition on a remote machine from a backup file stored on a different remote machine:
# ssh remote-machine-1 "cd /mnt/device && RSH=/usr/bin/ssh /usr/sbin/restore -rf remote-machine-2:backup-file"
Reboot:
# systemctl reboot
Example 5.4. Restoring Multiple ext4 Partitions
Rebuild partitions you want to restore by using the fdisk command.
Format the destination partitions:
# mkfs.ext4 /dev/sda1 # mkfs.ext4 /dev/sda2 # mkfs.ext4 /dev/sda3Re-label all the partitions so they match the /etc/fstab file:
# e2label /dev/sda1 Boot1 # e2label /dev/sda2 Root # e2label /dev/sda3 DataPrepare the working directories.
Mount the new partitions:
# mkdir /mnt/sda1 # mount -t ext4 /dev/sda1 /mnt/sda1 # mkdir /mnt/sda2 # mount -t ext4 /dev/sda2 /mnt/sda2 # mkdir /mnt/sda3 # mount -t ext4 /dev/sda3 /mnt/sda3Mount the partition that contains backup files:
# mkdir /backup-files # mount -t ext4 /dev/sda6 /backup-filesRestore the data from backup to the mounted partitions:
# cd /mnt/sda1 # restore -rf /backup-files/sda1.dump # cd /mnt/sda2 # restore -rf /backup-files/sda2.dump # cd /mnt/sda3 # restore -rf /backup-files/sda3.dumpReboot:
# systemctl reboot
Additional Resources
For more information, see the restore(8) man page.
5.6. Other ext4 File System Utilities
Red Hat Enterprise Linux 7 also features other utilities for managing ext4 file systems:
e2fsck
Used to repair an ext4 file system. This tool checks and repairs an ext4 file system more efficiently than ext3, thanks to updates in the ext4 disk structure.
e2labelChanges the label on an ext4 file system. This tool also works on ext2 and ext3 file systems.
quotaControls and reports on disk space (blocks) and file (inode) usage by users and groups on an ext4 file system. For more information on using quota, refer to man quota and Section 17.1, “Configuring Disk Quotas”.
fsfreezeTo suspend access to a file system, use the command # fsfreeze -f mount-point to freeze it and # fsfreeze -u mount-point to unfreeze it. This halts access to the file system and creates a stable image on disk.
It is unnecessary to use fsfreeze for device-mapper drives.
For more information see the fsfreeze(8) manpage.
As demonstrated in Section 5.2, “Mounting an ext4 File System”, the tune2fs utility can also adjust configurable file system parameters for ext2, ext3, and ext4 file systems. In addition, the following tools are also useful in debugging and analyzing ext4 file systems:
debugfs
Debugs ext2, ext3, or ext4 file systems.
e2imageSaves critical ext2, ext3, or ext4 file system metadata to a file.
For more information about these utilities, refer to their respective man pages.
Chapter 6. Btrfs (Technology Preview)
Btrfs is available as a Technology Preview feature in Red Hat Enterprise Linux 7 but has been deprecated since the Red Hat Enterprise Linux 7.4 release. It will be removed in a future major release of Red Hat Enterprise Linux.
For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.4 Release Notes.
Btrfs is a next generation Linux file system that offers advanced management, reliability, and scalability features. It is unique in offering snapshots, compression, and integrated device management.
6.1. Creating a btrfs File System
In order to make a basic btrfs file system, use the following command:
# mkfs.btrfs /dev/device6.2. Mounting a btrfs file system
To mount any device in the btrfs file system use the following command:
# mount /dev/device /mount-pointOther useful mount options include:
device=/dev/name
Appending this option to the mount command tells btrfs to scan the named device for a btrfs volume. This is used to ensure the mount will succeed as attempting to mount devices that are not btrfs will cause the mount to fail.
This does not mean all devices will be added to the file system, it only scans them.
max_inline=numberUse this option to set the maximum amount of space (in bytes) that can be used to inline data within a metadata B-tree leaf. The default is 8192 bytes. For 4k pages it is limited to 3900 bytes due to additional headers that need to fit into the leaf.
alloc_start=numberUse this option to set where in the disk allocations start.
thread_pool=numberUse this option to assign the number of worker threads allocated.
discardUse this option to enable discard/TRIM on freed blocks.
noaclUse this option to disable the use of ACL's.
space_cacheUse this option to store the free space data on disk to make caching a block group faster. This is a persistent change and is safe to boot into old kernels.
nospace_cacheUse this option to disable the above space_cache.
clear_cacheUse this option to clear all the free space caches during mount. This is a safe option but will trigger the space cache to be rebuilt. As such, leave the file system mounted in order to let the rebuild process finish. This mount option is intended to be used once and only after problems are apparent with the free space.
enospc_debugThis option is used to debug problems with "no space left".
recoveryUse this option to enable autorecovery upon mount.
6.3. Resizing a btrfs File System
It is not possible to resize a btrfs file system but it is possible to resize each of the devices it uses. If there is only one device in use then this works the same as resizing the file system. If there are multiple devices in use then they must be manually resized to achieve the desired result.
The unit size is not case specific; it accepts both G or g for GiB.
The command does not accept t for terabytes or p for petabytes. It only accepts k, m, and g.
Enlarging a btrfs File System
To enlarge the file system on a single device, use the command:
# btrfs filesystem resize amount /mount-pointFor example:
# btrfs filesystem resize +200M /btrfssingle Resize '/btrfssingle' of '+200M'To enlarge a multi-device file system, the device to be enlarged must be specified. First, show all devices that have a btrfs file system at a specified mount point:
# btrfs filesystem show /mount-pointFor example:
# btrfs filesystem show /btrfstest Label: none uuid: 755b41b7-7a20-4a24-abb3-45fdbed1ab39 Total devices 4 FS bytes used 192.00KiB devid 1 size 1.00GiB used 224.75MiB path /dev/vdc devid 2 size 524.00MiB used 204.75MiB path /dev/vdd devid 3 size 1.00GiB used 8.00MiB path /dev/vde devid 4 size 1.00GiB used 8.00MiB path /dev/vdf Btrfs v3.16.2Then, after identifying the devid of the device to be enlarged, use the following command:
# btrfs filesystem resize devid:amount /mount-pointFor example:
# btrfs filesystem resize 2:+200M /btrfstest Resize '/btrfstest/' of '2:+200M'The amount can also be max instead of a specified amount. This will use all remaining free space on the device.
Shrinking a btrfs File System
To shrink the file system on a single device, use the command:
# btrfs filesystem resize amount /mount-pointFor example:
# btrfs filesystem resize -200M /btrfssingle Resize '/btrfssingle' of '-200M'To shrink a multi-device file system, the device to be shrunk must be specified. First, show all devices that have a btrfs file system at a specified mount point:
# btrfs filesystem show /mount-pointFor example:
# btrfs filesystem show /btrfstest Label: none uuid: 755b41b7-7a20-4a24-abb3-45fdbed1ab39 Total devices 4 FS bytes used 192.00KiB devid 1 size 1.00GiB used 224.75MiB path /dev/vdc devid 2 size 524.00MiB used 204.75MiB path /dev/vdd devid 3 size 1.00GiB used 8.00MiB path /dev/vde devid 4 size 1.00GiB used 8.00MiB path /dev/vdf Btrfs v3.16.2Then, after identifying the devid of the device to be shrunk, use the following command:
# btrfs filesystem resize devid:amount /mount-pointFor example:
# btrfs filesystem resize 2:-200M /btrfstest Resize '/btrfstest' of '2:-200M'Set the File System Size
To set the file system to a specific size on a single device, use the command:
# btrfs filesystem resize amount /mount-pointFor example:
# btrfs filesystem resize 700M /btrfssingle Resize '/btrfssingle' of '700M'To set the file system size of a multi-device file system, the device to be changed must be specified. First, show all devices that have a btrfs file system at the specified mount point:
# btrfs filesystem show /mount-pointFor example:
# btrfs filesystem show /btrfstest Label: none uuid: 755b41b7-7a20-4a24-abb3-45fdbed1ab39 Total devices 4 FS bytes used 192.00KiB devid 1 size 1.00GiB used 224.75MiB path /dev/vdc devid 2 size 724.00MiB used 204.75MiB path /dev/vdd devid 3 size 1.00GiB used 8.00MiB path /dev/vde devid 4 size 1.00GiB used 8.00MiB path /dev/vdf Btrfs v3.16.2Then, after identifying the devid of the device to be changed, use the following command:
# btrfs filesystem resize devid:amount /mount-pointFor example:
# btrfs filesystem resize 2:300M /btrfstest Resize '/btrfstest' of '2:300M'6.4. Integrated Volume Management of Multiple Devices
A btrfs file system can be created on top of many devices, and more devices can be added after the file system has been created. By default, metadata will be mirrored across two devices and data will be striped across all devices present, however if only one device is present, metadata will be duplicated on that device.
6.4.1. Creating a File System with Multiple Devices
raid0
raid1
raid10
dup
single
The -m single option instructs that no duplication of metadata is done. This may be desired when using hardware raid.
RAID 10 requires at least four devices to run correctly.
Example 6.1. Creating a RAID 10 btrfs File System
Create a file system across four devices (metadata mirrored, data striped).
# mkfs.btrfs /dev/device1 /dev/device2 /dev/device3 /dev/device4Stripe the metadata without mirroring.
# mkfs.btrfs -m raid0 /dev/device1 /dev/device2Use raid10 for both data and metadata.
# mkfs.btrfs -m raid10 -d raid10 /dev/device1 /dev/device2 /dev/device3 /dev/device4Do not duplicate metadata on a single drive.
# mkfs.btrfs -m single /dev/deviceUse the single option to use the full capacity of each drive when the drives are different sizes.
# mkfs.btrfs -d single /dev/device1 /dev/device2 /dev/device3To add a new device to an already created multi-device file system, use the following command:
# btrfs device add /dev/device1 /mount-pointAfter rebooting or reloading the btrfs module, use the btrfs device scan command to discover all multi-device file systems. See Section 6.4.2, “Scanning for btrfs Devices” for more information.
6.4.2. Scanning for btrfs Devices
Use btrfs device scan to scan all block devices under /dev and probe for btrfs volumes. This must be performed after loading the btrfs module if running with more than one device in a file system.
To scan all devices, use the following command:
# btrfs device scanTo scan a single device, use the following command:
# btrfs device scan /dev/device6.4.3. Adding New Devices to a btrfs File System
Use the btrfs filesystem show command to list all the btrfs file systems and which devices they include.
The btrfs device add command is used to add new devices to a mounted file system.
The btrfs filesystem balance command balances (restripes) the allocated extents across all existing devices.
An example of all these commands together to add a new device is as follows:
Example 6.2. Add a New Device to a btrfs File System
# mkfs.btrfs /dev/device1 # mount /dev/device1
Next, add a second device to the mounted btrfs file system.
# btrfs device add /dev/device2 /mount-pointThe metadata and data on these devices are still stored only on /dev/device1. It must now be balanced to spread across all devices.
# btrfs filesystem balance /mount-pointBalancing a file system will take some time as it reads all of the file system's data and metadata and rewrites it across the new device.
6.4.4. Converting a btrfs File System
To convert a non-raid file system to a raid, add a device and run a balance filter that changes the chunk allocation profile.
Example 6.3. Converting a btrfs File System
To convert an existing single device system, /dev/sdb1 in this case, into a two device, raid1 system in order to protect against a single disk failure, use the following commands:
# mount /dev/sdb1 /mnt # btrfs device add /dev/sdc1 /mnt # btrfs balance start -dconvert=raid1 -mconvert=raid1 /mntIf the metadata is not converted from the single-device default, it remains as DUP. This does not guarantee that copies of the block are on separate devices. If data is not converted it does not have any redundant copies at all.
6.4.5. Removing btrfs Devices
Use the btrfs device delete command to remove an online device. It redistributes any extents in use to other devices in the file system in order to be safely removed.
Example 6.4. Removing a Device on a btrfs File System
First create and mount a few btrfs file systems.
# mkfs.btrfs /dev/sdb /dev/sdc /dev/sdd /dev/sde # mount /dev/sdb /mntAdd some data to the file system.
Finally, remove the required device.
# btrfs device delete /dev/sdc /mnt6.4.6. Replacing Failed Devices on a btrfs File System
Section 6.4.5, “Removing btrfs Devices” can be used to remove a failed device provided the super block can still be read. However, if a device is missing or the super block corrupted, the file system will need to be mounted in a degraded mode:
# mkfs.btrfs -m raid1 /dev/sdb /dev/sdc /dev/sdd /dev/sde ssd is destroyed or removed, use -o degraded to force the mount to ignore missing devices # mount -o degraded /dev/sdb /mnt 'missing' is a special device name # btrfs device delete missing /mntThe command btrfs device delete missing removes the first device that is described by the file system metadata but not present when the file system was mounted.
It is impossible to go below the minimum number of devices required for the specific raid layout, even including the missing one. It may be required to add a new device in order to remove the failed one.
For example, for a raid1 layout with two devices, if a device fails it is required to:
mount in degraded mode,
add a new device,
and, remove the missing device.
6.4.7. Registering a btrfs File System in /etc/fstab
If you do not have an initrd or it does not perform a btrfs device scan, it is possible to mount a multi-volume btrfs file system by passing all the devices in the file system explicitly to the mount command.
Example 6.5. Example /etc/fstab Entry
An example of a suitable /etc/fstab entry would be:
/dev/sdb /mnt btrfs device=/dev/sdb,device=/dev/sdc,device=/dev/sdd,device=/dev/sde 0Note that using universally unique identifiers (UUIDs) also works and is more stable than using device paths.
6.5. SSD Optimization
Using the btrfs file system can optimize SSD. There are two ways this can be done.
The first way is mkfs.btrfs turns off metadata duplication on a single device when /sys/block/device/queue/rotational is zero for the single specified device. This is equivalent to specifying -m single on the command line. It can be overridden and duplicate metadata forced by providing the -m dup option. Duplication is not required due to SSD firmware potentially losing both copies. This wastes space and is a performance cost.
The second way is through a group of SSD mount options: ssd, nossd, and ssd_spread.
The ssd option does several things:
It allows larger metadata cluster allocation.
It allocates data more sequentially where possible.
It disables btree leaf rewriting to match key and block order.
It commits log fragments without batching multiple processes.
The ssd mount option only enables the ssd option. Use the nossd option to disable it.
Some SSDs perform best when reusing block numbers often, while others perform much better when clustering strictly allocates big chunks of unused space. By default, mount -o ssd will find groupings of blocks where there are several free blocks that might have allocated blocks mixed in. The command mount -o ssd_spread ensures there are no allocated blocks mixed in. This improves performance on lower end SSDs.
The ssd_spread option enables both the ssd and the ssd_spread options. Use the nossd to disable both these options.
The ssd_spread option is never automatically set if none of the ssd options are provided and any of the devices are non-rotational.
These options will all need to be tested with your specific build to see if their use improves or reduces performance, as each combination of SSD firmware and application loads are different.
6.6. btrfs References
The man page btrfs(8) covers all important management commands. In particular this includes:
All the subvolume commands for managing snapshots.
The device commands for managing devices.
The scrub, balance, and defragment commands.
The man page mkfs.btrfs(8) contains information on creating a btrfs file system including all the options regarding it.
The man page btrfsck(8) for information regarding fsck on btrfs systems.
Chapter 7. Global File System 2
The Red Hat Global File System 2 (GFS2) is a native file system that interfaces directly with the Linux kernel file system interface (VFS layer). When implemented as a cluster file system, GFS2 employs distributed metadata and multiple journals.
GFS2 is based on 64-bit architecture, which can theoretically accommodate an 8 exabyte file system. However, the current supported maximum size of a GFS2 file system is 100 TB. If a system requires GFS2 file systems larger than 100 TB, contact your Red Hat service representative.
When determining the size of a file system, consider its recovery needs. Running the fsck command on a very large file system can take a long time and consume a large amount of memory. Additionally, in the event of a disk or disk-subsystem failure, recovery time is limited by the speed of backup media.
When configured in a Red Hat Cluster Suite, Red Hat GFS2 nodes can be configured and managed with Red Hat Cluster Suite configuration and management tools. Red Hat GFS2 then provides data sharing among GFS2 nodes in a Red Hat cluster, with a single, consistent view of the file system namespace across the GFS2 nodes. This allows processes on different nodes to share GFS2 files in the same way that processes on the same node can share files on a local file system, with no discernible difference. For information about the Red Hat Cluster Suite, see Red Hat's Cluster Administration guide.
A GFS2 must be built on a logical volume (created with LVM) that is a linear or mirrored volume. Logical volumes created with LVM in a Red Hat Cluster suite are managed with CLVM (a cluster-wide implementation of LVM), enabled by the CLVM daemon clvmd, and running in a Red Hat Cluster Suite cluster. The daemon makes it possible to use LVM2 to manage logical volumes across a cluster, allowing all nodes in the cluster to share the logical volumes. For information on the Logical Volume Manager, see Red Hat's Logical Volume Manager Administration guide.
The gfs2.ko kernel module implements the GFS2 file system and is loaded on GFS2 cluster nodes.
For comprehensive information on the creation and configuration of GFS2 file systems in clustered and non-clustered storage, see Red Hat's Global File System 2 guide.
Chapter 8. Network File System (NFS)
A Network File System (NFS) allows remote hosts to mount file systems over a network and interact with those file systems as though they are mounted locally. This enables system administrators to consolidate resources onto centralized servers on the network.
This chapter focuses on fundamental NFS concepts and supplemental information.
8.1. Introduction to NFS
Currently, there are two major versions of NFS included in Red Hat Enterprise Linux:
NFS version 3 (NFSv3) supports safe asynchronous writes and is more robust at error handling than the previous NFSv2. It also supports 64-bit file sizes and offsets, allowing clients to access more than 2 GB of file data.
NFS version 4 (NFSv4) works through firewalls and on the Internet, no longer requires an rpcbind service, supports ACLs, and utilizes stateful operations.
Red Hat Enterprise Linux fully supports NFS version 4.2 (NFSv4.2) since the Red Hat Enterprise Linux 7.4 release.
Following are the features of NFSv4.2 in Red Hat Enterprise Linux :
Sparse Files: It verifies space efficiency of a file and allows placeholder to improve storage efficiency. It is a file having one or more holes; holes are unallocated or uninitialized data blocks consisting only of zeroes. lseek() operation in NFSv4.2, supports seek_hole() and seek_data(), which allows application to map out the location of holes in the sparse file.
Space Reservation: It permits storage servers to reserve free space, which prohibits servers to run out of space. NFSv4.2 supports allocate() operation to reserve space, deallocate() operation to unreserve space, and fallocate() operation to preallocate or deallocate space in a file.
Labeled NFS: It enforces data access rights and enables SELinux labels between a client and a server for individual files on an NFS file system.
Layout Enhancements: NFSv4.2 provides new operation, layoutstats(), which the client can use to notify the metadata server about its communication with the layout.
Versions of Red Hat Enterprise Linux earlier than 7.4 support NFS up to version 4.1.
Following are the features of NFSv4.1:
Enhances performance and security of network, and also includes client-side support for Parallel NFS (pNFS).
No longer requires a separate TCP connection for callbacks, which allows an NFS server to grant delegations even when it cannot contact the client. For example, when NAT or a firewall interferes.
It provides exactly once semantics (except for reboot operations), preventing a previous issue whereby certain operations could return an inaccurate result if a reply was lost and the operation was sent twice.
NFS clients attempt to mount using NFSv4.1 by default, and fall back to NFSv4.0 when the server does not support NFSv4.1. The mount later fall back to NFSv3 when server does not support NFSv4.0.
NFS version 2 (NFSv2) is no longer supported by Red Hat.
All versions of NFS can use Transmission Control Protocol (TCP) running over an IP network, with NFSv4 requiring it. NFSv3 can use the User Datagram Protocol (UDP) running over an IP network to provide a stateless network connection between the client and server.
When using NFSv3 with UDP, the stateless UDP connection (under normal conditions) has less protocol overhead than TCP. This can translate into better performance on very clean, non-congested networks. However, because UDP is stateless, if the server goes down unexpectedly, UDP clients continue to saturate the network with requests for the server. In addition, when a frame is lost with UDP, the entire RPC request must be retransmitted; with TCP, only the lost frame needs to be resent. For these reasons, TCP is the preferred protocol when connecting to an NFS server.
The mounting and locking protocols have been incorporated into the NFSv4 protocol. The server also listens on the well-known TCP port 2049. As such, NFSv4 does not need to interact with rpcbind [1], lockd, and rpc.statd daemons. The rpc.mountd daemon is still required on the NFS server to set up the exports, but is not involved in any over-the-wire operations.
TCP is the default transport protocol for NFS version 3 under Red Hat Enterprise Linux. UDP can be used for compatibility purposes as needed, but is not recommended for wide usage. NFSv4 requires TCP.
All the RPC/NFS daemons have a '-p' command line option that can set the port, making firewall configuration easier.
After TCP wrappers grant access to the client, the NFS server refers to the /etc/exports configuration file to determine whether the client is allowed to access any exported file systems. Once verified, all file and directory operations are available to the user.
In order for NFS to work with a default installation of Red Hat Enterprise Linux with a firewall enabled, configure IPTables with the default TCP port 2049. Without proper IPTables configuration, NFS will not function properly.
The NFS initialization script and rpc.nfsd process now allow binding to any specified port during system start up. However, this can be error-prone if the port is unavailable, or if it conflicts with another daemon.
8.1.1. Required Services
Red Hat Enterprise Linux uses a combination of kernel-level support and daemon processes to provide NFS file sharing. All NFS versions rely on Remote Procedure Calls (RPC) between clients and servers. RPC services under Red Hat Enterprise Linux 7 are controlled by the rpcbind service. To share or mount NFS file systems, the following services work together depending on which version of NFS is implemented:
The portmap service was used to map RPC program numbers to IP address port number combinations in earlier versions of Red Hat Enterprise Linux. This service is now replaced by rpcbind in Red Hat Enterprise Linux 7 to enable IPv6 support.
nfs
systemctl start nfs starts the NFS server and the appropriate RPC processes to service requests for shared NFS file systems.
nfslocksystemctl start nfs-lock activates a mandatory service that starts the appropriate RPC processes allowing NFS clients to lock files on the server.
rpcbindrpcbind accepts port reservations from local RPC services. These ports are then made available (or advertised) so the corresponding remote RPC services can access them. rpcbind responds to requests for RPC services and sets up connections to the requested RPC service. This is not used with NFSv4.
The following RPC processes facilitate NFS services:
rpc.mountd
This process is used by an NFS server to process MOUNT requests from NFSv3 clients. It checks that the requested NFS share is currently exported by the NFS server, and that the client is allowed to access it. If the mount request is allowed, the rpc.mountd server replies with a Success status and provides the File-Handle for this NFS share back to the NFS client.
rpc.nfsdrpc.nfsd allows explicit NFS versions and protocols the server advertises to be defined. It works with the Linux kernel to meet the dynamic demands of NFS clients, such as providing server threads each time an NFS client connects. This process corresponds to the nfs service.
lockdlockd is a kernel thread which runs on both clients and servers. It implements the Network Lock Manager (NLM) protocol, which allows NFSv3 clients to lock files on the server. It is started automatically whenever the NFS server is run and whenever an NFS file system is mounted.
rpc.statdThis process implements the Network Status Monitor (NSM) RPC protocol, which notifies NFS clients when an NFS server is restarted without being gracefully brought down. rpc.statd is started automatically by the nfslock service, and does not require user configuration. This is not used with NFSv4.
rpc.rquotadThis process provides user quota information for remote users. rpc.rquotad is started automatically by the nfs service and does not require user configuration.
rpc.idmapdrpc.idmapd provides NFSv4 client and server upcalls, which map between on-the-wire NFSv4 names (strings in the form of user@domain) and local UIDs and GIDs. For idmapd to function with NFSv4, the /etc/idmapd.conf file must be configured. At a minimum, the "Domain" parameter should be specified, which defines the NFSv4 mapping domain. If the NFSv4 mapping domain is the same as the DNS domain name, this parameter can be skipped. The client and server must agree on the NFSv4 mapping domain for ID mapping to function properly.
In Red Hat Enterprise Linux 7, only the NFSv4 server uses rpc.idmapd. The NFSv4 client uses the keyring-based idmapper nfsidmap. nfsidmap is a stand-alone program that is called by the kernel on-demand to perform ID mapping; it is not a daemon. If there is a problem with nfsidmap does the client fall back to using rpc.idmapd. More information regarding nfsidmap can be found on the nfsidmap man page.
8.2. Configuring NFS Client
The mount command mounts NFS shares on the client side. Its format is as follows:
# mount -t nfs -o options server:/remote/export /local/directoryThis command uses the following variables:
optionsserver
The hostname, IP address, or fully qualified domain name of the server exporting the file system you wish to mount
/remote/exportThe file system or directory being exported from the server, that is, the directory you wish to mount
/local/directoryThe client location where /remote/export is mounted
The NFS protocol version used in Red Hat Enterprise Linux 7 is identified by the mount options nfsvers or vers. By default, mount uses NFSv4 with mount -t nfs. If the server does not support NFSv4, the client automatically steps down to a version supported by the server. If the nfsvers/vers option is used to pass a particular version not supported by the server, the mount fails. The file system type nfs4 is also available for legacy reasons; this is equivalent to running mount -t nfs -o nfsvers=4 host:/remote/export /local/directory.
For more information, see man mount.
If an NFS share was mounted manually, the share will not be automatically mounted upon reboot. Red Hat Enterprise Linux offers two methods for mounting remote file systems automatically at boot time: the /etc/fstab file and the autofs service. For more information, see Section 8.2.1, “Mounting NFS File Systems Using /etc/fstab” and Section 8.3, “autofs”.
8.2.1. Mounting NFS File Systems Using /etc/fstab
An alternate way to mount an NFS share from another machine is to add a line to the /etc/fstab file. The line must state the hostname of the NFS server, the directory on the server being exported, and the directory on the local machine where the NFS share is to be mounted. You must be root to modify the /etc/fstab file.
Example 8.1. Syntax Example
The general syntax for the line in /etc/fstab is as follows:
server:/usr/local/pub /pub nfs defaults 0 0The mount point /pub must exist on the client machine before this command can be executed. After adding this line to /etc/fstab on the client system, use the command mount /pub, and the mount point /pub is mounted from the server.
A valid /etc/fstab entry to mount an NFS export should contain the following information:
server:/remote/export /local/directory nfs options 0 0The variables server, /remote/export, /local/directory, and options are the same ones used when manually mounting an NFS share. For more information, see Section 8.2, “Configuring NFS Client”.
The mount point /local/directory must exist on the client before /etc/fstab is read. Otherwise, the mount fails.
After editing /etc/fstab, regenerate mount units so that your system registers the new configuration:
# systemctl daemon-reloadAdditional Resources
For more information about /etc/fstab, refer to man fstab.
8.3. autofs
One drawback of using /etc/fstab is that, regardless of how infrequently a user accesses the NFS mounted file system, the system must dedicate resources to keep the mounted file system in place. This is not a problem with one or two mounts, but when the system is maintaining mounts to many systems at one time, overall system performance can be affected. An alternative to /etc/fstab is to use the kernel-based automount utility. An automounter consists of two components:
a kernel module that implements a file system, and
a user-space daemon that performs all of the other functions.
The automount utility can mount and unmount NFS file systems automatically (on-demand mounting), therefore saving system resources. It can be used to mount other file systems including AFS, SMBFS, CIFS, and local file systems.
The nfs-utils package is now a part of both the 'NFS file server' and the 'Network File System Client' groups. As such, it is no longer installed by default with the Base group. Ensure that nfs-utils is installed on the system first before attempting to automount an NFS share.
autofs is also part of the 'Network File System Client' group.
autofs uses /etc/auto.master (master map) as its default primary configuration file. This can be changed to use another supported network source and name using the autofs configuration (in /etc/sysconfig/autofs) in conjunction with the Name Service Switch (NSS) mechanism. An instance of the autofs version 4 daemon was run for each mount point configured in the master map and so it could be run manually from the command line for any given mount point. This is not possible with autofs version 5, because it uses a single daemon to manage all configured mount points; as such, all automounts must be configured in the master map. This is in line with the usual requirements of other industry standard automounters. Mount point, hostname, exported directory, and options can all be specified in a set of files (or other supported network sources) rather than configuring them manually for each host.
8.3.1. Improvements in autofs Version 5 over Version 4
autofs version 5 features the following enhancements over version 4:
Direct map support
Direct maps in autofs provide a mechanism to automatically mount file systems at arbitrary points in the file system hierarchy. A direct map is denoted by a mount point of /- in the master map. Entries in a direct map contain an absolute path name as a key (instead of the relative path names used in indirect maps).
Lazy mount and unmount supportMulti-mount map entries describe a hierarchy of mount points under a single key. A good example of this is the -hosts map, commonly used for automounting all exports from a host under /net/host as a multi-mount map entry. When using the -hosts map, an ls of /net/host will mount autofs trigger mounts for each export from host. These will then mount and expire them as they are accessed. This can greatly reduce the number of active mounts needed when accessing a server with a large number of exports.
Enhanced LDAP supportThe autofs configuration file (/etc/sysconfig/autofs) provides a mechanism to specify the autofs schema that a site implements, thus precluding the need to determine this via trial and error in the application itself. In addition, authenticated binds to the LDAP server are now supported, using most mechanisms supported by the common LDAP server implementations. A new configuration file has been added for this support: /etc/autofs_ldap_auth.conf. The default configuration file is self-documenting, and uses an XML format.
Proper use of the Name Service Switch (nsswitch) configuration.The Name Service Switch configuration file exists to provide a means of determining from where specific configuration data comes. The reason for this configuration is to allow administrators the flexibility of using the back-end database of choice, while maintaining a uniform software interface to access the data. While the version 4 automounter is becoming increasingly better at handling the NSS configuration, it is still not complete. Autofs version 5, on the other hand, is a complete implementation.
For more information on the supported syntax of this file, see man nsswitch.conf. Not all NSS databases are valid map sources and the parser will reject ones that are invalid. Valid sources are files, yp, nis, nisplus, ldap, and hesiod.
Multiple master map entries per autofs mount pointOne thing that is frequently used but not yet mentioned is the handling of multiple master map entries for the direct mount point /-. The map keys for each entry are merged and behave as one map.
Example 8.2. Multiple Master Map Entries per autofs Mount Point
Following is an example in the connectathon test maps for the direct mounts:
/- /tmp/auto_dcthon /- /tmp/auto_test3_direct /- /tmp/auto_test4_direct8.3.2. Configuring autofs
The primary configuration file for the automounter is /etc/auto.master, also referred to as the master map which may be changed as described in the Section 8.3.1, “Improvements in autofs Version 5 over Version 4”. The master map lists autofs-controlled mount points on the system, and their corresponding configuration files or network sources known as automount maps. The format of the master map is as follows:
mount-point map-name optionsThe variables used in this format are:
mount-point
The autofs mount point, /home, for example.
map-nameThe name of a map source which contains a list of mount points, and the file system location from which those mount points should be mounted.
optionsIf supplied, these applies to all entries in the given map provided they do not themselves have options specified. This behavior is different from autofs version 4 where options were cumulative. This has been changed to implement mixed environment compatibility.
Example 8.3. /etc/auto.master File
The following is a sample line from /etc/auto.master file (displayed with cat /etc/auto.master):
/home /etc/auto.miscThe general format of maps is similar to the master map, however the "options" appear between the mount point and the location instead of at the end of the entry as in the master map:
mount-point [options] locationThe variables used in this format are:
mount-point
This refers to the autofs mount point. This can be a single directory name for an indirect mount or the full path of the mount point for direct mounts. Each direct and indirect map entry key (mount-point) may be followed by a space separated list of offset directories (subdirectory names each beginning with a /) making them what is known as a multi-mount entry.
optionsWhenever supplied, these are the mount options for the map entries that do not specify their own options.
locationThis refers to the file system location such as a local file system path (preceded with the Sun map format escape character ":" for map names beginning with /), an NFS file system or other valid file system location.
The following is a sample of contents from a map file (for example, /etc/auto.misc):
payroll -fstype=nfs personnel:/dev/hda3 sales -fstype=ext3 :/dev/hda4The first column in a map file indicates the autofs mount point (sales and payroll from the server called personnel). The second column indicates the options for the autofs mount while the third column indicates the source of the mount. Following the given configuration, the autofs mount points will be /home/payroll and /home/sales. The -fstype= option is often omitted and is generally not needed for correct operation.
The automounter create the directories if they do not exist. If the directories exist before the automounter was started, the automounter will not remove them when it exits.
To start the automount daemon, use the following command:
# systemctl start autofsTo restart the automount daemon, use the following command:
# systemctl restart autofsUsing the given configuration, if a process requires access to an autofs unmounted directory such as /home/payroll/2006/July.sxc, the automount daemon automatically mounts the directory. If a timeout is specified, the directory is automatically unmounted if the directory is not accessed for the timeout period.
To view the status of the automount daemon, use the following command:
# systemctl status autofs8.3.3. Overriding or Augmenting Site Configuration Files
It can be useful to override site defaults for a specific mount point on a client system. For example, consider the following conditions:
Automounter maps are stored in NIS and the /etc/nsswitch.conf file has the following directive:
automount: files nisThe auto.master file contains:
+auto.masterThe NIS auto.master map file contains:
/home auto.homeThe NIS auto.home map contains:
beth fileserver.example.com:/export/home/beth joe fileserver.example.com:/export/home/joe * fileserver.example.com:/export/home/&The file map /etc/auto.home does not exist.
Given these conditions, let's assume that the client system needs to override the NIS map auto.home and mount home directories from a different server. In this case, the client needs to use the following /etc/auto.master map:
/home /etc/auto.home +auto.masterThe /etc/auto.home map contains the entry:
* labserver.example.com:/export/home/&Because the automounter only processes the first occurrence of a mount point, /home contain the contents of /etc/auto.home instead of the NIS auto.home map.
Alternatively, to augment the site-wide auto.home map with just a few entries, create an /etc/auto.home file map, and in it put the new entries. At the end, include the NIS auto.home map. Then the /etc/auto.home file map looks similar to:
mydir someserver:/export/mydir +auto.homeWith these NIS auto.home map conditions, the ls /home command outputs:
beth joe mydirThis last example works as expected because autofs does not include the contents of a file map of the same name as the one it is reading. As such, autofs moves on to the next map source in the nsswitch configuration.
8.3.4. Using LDAP to Store Automounter Maps
LDAP client libraries must be installed on all systems configured to retrieve automounter maps from LDAP. On Red Hat Enterprise Linux, the openldap package should be installed automatically as a dependency of the automounter. To configure LDAP access, modify /etc/openldap/ldap.conf. Ensure that BASE, URI, and schema are set appropriately for your site.
The most recently established schema for storing automount maps in LDAP is described by rfc2307bis. To use this schema it is necessary to set it in the autofs configuration (/etc/sysconfig/autofs) by removing the comment characters from the schema definition. For example:
Example 8.4. Setting autofs Configuration
DEFAULT_MAP_OBJECT_CLASS="automountMap" DEFAULT_ENTRY_OBJECT_CLASS="automount" DEFAULT_MAP_ATTRIBUTE="automountMapName" DEFAULT_ENTRY_ATTRIBUTE="automountKey" DEFAULT_VALUE_ATTRIBUTE="automountInformation"
Ensure that these are the only schema entries not commented in the configuration. The automountKey replaces the cn attribute in the rfc2307bis schema. Following is an example of an LDAP Data Interchange Format (LDIF) configuration:
Example 8.5. LDF Configuration
# extended LDIF # # LDAPv3 # base <> with scope subtree # filter: (&(objectclass=automountMap)(automountMapName=auto.master)) # requesting: ALL # # auto.master, example.com dn: automountMapName=auto.master,dc=example,dc=com objectClass: top objectClass: automountMap automountMapName: auto.master # extended LDIF # # LDAPv3 # base <automountMapName=auto.master,dc=example,dc=com> with scope subtree # filter: (objectclass=automount) # requesting: ALL # # /home, auto.master, example.com dn: automountMapName=auto.master,dc=example,dc=com objectClass: automount cn: /home automountKey: /home automountInformation: auto.home # extended LDIF # # LDAPv3 # base <> with scope subtree # filter: (&(objectclass=automountMap)(automountMapName=auto.home)) # requesting: ALL # # auto.home, example.com dn: automountMapName=auto.home,dc=example,dc=com objectClass: automountMap automountMapName: auto.home # extended LDIF # # LDAPv3 # base <automountMapName=auto.home,dc=example,dc=com> with scope subtree # filter: (objectclass=automount) # requesting: ALL # # foo, auto.home, example.com dn: automountKey=foo,automountMapName=auto.home,dc=example,dc=com objectClass: automount automountKey: foo automountInformation: filer.example.com:/export/foo # /, auto.home, example.com dn: automountKey=/,automountMapName=auto.home,dc=example,dc=com objectClass: automount automountKey: / automountInformation: filer.example.com:/export/&
8.4. Common NFS Mount Options
Beyond mounting a file system with NFS on a remote host, it is also possible to specify other options at mount time to make the mounted share easier to use. These options can be used with manual mount commands, /etc/fstab settings, and autofs.
The following are options commonly used for NFS mounts:
lookupcache=mode
Specifies how the kernel should manage its cache of directory entries for a given mount point. Valid arguments for mode are all, none, or pos/positive.
nfsvers=versionSpecifies which version of the NFS protocol to use, where version is 3 or 4. This is useful for hosts that run multiple NFS servers. If no version is specified, NFS uses the highest version supported by the kernel and mount command.
The option vers is identical to nfsvers, and is included in this release for compatibility reasons.
noaclTurns off all ACL processing. This may be needed when interfacing with older versions of Red Hat Enterprise Linux, Red Hat Linux, or Solaris, since the most recent ACL technology is not compatible with older systems.
nolockDisables file locking. This setting is sometimes required when connecting to very old NFS servers.
noexecPrevents execution of binaries on mounted file systems. This is useful if the system is mounting a non-Linux file system containing incompatible binaries.
nosuidDisables set-user-identifier or set-group-identifier bits. This prevents remote users from gaining higher privileges by running a setuid program.
port=numSpecifies the numeric value of the NFS server port. If num is 0 (the default value), then mount queries the remote host's rpcbind service for the port number to use. If the remote host's NFS daemon is not registered with its rpcbind service, the standard NFS port number of TCP 2049 is used instead.
rsize=num and wsize=numThese options set the maximum number of bytes to be transfered in a single NFS read or write operation.
There is no fixed default value for rsize and wsize. By default, NFS uses the largest possible value that both the server and the client support. In Red Hat Enterprise Linux 7, the client and server maximum is 1,048,576 bytes. For more details, see the What are the default and maximum values for rsize and wsize with NFS mounts? KBase article.
sec=flavorsSecurity flavors to use for accessing files on the mounted export. The flavors value is a colon-separated list of one or more security flavors.
By default, the client attempts to find a security flavor that both the client and the server support. If the server does not support any of the selected flavors, the mount operation fails.
sec=sys uses local UNIX UIDs and GIDs. These use AUTH_SYS to authenticate NFS operations.
sec=krb5 uses Kerberos V5 instead of local UNIX UIDs and GIDs to authenticate users.
sec=krb5i uses Kerberos V5 for user authentication and performs integrity checking of NFS operations using secure checksums to prevent data tampering.
sec=krb5p uses Kerberos V5 for user authentication, integrity checking, and encrypts NFS traffic to prevent traffic sniffing. This is the most secure setting, but it also involves the most performance overhead.
tcpInstructs the NFS mount to use the TCP protocol.
udpInstructs the NFS mount to use the UDP protocol.
For more information, see man mount and man nfs.
8.5. Starting and Stopping the NFS Server
Prerequisites
For servers that support NFSv2 or NFSv3 connections, the rpcbind[1] service must be running. To verify that rpcbind is active, use the following command:
$ systemctl status rpcbindOn Red Hat Enterprise Linux 7.0, if your NFS server exports NFSv3 and is enabled to start at boot, you need to manually start and enable the nfs-lock service:
# systemctl start nfs-lock # systemctl enable nfs-lockOn Red Hat Enterprise Linux 7.1 and later, nfs-lock starts automatically if needed, and an attempt to enable it manually fails.
Procedures
To start an NFS server, use the following command:
# systemctl start nfsTo enable NFS to start at boot, use the following command:
# systemctl enable nfsTo stop the server, use:
# systemctl stop nfsThe restart option is a shorthand way of stopping and then starting NFS. This is the most efficient way to make configuration changes take effect after editing the configuration file for NFS. To restart the server type:
# systemctl restart nfsAfter you edit the /etc/sysconfig/nfs file, restart the nfs-config service by running the following command for the new values to take effect:
# systemctl restart nfs-configThe try-restart command only starts nfs if it is currently running. This command is the equivalent of condrestart (conditional restart) in Red Hat init scripts and is useful because it does not start the daemon if NFS is not running.
To conditionally restart the server, type:
# systemctl try-restart nfsTo reload the NFS server configuration file without restarting the service type:
# systemctl reload nfs
8.6. Configuring the NFS Server
There are two ways to configure exports on an NFS server:
Manually editing the NFS configuration file, that is, /etc/exports, and
Through the command line, that is, by using the command exportfs
8.6.1. The /etc/exports Configuration File
The /etc/exports file controls which file systems are exported to remote hosts and specifies options. It follows the following syntax rules:
Blank lines are ignored.
To add a comment, start a line with the hash mark (#).
You can wrap long lines with a backslash (\).
Each exported file system should be on its own individual line.
Any lists of authorized hosts placed after an exported file system must be separated by space characters.
Options for each of the hosts must be placed in parentheses directly after the host identifier, without any spaces separating the host and the first parenthesis.
Each entry for an exported file system has the following structure:
export host(options)The aforementioned structure uses the following variables:
export
The directory being exported
hostThe host or network to which the export is being shared
optionsThe options to be used for host
It is possible to specify multiple hosts, along with specific options for each host. To do so, list them on the same line as a space-delimited list, with each hostname followed by its respective options (in parentheses), as in:
export host1(options1) host2(options2) host3(options3)In its simplest form, the /etc/exports file only specifies the exported directory and the hosts permitted to access it, as in the following example:
Example 8.6. The /etc/exports File
/exported/directory bob.example.com
Here, bob.example.com can mount /exported/directory/ from the NFS server. Because no options are specified in this example, NFS uses default settings.
The default settings are:
ro
The exported file system is read-only. Remote hosts cannot change the data shared on the file system. To allow hosts to make changes to the file system (that is, read and write), specify the rw option.
syncThe NFS server will not reply to requests before changes made by previous requests are written to disk. To enable asynchronous writes instead, specify the option async.
wdelayThe NFS server will delay writing to the disk if it suspects another write request is imminent. This can improve performance as it reduces the number of times the disk must be accessed by separate write commands, thereby reducing write overhead. To disable this, specify the no_wdelay. no_wdelay is only available if the default sync option is also specified.
root_squashThis prevents root users connected remotely (as opposed to locally) from having root privileges; instead, the NFS server assigns them the user ID nfsnobody. This effectively "squashes" the power of the remote root user to the lowest local user, preventing possible unauthorized writes on the remote server. To disable root squashing, specify no_root_squash.
To squash every remote user (including root), use all_squash. To specify the user and group IDs that the NFS server should assign to remote users from a particular host, use the anonuid and anongid options, respectively, as in:
export host(anonuid=uid,anongid=gid)Here, uid and gid are user ID number and group ID number, respectively. The anonuid and anongid options allow you to create a special user and group account for remote NFS users to share.
By default, access control lists (ACLs) are supported by NFS under Red Hat Enterprise Linux. To disable this feature, specify the no_acl option when exporting the file system.
Each default for every exported file system must be explicitly overridden. For example, if the rw option is not specified, then the exported file system is shared as read-only. The following is a sample line from /etc/exports which overrides two default options:
/another/exported/directory 192.168.0.3(rw,async)
In this example 192.168.0.3 can mount /another/exported/directory/ read and write and all writes to disk are asynchronous. For more information on exporting options, see man exportfs.
Other options are available where no default value is specified. These include the ability to disable sub-tree checking, allow access from insecure ports, and allow insecure file locks (necessary for certain early NFS client implementations). For more information on these less-used options, see man exports.
The format of the /etc/exports file is very precise, particularly in regards to use of the space character. Remember to always separate exported file systems from hosts and hosts from one another with a space character. However, there should be no other space characters in the file except on comment lines.
For example, the following two lines do not mean the same thing:
/home bob.example.com(rw) /home bob.example.com (rw)The first line allows only users from bob.example.com read and write access to the /home directory. The second line allows users from bob.example.com to mount the directory as read-only (the default), while the rest of the world can mount it read/write.
8.6.2. The exportfs Command
Every file system being exported to remote users with NFS, as well as the access level for those file systems, are listed in the /etc/exports file. When the nfs service starts, the /usr/sbin/exportfs command launches and reads this file, passes control to rpc.mountd (if NFSv3) for the actual mounting process, then to rpc.nfsd where the file systems are then available to remote users.
When issued manually, the /usr/sbin/exportfs command allows the root user to selectively export or unexport directories without restarting the NFS service. When given the proper options, the /usr/sbin/exportfs command writes the exported file systems to /var/lib/nfs/xtab. Since rpc.mountd refers to the xtab file when deciding access privileges to a file system, changes to the list of exported file systems take effect immediately.
The following is a list of commonly-used options available for /usr/sbin/exportfs:
-r
Causes all directories listed in /etc/exports to be exported by constructing a new export list in /var/lib/nfs/etab. This option effectively refreshes the export list with any changes made to /etc/exports.
-aCauses all directories to be exported or unexported, depending on what other options are passed to /usr/sbin/exportfs. If no other options are specified, /usr/sbin/exportfs exports all file systems specified in /etc/exports.
-o file-systemsSpecifies directories to be exported that are not listed in /etc/exports. Replace file-systems with additional file systems to be exported. These file systems must be formatted in the same way they are specified in /etc/exports. This option is often used to test an exported file system before adding it permanently to the list of file systems to be exported. For more information on /etc/exports syntax, see Section 8.6.1, “The /etc/exports Configuration File”.
-iIgnores /etc/exports; only options given from the command line are used to define exported file systems.
-uUnexports all shared directories. The command /usr/sbin/exportfs -ua suspends NFS file sharing while keeping all NFS daemons up. To re-enable NFS sharing, use exportfs -r.
-vVerbose operation, where the file systems being exported or unexported are displayed in greater detail when the exportfs command is executed.
If no options are passed to the exportfs command, it displays a list of currently exported file systems. For more information about the exportfs command, see man exportfs.
8.6.2.1. Using exportfs with NFSv4
In Red Hat Enterprise Linux 7, no extra steps are required to configure NFSv4 exports as any filesystems mentioned are automatically available to NFSv3 and NFSv4 clients using the same path. This was not the case in previous versions.
To prevent clients from using NFSv4, turn it off by setting RPCNFSDARGS= -N 4 in /etc/sysconfig/nfs.
8.6.3. Running NFS Behind a Firewall
NFS requires rpcbind, which dynamically assigns ports for RPC services and can cause issues for configuring firewall rules. To allow clients to access NFS shares behind a firewall, edit the /etc/sysconfig/nfs file to set which ports the RPC services run on. To allow clients to access RPC Quota through a firewall, see Section 8.6.4, “Accessing RPC Quota through a Firewall”.
The /etc/sysconfig/nfs file does not exist by default on all systems. If /etc/sysconfig/nfs does not exist, create it and specify the following:
RPCMOUNTDOPTS="-p port"
This adds "-p port" to the rpc.mount command line: rpc.mount -p port.
To specify the ports to be used by the nlockmgr service, set the port number for the nlm_tcpport and nlm_udpport options in the /etc/modprobe.d/lockd.conf file.
If NFS fails to start, check /var/log/messages. Commonly, NFS fails to start if you specify a port number that is already in use. After editing /etc/sysconfig/nfs, you need to restart the nfs-config service for the new values to take effect in Red Hat Enterprise Linux 7.2 and prior by running:
# systemctl restart nfs-configThen, restart the NFS server:
# systemctl restart nfs-serverRun rpcinfo -p to confirm the changes have taken effect.
To allow NFSv4.0 callbacks to pass through firewalls set /proc/sys/fs/nfs/nfs_callback_tcpport and allow the server to connect to that port on the client.
This process is not needed for NFSv4.1 or higher, and the other ports for mountd, statd, and lockd are not required in a pure NFSv4 environment.
8.6.3.1. Discovering NFS exports
There are two ways to discover which file systems an NFS server exports.
On any server that supports NFSv3, use the showmount command:
$ showmount -e myserver Export list for mysever /exports/foo /exports/barOn any server that supports NFSv4, mount the root directory and look around.
# mount myserver:/ /mnt/ # cd /mnt/ exports # ls exports foo bar
On servers that support both NFSv4 and NFSv3, both methods work and give the same results.
Before Red Hat Enterprise Linux 6 on older NFS servers, depending on how they are configured, it is possible to export filesystems to NFSv4 clients at different paths. Because these servers do not enable NFSv4 by default, this should not be a problem.
8.6.4. Accessing RPC Quota through a Firewall
If you export a file system that uses disk quotas, you can use the quota Remote Procedure Call (RPC) service to provide disk quota data to NFS clients.
Procedure 8.1. Making RPC Quota Accessible Behind a Firewall
To enable the rpc-rquotad service, use the following command:
# systemctl enable rpc-rquotadTo start the rpc-rquotad service, use the following command:
# systemctl start rpc-rquotadNote that rpc-rquotad is, if enabled, started automatically after starting the nfs-server service.
To make the quota RPC service accessible behind a firewall, UDP or TCP port 875 need to be open. The default port number is defined in the /etc/services file.
You can override the default port number by appending -p port-number to the RPCRQUOTADOPTS variable in the /etc/sysconfig/rpc-rquotad file.
Restart rpc-rquotad for changes in the /etc/sysconfig/rpc-rquotad file to take effect:
# systemctl restart rpc-rquotad
Setting Quotas from Remote Hosts
By default, quotas can only be read by remote hosts. To allow setting quotas, append the -S option to the RPCRQUOTADOPTS variable in the /etc/sysconfig/rpc-rquotad file.
Restart rpc-rquotad for changes in the /etc/sysconfig/rpc-rquotad file to take effect:
# systemctl restart rpc-rquotad8.6.5. Hostname Formats
The host(s) can be in the following forms:
Single machine
A fully-qualified domain name (that can be resolved by the server), hostname (that can be resolved by the server), or an IP address.
Series of machines specified with wildcardsUse the * or ? character to specify a string match. Wildcards are not to be used with IP addresses; however, they may accidentally work if reverse DNS lookups fail. When specifying wildcards in fully qualified domain names, dots (.) are not included in the wildcard. For example, *.example.com includes one.example.com but does not include one.two.example.com.
IP networksUse a.b.c.d/z, where a.b.c.d is the network and z is the number of bits in the netmask (for example 192.168.0.0/24). Another acceptable format is a.b.c.d/netmask, where a.b.c.d is the network and netmask is the netmask (for example, 192.168.100.8/255.255.255.0).
NetgroupsUse the format @group-name, where group-name is the NIS netgroup name.
8.6.6. Enabling NFS over RDMA (NFSoRDMA)
The remote direct memory access (RDMA) service works automatically in Red Hat Enterprise Linux 7 if there is RDMA-capable hardware present.
To enable NFS over RDMA:
Install the rdma and rdma-core packages.
The /etc/rdma/rdma.conf file contains a line that sets XPRTRDMA_LOAD=yes by default, which requests the rdma service to load the NFSoRDMA client module.
To enable automatic loading of NFSoRDMA server modules, add SVCRDMA_LOAD=yes on a new line in /etc/rdma/rdma.conf.
RPCNFSDARGS="--rdma=20049" in the /etc/sysconfig/nfs file specifies the port number on which the NFSoRDMA service listens for clients. RFC 5667 specifies that servers must listen on port 20049 when providing NFSv4 services over RDMA.
Restart the nfs service after editing the /etc/rdma/rdma.conf file:
# systemctl restart nfsNote that with earlier kernel versions, a system reboot is needed after editing /etc/rdma/rdma.conf for the changes to take effect.
8.6.7. Configuring an NFSv4-only Server
By default, the NFS server supports NFSv2, NFSv3, and NFSv4 connections in Red Hat Enterprise Linux 7. However, you can also configure NFS to support only NFS version 4.0 and later. This minimizes the number of open ports and running services on the system, because NFSv4 does not require the rpcbind service to listen on the network.
When your NFS server is configured as NFSv4-only, clients attempting to mount shares using NFSv2 or NFSv3 fail with an error like the following:
Requested NFS version or transport protocol is not supported.Procedure 8.2. Configuring an NFSv4-only Server
To configure your NFS server to support only NFS version 4.0 and later:
Disable NFSv2, NFSv3, and UDP by adding the following line to the /etc/sysconfig/nfs configuration file:
RPCNFSDARGS="-N 2 -N 3 -U"Optionally, disable listening for the RPCBIND, MOUNT, and NSM protocol calls, which are not necessary in the NFSv4-only case.
The effects of disabling these options are:
Clients that attempt to mount shares from your server using NFSv2 or NFSv3 become unresponsive.
The NFS server itself is unable to mount NFSv2 and NFSv3 file systems.
To disable these options:
Add the following to the /etc/sysconfig/nfs file:
RPCMOUNTDOPTS="-N 2 -N 3"Disable related services:
# systemctl mask --now rpc-statd.service rpcbind.service rpcbind.socket
Restart the NFS server:
# systemctl restart nfsThe changes take effect as soon as you start or restart the NFS server.
Verifying the NFSv4-only Configuration
You can verify that your NFS server is configured in the NFSv4-only mode by using the netstat utility.
The following is an example netstat output on an NFSv4-only server; listening for RPCBIND, MOUNT, and NSM is also disabled. Here, nfs is the only listening NFS service:
# netstat -ltu Active Internet connections (only servers) Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 0.0.0.0:nfs 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:ssh 0.0.0.0:* LISTEN tcp 0 0 localhost:smtp 0.0.0.0:* LISTEN tcp6 0 0 [::]:nfs [::]:* LISTEN tcp6 0 0 [::]:12432 [::]:* LISTEN tcp6 0 0 [::]:12434 [::]:* LISTEN tcp6 0 0 localhost:7092 [::]:* LISTEN tcp6 0 0 [::]:ssh [::]:* LISTEN udp 0 0 localhost:323 0.0.0.0:* udp 0 0 0.0.0.0:bootpc 0.0.0.0:* udp6 0 0 localhost:323 [::]:*In comparison, the netstat output before configuring an NFSv4-only server includes the sunrpc and mountd services:
# netstat -ltu Active Internet connections (only servers) Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 0.0.0.0:nfs 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:36069 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:52364 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:sunrpc 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:mountd 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:ssh 0.0.0.0:* LISTEN tcp 0 0 localhost:smtp 0.0.0.0:* LISTEN tcp6 0 0 [::]:34941 [::]:* LISTEN tcp6 0 0 [::]:nfs [::]:* LISTEN tcp6 0 0 [::]:sunrpc [::]:* LISTEN tcp6 0 0 [::]:mountd [::]:* LISTEN tcp6 0 0 [::]:12432 [::]:* LISTEN tcp6 0 0 [::]:56881 [::]:* LISTEN tcp6 0 0 [::]:12434 [::]:* LISTEN tcp6 0 0 localhost:7092 [::]:* LISTEN tcp6 0 0 [::]:ssh [::]:* LISTEN udp 0 0 localhost:323 0.0.0.0:* udp 0 0 0.0.0.0:37190 0.0.0.0:* udp 0 0 0.0.0.0:876 0.0.0.0:* udp 0 0 localhost:877 0.0.0.0:* udp 0 0 0.0.0.0:mountd 0.0.0.0:* udp 0 0 0.0.0.0:38588 0.0.0.0:* udp 0 0 0.0.0.0:nfs 0.0.0.0:* udp 0 0 0.0.0.0:bootpc 0.0.0.0:* udp 0 0 0.0.0.0:sunrpc 0.0.0.0:* udp6 0 0 localhost:323 [::]:* udp6 0 0 [::]:57683 [::]:* udp6 0 0 [::]:876 [::]:* udp6 0 0 [::]:mountd [::]:* udp6 0 0 [::]:40874 [::]:* udp6 0 0 [::]:nfs [::]:* udp6 0 0 [::]:sunrpc [::]:*
8.7. Securing NFS
NFS is suitable for transparent sharing of entire file systems with a large number of known hosts. However, with ease-of-use comes a variety of potential security problems. To minimize NFS security risks and protect data on the server, consider the following sections when exporting NFS file systems on a server or mounting them on a client.
8.7.1. NFS Security with AUTH_SYS and Export Controls
Traditionally, NFS has given two options in order to control access to exported files.
First, the server restricts which hosts are allowed to mount which file systems either by IP address or by host name.
Second, the server enforces file system permissions for users on NFS clients in the same way it does local users. Traditionally it does this using AUTH_SYS (also called AUTH_UNIX) which relies on the client to state the UID and GID's of the user. Be aware that this means a malicious or misconfigured client can easily get this wrong and allow a user access to files that it should not.
To limit the potential risks, administrators often allow read-only access or squash user permissions to a common user and group ID. Unfortunately, these solutions prevent the NFS share from being used in the way it was originally intended.
Additionally, if an attacker gains control of the DNS server used by the system exporting the NFS file system, the system associated with a particular hostname or fully qualified domain name can be pointed to an unauthorized machine. At this point, the unauthorized machine is the system permitted to mount the NFS share, since no username or password information is exchanged to provide additional security for the NFS mount.
Wildcards should be used sparingly when exporting directories through NFS, as it is possible for the scope of the wildcard to encompass more systems than intended.
It is also possible to restrict access to the rpcbind[1] service with TCP wrappers. Creating rules with iptables can also limit access to ports used by rpcbind, rpc.mountd, and rpc.nfsd.
For more information on securing NFS and rpcbind, refer to man iptables.
8.7.2. NFS Security with AUTH_GSS
NFSv4 revolutionized NFS security by mandating the implementation of RPCSEC_GSS and the Kerberos version 5 GSS-API mechanism. However, RPCSEC_GSS and the Kerberos mechanism are also available for all versions of NFS. In FIPS mode, only FIPS-approved algorithms can be used.
Unlike AUTH_SYS, with the RPCSEC_GSS Kerberos mechanism, the server does not depend on the client to correctly represent which user is accessing the file. Instead, cryptography is used to authenticate users to the server, which prevents a malicious client from impersonating a user without having that user's Kerberos credentials. Using the RPCSEC_GSS Kerberos mechanism is the most straightforward way to secure mounts because after configuring Kerberos, no additional setup is needed.
Configuring Kerberos
Before configuring an NFSv4 Kerberos-aware server, you need to install and configure a Kerberos Key Distribution Centre (KDC). Kerberos is a network authentication system that allows clients and servers to authenticate to each other by using symmetric encryption and a trusted third party, the KDC. Red Hat recommends using Identity Management (IdM) for setting up Kerberos.
Procedure 8.3. Configuring an NFS Server and Client for IdM to Use RPCSEC_GSS
Create the nfs/hostname.domain@REALM principal on the NFS server side.
Create the host/hostname.domain@REALM principal on both the server and the client side.
Add the corresponding keys to keytabs for the client and server.
On the server side, use the sec= option to enable the wanted security flavors. To enable all security flavors as well as non-cryptographic mounts:
/export *(sec=sys:krb5:krb5i:krb5p)Valid security flavors to use with the sec= option are:
sys: no cryptographic protection, the default
krb5: authentication only
krb5i: integrity protection
krb5p: privacy protection
On the client side, add sec=krb5 (or sec=krb5i, or sec=krb5p, depending on the setup) to the mount options:
# mount -o sec=krb5 server:/export /mnt
Additional Resources
If you need to write files as root on the Kerberos-secured NFS share and keep root ownership on these files, see //access.redhat.com/articles/4040141. Note that this configuration is not recommended.
For further information on the RPCSEC_GSS framework, including how gssproxy and rpc.gssd inter-operate, see the GSSD flow description.
8.7.2.1. NFS Security with NFSv4
NFSv4 includes ACL support based on the Microsoft Windows NT model, not the POSIX model, because of the Microsoft Windows NT model's features and wide deployment.
Another important security feature of NFSv4 is the removal of the use of the MOUNT protocol for mounting file systems. The MOUNT protocol presented a security risk because of the way the protocol processed file handles.
8.7.3. File Permissions
Once the NFS file system is mounted as either read or read and write by a remote host, the only protection each shared file has is its permissions. If two users that share the same user ID value mount the same NFS file system, they can modify each others' files. Additionally, anyone logged in as root on the client system can use the su - command to access any files with the NFS share.
By default, access control lists (ACLs) are supported by NFS under Red Hat Enterprise Linux. Red Hat recommends that this feature is kept enabled.
By default, NFS uses root squashing when exporting a file system. This sets the user ID of anyone accessing the NFS share as the root user on their local machine to nobody. Root squashing is controlled by the default option root_squash; for more information about this option, refer to Section 8.6.1, “The /etc/exports Configuration File”. If possible, never disable root squashing.
When exporting an NFS share as read-only, consider using the all_squash option. This option makes every user accessing the exported file system take the user ID of the nfsnobody user.
8.8. NFS and rpcbind
The following section only applies to NFSv3 implementations that require the rpcbind service for backward compatibility.
The rpcbind[1] utility maps RPC services to the ports on which they listen. RPC processes notify rpcbind when they start, registering the ports they are listening on and the RPC program numbers they expect to serve. The client system then contacts rpcbind on the server with a particular RPC program number. The rpcbind service redirects the client to the proper port number so it can communicate with the requested service.
Because RPC-based services rely on rpcbind to make all connections with incoming client requests, rpcbind must be available before any of these services start.
The rpcbind service uses TCP wrappers for access control, and access control rules for rpcbind affect all RPC-based services. Alternatively, it is possible to specify access control rules for each of the NFS RPC daemons. The man pages for rpc.mountd and rpc.statd contain information regarding the precise syntax for these rules.
8.8.1. Troubleshooting NFS and rpcbind
Because rpcbind[1] provides coordination between RPC services and the port numbers used to communicate with them, it is useful to view the status of current RPC services using rpcbind when troubleshooting. The rpcinfo command shows each RPC-based service with port numbers, an RPC program number, a version number, and an IP protocol type (TCP or UDP).
To make sure the proper NFS RPC-based services are enabled for rpcbind, use the following command:
# rpcinfo -pExample 8.7. rpcinfo -p command output
The following is sample output from this command:
program vers proto port service 100021 1 udp 32774 nlockmgr 100021 3 udp 32774 nlockmgr 100021 4 udp 32774 nlockmgr 100021 1 tcp 34437 nlockmgr 100021 3 tcp 34437 nlockmgr 100021 4 tcp 34437 nlockmgr 100011 1 udp 819 rquotad 100011 2 udp 819 rquotad 100011 1 tcp 822 rquotad 100011 2 tcp 822 rquotad 100003 2 udp 2049 nfs 100003 3 udp 2049 nfs 100003 2 tcp 2049 nfs 100003 3 tcp 2049 nfs 100005 1 udp 836 mountd 100005 1 tcp 839 mountd 100005 2 udp 836 mountd 100005 2 tcp 839 mountd 100005 3 udp 836 mountd 100005 3 tcp 839 mountdIf one of the NFS services does not start up correctly, rpcbind will be unable to map RPC requests from clients for that service to the correct port. In many cases, if NFS is not present in rpcinfo output, restarting NFS causes the service to correctly register with rpcbind and begin working.
For more information and a list of options on rpcinfo, see its man page.
8.9. pNFS
Support for Parallel NFS (pNFS) as part of the NFS v4.1 standard is available as of Red Hat Enterprise Linux 6.4. The pNFS architecture improves the scalability of NFS, with possible improvements to performance. That is, when a server implements pNFS as well, a client is able to access data through multiple servers concurrently. It supports three storage protocols or layouts: files, objects, and blocks.
The protocol allows for three possible pNFS layout types: files, objects, and blocks. While the Red Hat Enterprise Linux 6.4 client only supported the files layout type, Red Hat Enterprise Linux 7 supports the files layout type, with objects and blocks layout types being included as a technology preview.
pNFS Flex Files
Flexible Files is a new layout for pNFS that enables the aggregation of standalone NFSv3 and NFSv4 servers into a scale out name space. The Flex Files feature is part of the NFSv4.2 standard as described in the RFC 7862 specification.
Red Hat Enterprise Linux can mount NFS shares from Flex Files servers since Red Hat Enterprise Linux 7.4.
Mounting pNFS Shares
To enable pNFS functionality, mount shares from a pNFS-enabled server with NFS version 4.1 or later:
# mount -t nfs -o v4.1 server:/remote-export /local-directoryAfter the server is pNFS-enabled, the nfs_layout_nfsv41_files kernel is automatically loaded on the first mount. The mount entry in the output should contain minorversion=1. Use the following command to verify the module was loaded:
$ lsmod | grep nfs_layout_nfsv41_filesTo mount an NFS share with the Flex Files feature from a server that supports Flex Files, use NFS version 4.2 or later:
# mount -t nfs -o v4.2 server:/remote-export /local-directoryVerify that the nfs_layout_flexfiles module has been loaded:
$ lsmod | grep nfs_layout_flexfiles
Additional Resources
8.10. Enabling pNFS SCSI Layouts in NFS
You can configure the NFS server and client to use the pNFS SCSI layout for accessing data. pNFS SCSI is beneficial in use cases that involve longer-duration single-client access to a file.
Prerequisites
Both the client and the server must be able to send SCSI commands to the same block device. That is, the block device must be on a shared SCSI bus.
The block device must contain an XFS file system.
The SCSI device must support SCSI Persistent Reservations as described in the SCSI-3 Primary Commands specification.
8.10.1. pNFS SCSI Layouts
The SCSI layout builds on the work of pNFS block layouts. The layout is defined across SCSI devices. It contains a sequential series of fixed-size blocks as logical units (LUs) that must be capable of supporting SCSI persistent reservations. The LU devices are identified by their SCSI device identification.
pNFS SCSI performs well in use cases that involve longer-duration single-client access to a file. An example might be a mail server or a virtual machine housing a cluster.
Operations Between the Client and the Server
When an NFS client reads from a file or writes to it, the client performs a LAYOUTGET operation. The server responds with the location of the file on the SCSI device. The client might need to perform an additional operation of GETDEVICEINFO to determine which SCSI device to use. If these operations work correctly, the client can issue I/O requests directly to the SCSI device instead of sending READ and WRITE operations to the server.
Errors or contention between clients might cause the server to recall layouts or not issue them to the clients. In those cases, the clients fall back to issuing READ and WRITE operations to the server instead of sending I/O requests directly to the SCSI device.
Device Reservations
pNFS SCSI handles fencing through the assignment of reservations. Before the server issues layouts to clients, it reserves the SCSI device to ensure that only registered clients may access the device. If a client can issue commands to that SCSI device but is not registered with the device, many operations from the client on that device fail. For example, the blkid command on the client fails to show the UUID of the XFS file system if the server has not given a layout for that device to the client.
The server does not remove its own persistent reservation. This protects the data within the file system on the device across restarts of clients and servers. In order to repurpose the SCSI device, you might need to manually remove the persistent reservation on the NFS server.
8.10.2. Checking for a SCSI Device Compatible with pNFS
This procedure checks if a SCSI device supports the pNFS SCSI layout.
Prerequisites
Install the sg3_utils package:
# yum install sg3_utils
Procedure 8.4. Checking for a SCSI Device Compatible with pNFS
On both the server and client, check for the proper SCSI device support:
# sg_persist --in --report-capabilities --verbose path-to-scsi-deviceEnsure that the Persist Through Power Loss Active (PTPL_A) bit is set.
Example 8.8. A SCSI device that supports pNFS SCSI
The following is an example of sg_persist output for a SCSI device that supports pNFS SCSI. The PTPL_A bit reports 1.
inquiry cdb: 12 00 00 00 24 00 Persistent Reservation In cmd: 5e 02 00 00 00 00 00 20 00 00 LIO-ORG block11 4.0 Peripheral device type: disk Report capabilities response: Compatible Reservation Handling(CRH): 1 Specify Initiator Ports Capable(SIP_C): 1 All Target Ports Capable(ATP_C): 1 Persist Through Power Loss Capable(PTPL_C): 1 Type Mask Valid(TMV): 1 Allow Commands: 1 Persist Through Power Loss Active(PTPL_A): 1 Support indicated in Type mask: Write Exclusive, all registrants: 1 Exclusive Access, registrants only: 1 Write Exclusive, registrants only: 1 Exclusive Access: 1 Write Exclusive: 1 Exclusive Access, all registrants: 1
Additional Resources
The sg_persist(8) man page
8.10.3. Setting up pNFS SCSI on the Server
This procedure configures an NFS server to export a pNFS SCSI layout.
Procedure 8.5. Setting up pNFS SCSI on the Server
On the server, mount the XFS file system created on the SCSI device.
Configure the NFS server to export NFS version 4.1 or higher. Set the following option in the [nfsd] section of the /etc/nfs.conf file:
[nfsd] vers4.1=yConfigure the NFS server to export the XFS file system over NFS with the pnfs option:
Example 8.9. An Entry in /etc/exports to Export pNFS SCSI
The following entry in the /etc/exports configuration file exports the file system mounted at /exported/directory/ to the allowed.example.com client as a pNFS SCSI layout:
/exported/directory allowed.example.com(pnfs)
Additional Resources
8.10.4. Setting up pNFS SCSI on the Client
This procedure configures an NFS client to mount a pNFS SCSI layout.
Prerequisites
Procedure 8.6. Setting up pNFS SCSI on the Client
On the client, mount the exported XFS file system using NFS version 4.1 or higher:
# mount -t nfs -o nfsvers=4.1 host:/remote/export /local/directoryDo not mount the XFS file system directly without NFS.
Additional Resources
8.10.5. Releasing the pNFS SCSI Reservation on the Server
This procedure releases the persistent reservation that an NFS server holds on a SCSI device. This enables you to repurpose the SCSI device when you no longer need to export pNFS SCSI.
You must remove the reservation from the server. It cannot be removed from a different IT Nexus.
Prerequisites
Install the sg3_utils package:
# yum install sg3_utils
Procedure 8.7. Releasing the pNFS SCSI Reservation on the Server
Query an existing reservation on the server:
# sg_persist --read-reservation path-to-scsi-deviceExample 8.10. Querying a Reservation on /dev/sda
# sg_persist --read-reservation /dev/sda LIO-ORG block_1 4.0 Peripheral device type: disk PR generation=0x8, Reservation follows: Key=0x100000000000000 scope: LU_SCOPE, type: Exclusive Access, registrants only
Remove the existing registration on the server:
# sg_persist --out \ --release \ --param-rk=reservation-key \ --prout-type=6 \ path-to-scsi-deviceExample 8.11. Removing a Reservation on /dev/sda
# sg_persist --out \ --release \ --param-rk=0x100000000000000 \ --prout-type=6 \ /dev/sda LIO-ORG block_1 4.0 Peripheral device type: disk
Additional Resources
The sg_persist(8) man page
8.10.6. Monitoring pNFS SCSI Layouts Functionality
You can monitor if the pNFS client and server exchange proper pNFS SCSI operations or if they fall back on regular NFS operations.
Prerequisites
A pNFS SCSI client and server are configured.
8.10.6.1. Checking pNFS SCSI Operations from the Server Using nfsstat
This procedure uses the nfsstat utility to monitor pNFS SCSI operations from the server.
Procedure 8.8. Checking pNFS SCSI Operations from the Server Using nfsstat
Monitor the operations serviced from the server:
# watch --differences \ "nfsstat --server | egrep --after-context=1 read\|write\|layout" Every 2.0s: nfsstat --server | egrep --after-context=1 read\|write\|layout putrootfh read readdir readlink remove rename 2 0% 0 0% 1 0% 0 0% 0 0% 0 0% -- setcltidconf verify write rellockowner bc_ctl bind_conn 0 0% 0 0% 0 0% 0 0% 0 0% 0 0% -- getdevlist layoutcommit layoutget layoutreturn secinfononam sequence 0 0% 29 1% 49 1% 5 0% 0 0% 2435 86%The client and server use pNFS SCSI operations when:
The layoutget, layoutreturn, and layoutcommit counters increment. This means that the server is serving layouts.
The server read and write counters do not increment. This means that the clients are performing I/O requests directly to the SCSI devices.
8.10.6.2. Checking pNFS SCSI Operations from the Client Using mountstats
This procedure uses the /proc/self/mountstats file to monitor pNFS SCSI operations from the client.
Procedure 8.9. Checking pNFS SCSI Operations from the Client Using mountstats
List the per-mount operation counters:
# cat /proc/self/mountstats \ | awk /scsi_lun_0/,/^$/ \ | egrep device\|READ\|WRITE\|LAYOUT device 192.168.122.73:/exports/scsi_lun_0 mounted on /mnt/rhel7/scsi_lun_0 with fstype nfs4 statvers=1.1 nfsv4: bm0=0xfdffbfff,bm1=0x40f9be3e,bm2=0x803,acl=0x3,sessions,pnfs=LAYOUT_SCSI READ: 0 0 0 0 0 0 0 0 WRITE: 0 0 0 0 0 0 0 0 READLINK: 0 0 0 0 0 0 0 0 READDIR: 0 0 0 0 0 0 0 0 LAYOUTGET: 49 49 0 11172 9604 2 19448 19454 LAYOUTCOMMIT: 28 28 0 7776 4808 0 24719 24722 LAYOUTRETURN: 0 0 0 0 0 0 0 0 LAYOUTSTATS: 0 0 0 0 0 0 0 0In the results:
The LAYOUT statistics indicate requests where the client and server use pNFS SCSI operations.
The READ and WRITE statistics indicate requests where the client and server fall back to NFS operations.
8.11. NFS References
Administering an NFS server can be a challenge. Many options, including quite a few not mentioned in this chapter, are available for exporting or mounting NFS shares. For more information, see the following sources:
Installed Documentation
man mount — Contains a comprehensive look at mount options for both NFS server and client configurations.
man fstab — Provides detail for the format of the /etc/fstab file used to mount file systems at boot-time.
man nfs — Provides details on NFS-specific file system export and mount options.
man exports — Shows common options used in the /etc/exports file when exporting NFS file systems.
Useful Websites
//linux-nfs.org — The current site for developers where project status updates can be viewed.
Chapter 9. Server Message Block (SMB)
The Server Message Block (SMB) protocol implements an application-layer network protocol used to access resources on a server, such as file shares and shared printers. On Microsoft Windows, SMB is implemented by default. If you run Red Hat Enterprise Linux, use Samba to provide SMB shares and the cifs-utils utility to mount SMB shares from a remote server.
In the context of SMB, you sometimes read about the Common Internet File System (CIFS) protocol, which is a dialect of SMB. Both the SMB and CIFS protocol are supported and the kernel module and utilities involved in mounting SMB and CIFS shares both use the name cifs.
Chapter 10. FS-Cache
FS-Cache is a persistent local cache that can be used by file systems to take data retrieved from over the network and cache it on local disk. This helps minimize network traffic for users accessing data from a file system mounted over the network (for example, NFS).
The following diagram is a high-level illustration of how FS-Cache works:
Figure 10.1. FS-Cache Overview
FS-Cache is designed to be as transparent as possible to the users and administrators of a system. Unlike cachefs on Solaris, FS-Cache allows a file system on a server to interact directly with a client's local cache without creating an overmounted file system. With NFS, a mount option instructs the client to mount the NFS share with FS-cache enabled.
FS-Cache does not alter the basic operation of a file system that works over the network - it merely provides that file system with a persistent place in which it can cache data. For instance, a client can still mount an NFS share whether or not FS-Cache is enabled. In addition, cached NFS can handle files that won't fit into the cache (whether individually or collectively) as files can be partially cached and do not have to be read completely up front. FS-Cache also hides all I/O errors that occur in the cache from the client file system driver.
To provide caching services, FS-Cache needs a cache back end. A cache back end is a storage driver configured to provide caching services (i.e. cachefiles). In this case, FS-Cache requires a mounted block-based file system that supports bmap and extended attributes (e.g. ext3) as its cache back end.
FS-Cache cannot arbitrarily cache any file system, whether through the network or otherwise: the shared file system's driver must be altered to allow interaction with FS-Cache, data storage/retrieval, and metadata setup and validation. FS-Cache needs indexing keys and coherency data from the cached file system to support persistence: indexing keys to match file system objects to cache objects, and coherency data to determine whether the cache objects are still valid.
In Red Hat Enterprise Linux 7, the cachefilesd package is not installed by default and needs to be installed manually.
10.1. Performance Guarantee
FS-Cache does not guarantee increased performance, however it ensures consistent performance by avoiding network congestion. Using a cache back end incurs a performance penalty: for example, cached NFS shares add disk accesses to cross-network lookups. While FS-Cache tries to be as asynchronous as possible, there are synchronous paths (e.g. reads) where this isn't possible.
For example, using FS-Cache to cache an NFS share between two computers over an otherwise unladen GigE network will not demonstrate any performance improvements on file access. Rather, NFS requests would be satisfied faster from server memory rather than from local disk.
The use of FS-Cache, therefore, is a compromise between various factors. If FS-Cache is being used to cache NFS traffic, for instance, it may slow the client down a little, but massively reduce the network and server loading by satisfying read requests locally without consuming network bandwidth.
10.2. Setting up a Cache
Currently, Red Hat Enterprise Linux 7 only provides the cachefiles caching back end. The cachefilesd daemon initiates and manages cachefiles. The /etc/cachefilesd.conf file controls how cachefiles provides caching services.
The first setting to configure in a cache back end is which directory to use as a cache. To configure this, use the following parameter:
$ dir /path/to/cacheTypically, the cache back end directory is set in /etc/cachefilesd.conf as /var/cache/fscache, as in:
$ dir /var/cache/fscacheIf you want to change the cache back end directory, the selinux context must be same as /var/cache/fscache:
# semanage fcontext -a -e /var/cache/fscache /path/to/cache # restorecon -Rv /path/to/cacheReplace /path/to/cache with the directory name while setting up cache.
If the given commands for setting selinux context did not work, use the following commands:
# semanage permissive -a cachefilesd_t # semanage permissive -a cachefiles_kernel_tFS-Cache will store the cache in the file system that hosts /path/to/cache. On a laptop, it is advisable to use the root file system (/) as the host file system, but for a desktop machine it would be more prudent to mount a disk partition specifically for the cache.
File systems that support functionalities required by FS-Cache cache back end include the Red Hat Enterprise Linux 7 implementations of the following file systems:
ext3 (with extended attributes enabled)
ext4
Btrfs
XFS
The host file system must support user-defined extended attributes; FS-Cache uses these attributes to store coherency maintenance information. To enable user-defined extended attributes for ext3 file systems (i.e. device), use:
# tune2fs -o user_xattr /dev/deviceAlternatively, extended attributes for a file system can be enabled at mount time, as in:
# mount /dev/device /path/to/cache -o user_xattrThe cache back end works by maintaining a certain amount of free space on the partition hosting the cache. It grows and shrinks the cache in response to other elements of the system using up free space, making it safe to use on the root file system (for example, on a laptop). FS-Cache sets defaults on this behavior, which can be configured via cache cull limits. For more information about configuring cache cull limits, refer to Section 10.4, “Setting Cache Cull Limits”.
Once the configuration file is in place, start up the cachefilesd service:
# systemctl start cachefilesdTo configure cachefilesd to start at boot time, execute the following command as root:
# systemctl enable cachefilesd10.3. Using the Cache with NFS
NFS will not use the cache unless explicitly instructed. To configure an NFS mount to use FS-Cache, include the -o fsc option to the mount command:
# mount nfs-share:/ /mount/point -o fscAll access to files under /mount/point will go through the cache, unless the file is opened for direct I/O or writing. For more information, see Section 10.3.2, “Cache Limitations with NFS”. NFS indexes cache contents using NFS file handle, not the file name, which means hard-linked files share the cache correctly.
Caching is supported in version 2, 3, and 4 of NFS. However, each version uses different branches for caching.
10.3.1. Cache Sharing
There are several potential issues to do with NFS cache sharing. Because the cache is persistent, blocks of data in the cache are indexed on a sequence of four keys:
Level 1: Server details
Level 2: Some mount options; security type; FSID; uniquifier
Level 3: File Handle
Level 4: Page number in file
To avoid coherency management problems between superblocks, all NFS superblocks that wish to cache data have unique Level 2 keys. Normally, two NFS mounts with same source volume and options share a superblock, and thus share the caching, even if they mount different directories within that volume.
Example 10.1. Cache Sharing
Take the following two mount commands:
mount home0:/disk0/fred /home/fred -o fsc
mount home0:/disk0/jim /home/jim -o fsc
Here, /home/fred and /home/jim likely share the superblock as they have the same options, especially if they come from the same volume/partition on the NFS server (home0). Now, consider the next two subsequent mount commands:
mount home0:/disk0/fred /home/fred -o fsc,rsize=230
mount home0:/disk0/jim /home/jim -o fsc,rsize=231
In this case, /home/fred and /home/jim will not share the superblock as they have different network access parameters, which are part of the Level 2 key. The same goes for the following mount sequence:
mount home0:/disk0/fred /home/fred1 -o fsc,rsize=230
mount home0:/disk0/fred /home/fred2 -o fsc,rsize=231
Here, the contents of the two subtrees (/home/fred1 and /home/fred2) will be cached twice.
Another way to avoid superblock sharing is to suppress it explicitly with the nosharecache parameter. Using the same example:
mount home0:/disk0/fred /home/fred -o nosharecache,fsc
mount home0:/disk0/jim /home/jim -o nosharecache,fsc
However, in this case only one of the superblocks is permitted to use cache since there is nothing to distinguish the Level 2 keys of home0:/disk0/fred and home0:/disk0/jim. To address this, add a unique identifier on at least one of the mounts, i.e. fsc=unique-identifier. For example:
mount home0:/disk0/fred /home/fred -o nosharecache,fsc
mount home0:/disk0/jim /home/jim -o nosharecache,fsc=jim
Here, the unique identifier jim is added to the Level 2 key used in the cache for /home/jim.
10.3.2. Cache Limitations with NFS
Opening a file from a shared file system for direct I/O automatically bypasses the cache. This is because this type of access must be direct to the server.
Opening a file from a shared file system for writing will not work on NFS version 2 and 3. The protocols of these versions do not provide sufficient coherency management information for the client to detect a concurrent write to the same file from another client.
Opening a file from a shared file system for either direct I/O or writing flushes the cached copy of the file. FS-Cache will not cache the file again until it is no longer opened for direct I/O or writing.
Furthermore, this release of FS-Cache only caches regular NFS files. FS-Cache will not cache directories, symlinks, device files, FIFOs and sockets.
10.4. Setting Cache Cull Limits
The cachefilesd daemon works by caching remote data from shared file systems to free space on the disk. This could potentially consume all available free space, which could be bad if the disk also housed the root partition. To control this, cachefilesd tries to maintain a certain amount of free space by discarding old objects (i.e. accessed less recently) from the cache. This behavior is known as cache culling.
Cache culling is done on the basis of the percentage of blocks and the percentage of files available in the underlying file system. There are six limits controlled by settings in /etc/cachefilesd.conf:
brun N% (percentage of blocks) , frun N% (percentage of files)
If the amount of free space and the number of available files in the cache rises above both these limits, then culling is turned off.
bcull N% (percentage of blocks), fcull N% (percentage of files)If the amount of available space or the number of files in the cache falls below either of these limits, then culling is started.
bstop N% (percentage of blocks), fstop N% (percentage of files)If the amount of available space or the number of available files in the cache falls below either of these limits, then no further allocation of disk space or files is permitted until culling has raised things above these limits again.
The default value of N for each setting is as follows:
brun/frun - 10%
bcull/fcull - 7%
bstop/fstop - 3%
When configuring these settings, the following must hold true:
0 ≤ bstop < bcull < brun < 100
0 ≤ fstop < fcull < frun < 100
These are the percentages of available space and available files and do not appear as 100 minus the percentage displayed by the df program.
Culling depends on both bxxx and fxxx pairs simultaneously; they can not be treated separately.
10.5. Statistical Information
FS-Cache also keeps track of general statistical information. To view this information, use:
# cat /proc/fs/fscache/statsFS-Cache statistics includes information on decision points and object counters. For more information, see the following kernel document:
/usr/share/doc/kernel-doc-version/Documentation/filesystems/caching/fscache.txt
10.6. FS-Cache References
For more information on cachefilesd and how to configure it, see man cachefilesd and man cachefilesd.conf. The following kernel documents also provide additional information:
/usr/share/doc/cachefilesd-version-number/README
/usr/share/man/man5/cachefilesd.conf.5.gz
/usr/share/man/man8/cachefilesd.8.gz
For general information about FS-Cache, including details on its design constraints, available statistics, and capabilities, see the following kernel document: /usr/share/doc/kernel-doc-version/Documentation/filesystems/caching/fscache.txt
Part II. Storage Administration
The Storage Administration section starts with storage considerations for Red Hat Enterprise Linux 7. Instructions regarding partitions, logical volume management, and swap partitions follow this. Disk Quotas, RAID systems are next, followed by the functions of mount command, volume_key, and acls. SSD tuning, write barriers, I/O limits and diskless systems follow this. The large chapter of Online Storage is next, and finally device mapper multipathing and virtual storage to finish.
Chapter 11. Storage Considerations During Installation
Many storage device and file system settings can only be configured at install time. Other settings, such as file system type, can only be modified up to a certain point without requiring a reformat. As such, it is prudent that you plan your storage configuration accordingly before installing Red Hat Enterprise Linux 7.
This chapter discusses several considerations when planning a storage configuration for your system. For installation instructions (including storage configuration during installation), see the Installation Guide provided by Red Hat.
11.1. Special Considerations
This section enumerates several issues and factors to consider for specific storage configurations.
Separate Partitions for /home, /opt, /usr/local
If it is likely that you will upgrade your system in the future, place /home, /opt, and /usr/local on a separate device. This allows you to reformat the devices or file systems containing the operating system while preserving your user and application data.
DASD and zFCP Devices on IBM System Z
On the IBM System Z platform, DASD and zFCP devices are configured via the Channel Command Word (CCW) mechanism. CCW paths must be explicitly added to the system and then brought online. For DASD devices, this means listing the device numbers (or device number ranges) as the DASD= parameter at the boot command line or in a CMS configuration file.
For zFCP devices, you must list the device number, logical unit number (LUN), and world wide port name (WWPN). Once the zFCP device is initialized, it is mapped to a CCW path. The FCP_x= lines on the boot command line (or in a CMS configuration file) allow you to specify this information for the installer.
Encrypting Block Devices Using LUKS
Formatting a block device for encryption using LUKS/dm-crypt destroys any existing formatting on that device. As such, you should decide which devices to encrypt (if any) before the new system's storage configuration is activated as part of the installation process.
Stale BIOS RAID Metadata
Moving a disk from a system configured for firmware RAID without removing the RAID metadata from the disk can prevent Anaconda from correctly detecting the disk.
Removing/deleting RAID metadata from disk could potentially destroy any stored data. Red Hat recommends that you back up your data before proceeding.
If you have created the RAID volume using dmraid, which is now deprecated, use the dmraid utility to delete it:
# dmraid -r -E /device/iSCSI Detection and Configuration
For plug and play detection of iSCSI drives, configure them in the firmware of an iBFT boot-capable network interface card (NIC). CHAP authentication of iSCSI targets is supported during installation. However, iSNS discovery is not supported during installation.
FCoE Detection and Configuration
For plug and play detection of Fibre Channel over Ethernet (FCoE) drives, configure them in the firmware of an EDD boot-capable NIC.
DASD
Direct-access storage devices (DASD) cannot be added or configured during installation. Such devices are specified in the CMS configuration file.
Block Devices with DIF/DIX Enabled
DIF/DIX is a hardware checksum feature provided by certain SCSI host bus adapters and block devices. When DIF/DIX is enabled, error occurs if the block device is used as a general-purpose block device. Buffered I/O or mmap(2)-based I/O will not work reliably, as there are no interlocks in the buffered write path to prevent buffered data from being overwritten after the DIF/DIX checksum has been calculated.
This causes the I/O to later fail with a checksum error. This problem is common to all block device (or file system-based) buffered I/O or mmap(2) I/O, so it is not possible to work around these errors caused by overwrites.
As such, block devices with DIF/DIX enabled should only be used with applications that use O_DIRECT. Such applications should use the raw block device. Alternatively, it is also safe to use the XFS file system on a DIF/DIX enabled block device, as long as only O_DIRECT I/O is issued through the file system. XFS is the only file system that does not fall back to buffered I/O when doing certain allocation operations.
The responsibility for ensuring that the I/O data does not change after the DIF/DIX checksum has been computed always lies with the application, so only applications designed for use with O_DIRECT I/O and DIF/DIX hardware should use DIF/DIX.
Chapter 12. File System Check
File systems may be checked for consistency, and optionally repaired, with file system-specific userspace tools. These tools are often referred to as fsck tools, where fsck is a shortened version of file system check.
These file system checkers only guarantee metadata consistency across the file system; they have no awareness of the actual data contained within the file system and are not data recovery tools.
File system inconsistencies can occur for various reasons, including but not limited to hardware errors, storage administration errors, and software bugs.
Before modern metadata-journaling file systems became common, a file system check was required any time a system crashed or lost power. This was because a file system update could have been interrupted, leading to an inconsistent state. As a result, a file system check is traditionally run on each file system listed in /etc/fstab at boot-time. For journaling file systems, this is usually a very short operation, because the file system's metadata journaling ensures consistency even after a crash.
However, there are times when a file system inconsistency or corruption may occur, even for journaling file systems. When this happens, the file system checker must be used to repair the file system. The following provides best practices and other useful information when performing this procedure.
Red Hat does not recommend this unless the machine does not boot, the file system is extremely large, or the file system is on remote storage. It is possible to disable file system check at boot by setting the sixth field in /etc/fstab to 0.
12.1. Best Practices for fsck
Generally, running the file system check and repair tool can be expected to automatically repair at least some of the inconsistencies it finds. In some cases, severely damaged inodes or directories may be discarded if they cannot be repaired. Significant changes to the file system may occur. To ensure that unexpected or undesirable changes are not permanently made, perform the following precautionary steps:
Dry run
Most file system checkers have a mode of operation which checks but does not repair the file system. In this mode, the checker prints any errors that it finds and actions that it would have taken, without actually modifying the file system.
Later phases of consistency checking may print extra errors as it discovers inconsistencies which would have been fixed in early phases if it were running in repair mode.
Operate first on a file system imageMost file systems support the creation of a metadata image, a sparse copy of the file system which contains only metadata. Because file system checkers operate only on metadata, such an image can be used to perform a dry run of an actual file system repair, to evaluate what changes would actually be made. If the changes are acceptable, the repair can then be performed on the file system itself.
Severely damaged file systems may cause problems with metadata image creation.
Save a file system image for support investigationsA pre-repair file system metadata image can often be useful for support investigations if there is a possibility that the corruption was due to a software bug. Patterns of corruption present in the pre-repair image may aid in root-cause analysis.
Operate only on unmounted file systemsA file system repair must be run only on unmounted file systems. The tool must have sole access to the file system or further damage may result. Most file system tools enforce this requirement in repair mode, although some only support check-only mode on a mounted file system. If check-only mode is run on a mounted file system, it may find spurious errors that would not be found when run on an unmounted file system.
Disk errorsFile system check tools cannot repair hardware problems. A file system must be fully readable and writable if repair is to operate successfully. If a file system was corrupted due to a hardware error, the file system must first be moved to a good disk, for example with the dd(8) utility.
12.2. File System-Specific Information for fsck
12.2.1. ext2, ext3, and ext4
All of these file sytems use the e2fsck binary to perform file system checks and repairs. The file names fsck.ext2, fsck.ext3, and fsck.ext4 are hardlinks to this same binary. These binaries are run automatically at boot time and their behavior differs based on the file system being checked and the state of the file system.
A full file system check and repair is invoked for ext2, which is not a metadata journaling file system, and for ext4 file systems without a journal.
For ext3 and ext4 file systems with metadata journaling, the journal is replayed in userspace and the binary exits. This is the default action as journal replay ensures a consistent file system after a crash.
If these file systems encounter metadata inconsistencies while mounted, they record this fact in the file system superblock. If e2fsck finds that a file system is marked with such an error, e2fsck performs a full check after replaying the journal (if present).
e2fsck may ask for user input during the run if the -p option is not specified. The -p option tells e2fsck to automatically do all repairs that may be done safely. If user intervention is required, e2fsck indicates the unfixed problem in its output and reflect this status in the exit code.
Commonly used e2fsck run-time options include:
-n
No-modify mode. Check-only operation.
-b superblockSpecify block number of an alternate suprerblock if the primary one is damaged.
-fForce full check even if the superblock has no recorded errors.
-j journal-devSpecify the external journal device, if any.
-pAutomatically repair or "preen" the file system with no user input.
-yAssume an answer of "yes" to all questions.
All options for e2fsck are specified in the e2fsck(8) manual page.
The following five basic phases are performed by e2fsck while running:
Inode, block, and size checks.
Directory structure checks.
Directory connectivity checks.
Reference count checks.
Group summary info checks.
The e2image(8) utility can be used to create a metadata image prior to repair for diagnostic or testing purposes. The -r option should be used for testing purposes in order to create a sparse file of the same size as the file system itself. e2fsck can then operate directly on the resulting file. The -Q option should be specified if the image is to be archived or provided for diagnostic. This creates a more compact file format suitable for transfer.
12.2.2. XFS
No repair is performed automatically at boot time. To initiate a file system check or repair, use the xfs_repair tool.
Although an fsck.xfs binary is present in the xfsprogs package, this is present only to satisfy initscripts that look for an fsck.file system binary at boot time. fsck.xfs immediately exits with an exit code of 0.
Older xfsprogs packages contain an xfs_check tool. This tool is very slow and does not scale well for large file systems. As such, it has been deprecated in favor of xfs_repair -n.
A clean log on a file system is required for xfs_repair to operate. If the file system was not cleanly unmounted, it should be mounted and unmounted prior to using xfs_repair. If the log is corrupt and cannot be replayed, the -L option may be used to zero the log.
The -L option must only be used if the log cannot be replayed. The option discards all metadata updates in the log and results in further inconsistencies.
It is possible to run xfs_repair in a dry run, check-only mode by using the -n option. No changes will be made to the file system when this option is specified.
xfs_repair takes very few options. Commonly used options include:
-n
No modify mode. Check-only operation.
-LZero metadata log. Use only if log cannot be replayed with mount.
-m maxmemLimit memory used during run to maxmem MB. 0 can be specified to obtain a rough estimate of the minimum memory required.
-l logdevSpecify the external log device, if present.
All options for xfs_repair are specified in the xfs_repair(8) manual page.
The following eight basic phases are performed by xfs_repair while running:
Inode and inode blockmap (addressing) checks.
Inode allocation map checks.
Inode size checks.
Directory checks.
Pathname checks.
Link count checks.
Freemap checks.
Super block checks.
For more information, see the xfs_repair(8) manual page.
xfs_repair is not interactive. All operations are performed automatically with no input from the user.
If it is desired to create a metadata image prior to repair for diagnostic or testing purposes, the xfs_metadump(8) and xfs_mdrestore(8) utilities may be used.
12.2.3. Btrfs
Btrfs is available as a Technology Preview feature in Red Hat Enterprise Linux 7 but has been deprecated since the Red Hat Enterprise Linux 7.4 release. It will be removed in a future major release of Red Hat Enterprise Linux.
For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.4 Release Notes.
The btrfsck tool is used to check and repair btrfs file systems. This tool is still in early development and may not detect or repair all types of file system corruption.
By default, btrfsck does not make changes to the file system; that is, it runs check-only mode by default. If repairs are desired the --repair option must be specified.
The following three basic phases are performed by btrfsck while running:
Extent checks.
File system root checks.
Root reference count checks.
The btrfs-image(8) utility can be used to create a metadata image prior to repair for diagnostic or testing purposes.
Chapter 13. Partitions
With the parted utility, you can:
View the existing partition table.
Change the size of existing partitions.
Add partitions from free space or additional hard drives.
The parted package is installed by default on Red Hat Enterprise Linux 7. To start parted, log in as root and enter the following command:
# parted /dev/sdaReplace /dev/sda with the device name for the drive to configure.
Manipulating Partitions on Devices in Use
For a device to not be in use, none of the partitions on the device can be mounted, and no swap space on the device can be enabled.
If you want to remove or resize a partition, the device on which that partition resides must not be in use.
It is possible to create a new partition on a device that is in use, but this is not recommended.
Modifying the Partition Table
Modifying the partition table while another partition on the same disk is in use is generally not recommended because the kernel is not able to reread the partition table. As a consequence, changes are not applied to a running system. In the described situation, reboot the system, or use the following command to make the system register new or modified partitions:
# partx --update --nr partition-number diskThe easiest way to modify disks that are currently in use is:
Boot the system in rescue mode if the partitions on the disk are impossible to unmount, for example in the case of a system disk.
When prompted to mount the file system, select Skip.
If the drive does not contain any partitions in use, that is there are no system processes that use or lock the file system from being unmounted, you can unmount the partitions with the umount command and turn off all the swap space on the hard drive with the swapoff command.
Do not use the parted utility to create file systems. Use the mkfs tool instead.
Table 13.1. parted Commands
help | Display list of available commands |
mklabel label | Create a disk label for the partition table |
mkpart part-type [fs-type] start-mb end-mb | Make a partition without creating a new file system |
name minor-num name | Name the partition for Mac and PC98 disklabels only |
Display the partition table | |
quit | Quit parted |
rescue start-mb end-mb | Rescue a lost partition from start-mb to end-mb |
rm minor-num | Remove the partition |
select device | Select a different device to configure |
set minor-num flag state | Set the flag on a partition; state is either on or off |
toggle [NUMBER [FLAG] | Toggle the state of FLAG on partition NUMBER |
unit UNIT | Set the default unit to UNIT |
13.1. Viewing the Partition Table
To view the partition table:
Start parted.
Use the following command to view the partition table:
(parted) print
A table similar to the following one appears:
Example 13.1. Partition Table
Model: ATA ST3160812AS (scsi) Disk /dev/sda: 160GB Sector size (logical/physical): 512B/512B Partition Table: msdos Number Start End Size Type File system Flags 1 32.3kB 107MB 107MB primary ext3 boot 2 107MB 105GB 105GB primary ext3 3 105GB 107GB 2147MB primary linux-swap 4 107GB 160GB 52.9GB extended root 5 107GB 133GB 26.2GB logical ext3 6 133GB 133GB 107MB logical ext3 7 133GB 160GB 26.6GB logical lvm
Following is the description of the partition table:
Model: ATA ST3160812AS (scsi): explains the disk type, manufacturer, model number, and interface.
Disk /dev/sda: 160GB: displays the file path to the block device and the storage capacity.
Partition Table: msdos: displays the disk label type.
In the partition table, Number is the partition number. For example, the partition with minor number 1 corresponds to /dev/sda1. The Start and End values are in megabytes. Valid Types are metadata, free, primary, extended, or logical. The File system is the file system type. The Flags column lists the flags set for the partition. Available flags are boot, root, swap, hidden, raid, lvm, or lba.
The File system in the partition table can be any of the following:
ext2
ext3
fat16
fat32
hfs
jfs
linux-swap
ntfs
reiserfs
hp-ufs
sun-ufs
xfs
If a File system of a device shows no value, this means that its file system type is unknown.
To select a different device without having to restart parted, use the following command and replace /dev/sda with the device you want to select:
(parted) select /dev/sdaIt allows you to view or configure the partition table of a device.
13.2. Creating a Partition
Do not attempt to create a partition on a device that is in use.
Procedure 13.1. Creating a Partition
Before creating a partition, boot into rescue mode, or unmount any partitions on the device and turn off any swap space on the device.
Start parted:
# parted /dev/sdaReplace /dev/sda with the device name on which you want to create the partition.
View the current partition table to determine if there is enough free space:
(parted) printFrom the partition table, determine the start and end points of the new partition and what partition type it should be. You can only have four primary partitions, with no extended partition, on a device. If you need more than four partitions, you can have three primary partitions, one extended partition, and multiple logical partitions within the extended. For an overview of disk partitions, see the appendix An Introduction to Disk Partitions in the Red Hat Enterprise Linux 7 Installation Guide.
To create partition:
(parted) mkpart part-type name fs-type start endReplace part-type with with primary, logical, or extended as per your requirement.
Replace name with partition-name; name is required for GPT partition tables.
Replace fs-type with any one of btrfs, ext2, ext3, ext4, fat16, fat32, hfs, hfs+, linux-swap, ntfs, reiserfs, or xfs; fs-type is optional.
Replace start end with the size in megabytes as per your requirement.
For example, to create a primary partition with an ext3 file system from 1024 megabytes until 2048 megabytes on a hard drive, type the following command:
(parted) mkpart primary 1024 2048If you use the mkpartfs command instead, the file system is created after the partition is created. However, parted does not support creating an ext3 file system. Thus, if you wish to create an ext3 file system, use mkpart and create the file system with the mkfs command as described later.
The changes start taking place as soon as you press Enter, so review the command before executing to it.
View the partition table to confirm that the created partition is in the partition table with the correct partition type, file system type, and size using the following command:
(parted) printAlso remember the minor number of the new partition so that you can label any file systems on it.
Exit the parted shell:
(parted) quitUse the following command after parted is closed to make sure the kernel recognizes the new partition:
# cat /proc/partitions
The maximum number of partitions parted can create is 128. While the GUID Partition Table (GPT) specification allows for more partitions by growing the area reserved for the partition table, common practice used by parted is to limit it to enough area for 128 partitions.
13.2.1. Formatting and Labeling the Partition
To format and label the partition use the following procedure:
Procedure 13.2. Format and Label the Partition
The partition does not have a file system. To create the ext4 file system, use:
# mkfs.ext4 /dev/sda6Formatting the partition permanently destroys any data that currently exists on the partition.
Label the file system on the partition. For example, if the file system on the new partition is /dev/sda6 and you want to label it Work, use:
# e2label /dev/sda6 "Work"By default, the installation program uses the mount point of the partition as the label to make sure the label is unique. You can use any label you want.
Create a mount point (e.g. /work) as root.
13.2.2. Add the Partition to /etc/fstab
As root, edit the /etc/fstab file to include the new partition using the partition's UUID.
Use the command blkid -o list for a complete list of the partition's UUID, or blkid device for individual device details.
In /etc/fstab:
The first column should contain UUID= followed by the file system's UUID.
The second column should contain the mount point for the new partition.
The third column should be the file system type: for example, ext4 or swap.
The fourth column lists mount options for the file system. The word defaults here means that the partition is mounted at boot time with default options.
The fifth and sixth field specify backup and check options. Example values for a non-root partition are 0 2.
Regenerate mount units so that your system registers the new configuration:
# systemctl daemon-reloadTry mounting the file system to verify that the configuration works:
# mount /work
Additional Information
If you need more information about the format of /etc/fstab, see the fstab(5) man page.
13.3. Removing a Partition
Do not attempt to remove a partition on a device that is in use.
Procedure 13.3. Remove a Partition
Before removing a partition, do one of the following:
Boot into rescue mode, or
Unmount any partitions on the device and turn off any swap space on the device.
Start the parted utility:
# parted deviceReplace device with the device on which to remove the partition: for example, /dev/sda.
View the current partition table to determine the minor number of the partition to remove:
(parted) printRemove the partition with the command rm. For example, to remove the partition with minor number 3:
(parted) rm 3The changes start taking place as soon as you press Enter, so review the command before committing to it.
After removing the partition, use the print command to confirm that it is removed from the partition table:
(parted) printExit from the parted shell:
(parted) quitExamine the content of the /proc/partitions file to make sure the kernel knows the partition is removed:
# cat /proc/partitionsRemove the partition from the /etc/fstab file. Find the line that declares the removed partition, and remove it from the file.
Regenerate mount units so that your system registers the new /etc/fstab configuration:
# systemctl daemon-reload
13.4. Setting a Partition Type
The partition type, not to be confused with the file system type, is used by a running system only rarely. However, the partition type matters to on-the-fly generators, such as systemd-gpt-auto-generator, which use the partition type to, for example, automatically identify and mount devices.
You can start the fdisk utility and use the t command to set the partition type. The following example shows how to change the partition type of the first partition to 0x83, default on Linux:
# fdisk /dev/sdc Command (m for help): t Selected partition 1 Partition type (type L to list all types): 83 Changed type of partition 'Linux LVM' to 'Linux'.The parted utility provides some control of partition types by trying to map the partition type to 'flags', which is not convenient for end users. The parted utility can handle only certain partition types, for example LVM or RAID. To remove, for example, the lvm flag from the first partition with parted, use:
# parted /dev/sdc 'set 1 lvm off'For a list of commonly used partition types and hexadecimal numbers used to represent them, see the Partition Types table in the Partitions: Turning One Drive Into Many appendix of the Red Hat Enterprise Linux 7 Installation Guide.
13.5. Resizing a Partition with fdisk
The fdisk utility allows you to create and manipulate GPT, MBR, Sun, SGI, and BSD partition tables. On disks with a GUID Partition Table (GPT), using the parted utility is recommended, as fdisk GPT support is in an experimental phase.
Before resizing a partition, back up the data stored on the file system and test the procedure, as the only way to change a partition size using fdisk is by deleting and recreating the partition.
The partition you are resizing must be the last partition on a particular disk.
Red Hat only supports extending and resizing LVM partitions.
Procedure 13.4. Resizing a Partition
The following procedure is provided only for reference. To resize a partition using fdisk:
Unmount the device:
# umount /dev/vdaRun fdisk disk_name. For example:
# fdisk /dev/vda Welcome to fdisk (util-linux 2.23.2). Changes will remain in memory only, until you decide to write them. Be careful before using the write command. Command (m for help):Use the p option to determine the line number of the partition to be deleted.
Command (m for help): p Disk /dev/vda: 16.1 GB, 16106127360 bytes, 31457280 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk label type: dos Disk identifier: 0x0006d09a Device Boot Start End Blocks Id System /dev/vda1 * 2048 1026047 512000 83 Linux /dev/vda2 1026048 31457279 15215616 8e Linux LVMUse the d option to delete a partition. If there is more than one partition available, fdisk prompts you to provide a number of the partition to delete:
Command (m for help): d Partition number (1,2, default 2): 2 Partition 2 is deletedUse the n option to create a partition and follow the prompts. Allow enough space for any future resizing. The fdisk default behavior (press Enter) is to use all space on the device. You can specify the end of the partition by sectors, or specify a human-readable size by using +<size><suffix>, for example +500M, or +10G.
Red Hat recommends using the human-readable size specification if you do not want to use all free space, as fdisk aligns the end of the partition with the physical sectors. If you specify the size by providing an exact number (in sectors), fdisk does not align the end of the partition.
Command (m for help): n Partition type: p primary (1 primary, 0 extended, 3 free) e extended Select (default p): *Enter* Using default response p Partition number (2-4, default 2): *Enter* First sector (1026048-31457279, default 1026048): *Enter* Using default value 1026048 Last sector, +sectors or +size{K,M,G} (1026048-31457279, default 31457279): +500M Partition 2 of type Linux and of size 500 MiB is setSet the partition type to LVM:
Command (m for help): t Partition number (1,2, default 2): *Enter* Hex code (type L to list all codes): 8e Changed type of partition 'Linux' to 'Linux LVM'Write the changes with the w option when you are sure the changes are correct, as errors can cause instability with the selected partition.
Run e2fsck on the device to check for consistency:
# e2fsck /dev/vda e2fsck 1.41.12 (17-May-2010) Pass 1:Checking inodes, blocks, and sizes Pass 2:Checking directory structure Pass 3:Checking directory connectivity Pass 4:Checking reference counts Pass 5:Checking group summary information ext4-1:11/131072 files (0.0% non-contiguous),27050/524128 blocksMount the device:
# mount /dev/vda
For more information, see the fdisk(8) manual page.
Chapter 14. Creating and Maintaining Snapshots with Snapper
A snapshot volume is a point in time copy of a target volume that provides a way to revert a file system back to an earlier state. Snapper is a command-line tool to create and maintain snapshots for Btrfs and thinly-provisioned LVM file systems.
14.1. Creating Initial Snapper Configuration
Snapper requires discrete configuration files for each volume it operates on. You must set up the configuration files manually. By default, only the root user is allowed to perform snapper commands.
If you are using thin provisioning, monitor free space in the thin-pool. A fully consumed thin pool can result in data loss. For more information, see Thinly-Provisioned Logical Volumes (Thin Volumes) in the Red Hat Enterprise Linux 7 Logical Volume Manager Administration Guide.
Note that the Btrfs tools and file system are provided as a Technology Preview, which make them unsuitable for production systems.
Although it is possible to allow a user or group other than root to use certain Snapper commands, Red Hat recommends that you do not add elevated permissions to otherwise unprivileged users or groups. Such a configuration bypasses SELinux and could pose a security risk. Red Hat recommends that you review these capabilities with your Security Team and consider using the sudo infrastructure instead.
Btrfs is available as a Technology Preview feature in Red Hat Enterprise Linux 7 but has been deprecated since the Red Hat Enterprise Linux 7.4 release. It will be removed in a future major release of Red Hat Enterprise Linux.
For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.4 Release Notes.
Procedure 14.1. Creating a Snapper Configuration File
Create or choose either:
A thinly-provisioned logical volume with a Red Hat supported file system on top of it, or
A Btrfs subvolume.
Mount the file system.
Create the configuration file that defines this volume.
For LVM2:
# snapper -c config_name create-config -f "lvm(fs_type)" /mount-pointFor example, to create a configuration file called lvm_config on an LVM2 subvolume with an ext4 file system, mounted at /lvm_mount, use:
# snapper -c lvm_config create-config -f "lvm(ext4)" /lvm_mountFor Btrfs:
# snapper -c config_name create-config -f btrfs /mount-pointThe -c config_name option specifies the name of the configuration file.
The create-config tells snapper to create a configuration file.
The -f file_system tells snapper what file system to use; if this is omitted snapper will attempt to detect the file system.
The /mount-point is where the subvolume or thinly-provisioned LVM2 file system is mounted.
Alternatively, to create a configuration file called btrfs_config, on a Btrfs subvolume that is mounted at /btrfs_mount, use:
# snapper -c btrfs_config create-config -f btrfs /btrfs_mount
The configuration files are stored in the /etc/snapper/configs/ directory.
14.2. Creating a Snapper Snapshot
Snapper can create the following kinds of snapshots:
Pre Snapshot
A pre snapshot serves as a point of origin for a post snapshot. The two are closely tied and designed to track file system modification between the two points. The pre snapshot must be created before the post snapshot.
Post SnapshotA post snapshot serves as the end point to the pre snapshot. The coupled pre and post snapshots define a range for comparison. By default, every new snapper volume is configured to create a background comparison after a related post snapshot is created successfully.
Single SnapshotA single snapshot is a standalone snapshot created at a specific moment. These can be used to track a timeline of modifications and have a general point to return to later.
14.2.1. Creating a Pre and Post Snapshot Pair
14.2.1.1. Creating a Pre Snapshot with Snapper
To create a pre snapshot, use:
# snapper -c config_name create -t preThe create -t option specifies what type of snapshot to create. Accepted entries are pre, post, or single.
# snapper -c SnapperExample create -t pre -p 1The -p option prints the number of the created snapshot and is optional.
14.2.1.2. Creating a Post Snapshot with Snapper
Procedure 14.2. Creating a Post Snapshot
Determine the number of the pre snapshot:
# snapper -c config_name listFor example, to display the list of snapshots created using the configuration file lvm_config, use the following:
# snapper -c lvm_config list Type | # | Pre # | Date | User | Cleanup | Description | Userdata -------+---+-------+-------------------+------+----------+-------------+--------- single | 0 | | | root | | current | pre | 1 | | Mon 06<...> | root | | |This output shows that the pre snapshot is number 1.
Create a post snapshot that is linked to a previously created pre snapshot:
# snapper -c config_file create -t post --pre-num pre_snapshot_numberThe -t post option specifies the creation of the post snapshot type.
The --pre-num option specifies the corresponding pre snapshot.
For example, to create a post snapshot using the lvm_config configuration file and is linked to pre snapshot number 1, use:
# snapper -c lvm_config create -t post --pre-num 1 -p 2The -p option prints the number of the created snapshot and is optional.
The pre and post snapshots 1 and 2 are now created and paired. Verify this with the list command:
# snapper -c lvm_config list Type | # | Pre # | Date | User | Cleanup | Description | Userdata -------+---+-------+-------------------+------+----------+-------------+--------- single | 0 | | | root | | current | pre | 1 | | Mon 06<...> | root | | | post | 2 | 1 | Mon 06<...> | root | | |
14.2.1.3. Wrapping a Command in Pre and Post Snapshots
You can also wrap a command within a pre and post snapshot, which can be useful when testing. See Procedure 14.3, “Wrapping a Command in Pre and Post Snapshots”, which is a shortcut for the following steps:
Running the snapper create pre snapshot command.
Running a command or a list of commands to perform actions with a possible impact on the file system content.
Running the snapper create post snapshot command.
Procedure 14.3. Wrapping a Command in Pre and Post Snapshots
To wrap a command in pre and post snapshots:
# snapper -c lvm_config create --command "command_to_be_tracked"For example, to track the creation of the /lvm_mount/hello_file file:
# snapper -c lvm_config create --command "echo Hello > /lvm_mount/hello_file"To verify this, use the status command:
# snapper -c config_file status first_snapshot_number..second_snapshot_numberFor example, to track the changes made in the first step:
# snapper -c lvm_config status 3..4 +..... /lvm_mount/hello_fileUse the list command to verify the number of the snapshot if needed.
Note that there is no guarantee that the command in the given example is the only thing the snapshots capture. Snapper also records anything that is modified by the system, not just what a user modifies.
14.2.2. Creating a Single Snapper Snapshot
Creating a single snapper snapshot is similar to creating a pre or post snapshot, only the create -t option specifies single. The single snapshot is used to create a single snapshot in time without having it relate to any others. However, if you are interested in a straightforward way to create snapshots of LVM2 thin volumes without the need to automatically generate comparisons or list additional information, Red Hat recommends using the System Storage Manager instead of Snapper for this purpose, as described in Section 16.2.6, “Snapshot”.
To create a single snapshot, use:
# snapper -c config_name create -t singleFor example, the following command creates a single snapshot using the lvm_config configuration file.
# snapper -c lvm_config create -t singleAlthough single snapshots are not specifically designed to track changes, you can use the snapper diff, xadiff, and status commands to compare any two snapshots. For more information on these commands, see Section 14.3, “Tracking Changes Between Snapper Snapshots”.
14.2.3. Configuring Snapper to Take Automated Snapshots
Taking automated snapshots is one of key features of Snapper. By default, when you configure Snapper for a volume, Snapper starts taking a snapshot of the volume every hour.
Under the default configuration, Snapper keeps:
10 hourly snapshots, and the final hourly snapshot is saved as a “daily” snapshot.
10 daily snapshots, and the final daily snapshot for a month is saved as a “monthly” snapshot.
10 monthly snapshots, and the final monthly snapshot is saved as a “yearly” snapshot.
10 yearly snapshots.
Note that Snapper keeps by default no more that 50 snapshots in total. However, Snapper keeps by default all snapshots created less than 1,800 seconds ago.
The default configuration is specified in the /etc/snapper/config-templates/default file. When you use the snapper create-config command to create a configuration, any unspecified values are set based on the default configuration. You can edit the configuration for any defined volume in the /etc/snapper/configs/config_name file.
14.3. Tracking Changes Between Snapper Snapshots
Use the status, diff, and xadiff commands to track the changes made to a subvolume between snapshots:
status
The status command shows a list of files and directories that have been created, modified, or deleted between two snapshots, that is a comprehensive list of changes between two snapshots. You can use this command to get an overview of the changes without excessive details.
diffThe diff command shows a diff of modified files and directories between two snapshots as received from the status command if there is at least one modification detected.
xadiffThe xadiff command compares how the extended attributes of a file or directory have changed between two snapshots.
14.3.1. Comparing Changes with the status Command
The status command shows a list of files and directories that have been created, modified, or deleted between two snapshots.
To display the status of files between two snapshots, use:
# snapper -c config_file status first_snapshot_number..second_snapshot_numberUse the list command to determine snapshot numbers if needed.
For example, the following command displays the changes made between snapshot 1 and 2, using the configuration file lvm_config.
#snapper -c lvm_config status 1..2 tp.... /lvm_mount/dir1 -..... /lvm_mount/dir1/file_a c.ug.. /lvm_mount/file2 +..... /lvm_mount/file3 ....x. /lvm_mount/file4 cp..xa /lvm_mount/file5Read letters and dots in the first part of the output as columns:
+..... /lvm_mount/file3 |||||| 123456Column 1 indicates any modification of the file (directory entry) type. Possible values are:
Column 1
. | Nothing has changed. |
+ | File created. |
- | File deleted. |
c | Content changed. |
t | The type of directory entry has changed. For example, a former symbolic link has changed to a regular file with the same file name. |
Column 2 indicates any changes in the file permissions. Possible values are:
Column 2
. | No permissions changed. |
p | Permissions changed. |
Column 3 indicates any changes in the user ownership. Possible values are:
Column 3
. | No user ownership changed. |
u | User ownership has changed. |
Column 4 indicates any changes in the group ownership. Possible values are:
Column 4
. | No group ownership changed. |
g | Group ownership has changed. |
Column 5 indicates any changes in the extended attributes. Possible values are:
Column 5
. | No extended attributes changed. |
x | Extended attributes changed. |
Column 6 indicates any changes in the access control lists (ACLs). Possible values are:
Column 6
. | No ACLs changed. |
a | ACLs modified. |
14.3.2. Comparing Changes with the diff Command
The diff command shows the changes of modified files and directories between two snapshots.
# snapper -c config_name diff first_snapshot_number..second_snapshot_numberUse the list command to determine the number of the snapshot if needed.
For example, to compare the changes made in files between snapshot 1 and snapshot 2 that were made using the lvm_config configuration file, use:
# snapper -c lvm_config diff 1..2 --- /lvm_mount/.snapshots/13/snapshot/file4 19<...> +++ /lvm_mount/.snapshots/14/snapshot/file4 20<...> @@ -0,0 +1 @@ +wordsThis output shows that file4 had been modified to add "words" into the file.
14.3.3. Comparing Changes with the xadiff Command
The xadiff command compares how the extended attributes of a file or directory have changed between two snapshots:
# snapper -c config_name xadiff first_snapshot_number..second_snapshot_numberUse the list command to determine the number of the snapshot if needed.
For example, to show the xadiff output between snapshot number 1 and snapshot number 2 that were made using the lvm_config configuration file, use:
# snapper -c lvm_config xadiff 1..214.4. Reversing Changes in Between Snapshots
To reverse changes made between two existing Snapper snapshots, use the undochange command in the following format, where 1 is the first snapshot and 2 is the second snapshot:
snapper -c config_name undochange 1..2Using the undochange command does not revert the Snapper volume back to its original state and does not provide data consistency. Any file modification that occurs outside of the specified range, for example after snapshot 2, will remain unchanged after reverting back, for example to the state of snapshot 1. For example, if undochange is run to undo the creation of a user, any files owned by that user can still remain.
There is also no mechanism to ensure file consistency as a snapshot is made, so any inconsistencies that already exist can be transferred back to the snapshot when the undochange command is used.
Do not use the Snapper undochange command with the root file system, as doing so is likely to lead to a failure.
The following diagram demonstrates how the undochange command works:
Figure 14.1. Snapper Status over Time
The diagram shows the point in time in which snapshot_1 is created, file_a is created, then file_b deleted. Snapshot_2 is then created, after which file_a is edited and file_c is created. This is now the current state of the system. The current system has an edited version of file_a, no file_b, and a newly created file_c.
When the undochange command is called, Snapper generates a list of modified files between the first listed snapshot and the second. In the diagram, if you use the snapper -c SnapperExample undochange 1..2 command, Snapper creates a list of modified files (that is, file_a is created; file_b is deleted) and applies them to the current system. Therefore:
the current system will not have file_a, as it has yet to be created when snapshot_1 was created.
file_b will exist, copied from snapshot_1 into the current system.
file_c will exist, as its creation was outside the specified time.
Be aware that if file_b and file_c conflict, the system can become corrupted.
You can also use the snapper -c SnapperExample undochange 2..1 command. In this case, the current system replaces the edited version of file_a with one copied from snapshot_1, which undoes edits of that file made after snapshot_2 was created.
Using the mount and unmount Commands to Reverse Changes
The undochange command is not always the best way to revert modifications. With the status and diff command, you can make a qualified decision, and use the mount and unmount commands instead of Snapper. The mount and unmount commands are only useful if you want to mount snapshots and browse their content independently of Snapper workflow.
If needed, the mount command activates respective LVM Snapper snapshot before mounting. Use the mount and unmount commands if you are, for example, interested in mounting snapshots and extracting older version of several files manually. To revert files manually, copy them from a mounted snapshot to the current file system. The current file system, snapshot 0, is the live file system created in Procedure 14.1, “Creating a Snapper Configuration File”. Copy the files to the subtree of the original /mount-point.
Use the mount and unmount commands for explicit client-side requests. The /etc/snapper/configs/config_name file contains the ALLOW_USERS= and ALLOW_GROUPS= variables where you can add users and groups. Then, snapperd allows you to perform mount operations for the added users and groups.
14.5. Deleting a Snapper Snapshot
To delete a snapshot:
# snapper -c config_name delete snapshot_numberYou can use the list command to verify that the snapshot was successfully deleted.
Chapter 15. Swap Space
Swap space in Linux is used when the amount of physical memory (RAM) is full. If the system needs more memory resources and the RAM is full, inactive pages in memory are moved to the swap space. While swap space can help machines with a small amount of RAM, it should not be considered a replacement for more RAM. Swap space is located on hard drives, which have a slower access time than physical memory. Swap space can be a dedicated swap partition (recommended), a swap file, or a combination of swap partitions and swap files. Note that Btrfs does not support swap space.
In years past, the recommended amount of swap space increased linearly with the amount of RAM in the system. However, modern systems often include hundreds of gigabytes of RAM. As a consequence, recommended swap space is considered a function of system memory workload, not system memory.
Table 15.1, “Recommended System Swap Space” illustrates the recommended size of a swap partition depending on the amount of RAM in your system and whether you want sufficient memory for your system to hibernate. The recommended swap partition size is established automatically during installation. To allow for hibernation, however, you need to edit the swap space in the custom partitioning stage.
Recommendations in Table 15.1, “Recommended System Swap Space” are especially important on systems with low memory (1 GB and less). Failure to allocate sufficient swap space on these systems can cause issues such as instability or even render the installed system unbootable.
Table 15.1. Recommended System Swap Space
⩽ 2 GB | 2 times the amount of RAM | 3 times the amount of RAM |
> 2 GB – 8 GB | Equal to the amount of RAM | 2 times the amount of RAM |
> 8 GB – 64 GB | At least 4 GB | 1.5 times the amount of RAM |
> 64 GB | At least 4 GB | Hibernation not recommended |
There are two reasons why hibernation is not recommended with systems with more than 64 GB of RAM. Firstly, hibernation requires extra space for an inflated (and perhaps infrequently utilized) swap area. Secondly, the process of moving resident data from RAM to disk and back on can take a lot of time to complete.
At the border between each range listed in Table 15.1, “Recommended System Swap Space”, for example a system with 2 GB, 8 GB, or 64 GB of system RAM, discretion can be exercised with regard to chosen swap space and hibernation support. If your system resources allow for it, increasing the swap space may lead to better performance.
Note that distributing swap space over multiple storage devices also improves swap space performance, particularly on systems with fast drives, controllers, and interfaces.
File systems and LVM2 volumes assigned as swap space should not be in use when being modified. Any attempts to modify swap fail if a system process or the kernel is using swap space. Use the free and cat /proc/swaps commands to verify how much and where swap is in use.
You should modify swap space while the system is booted in rescue mode, see Booting Your Computer in Rescue Mode in the Red Hat Enterprise Linux 7 Installation Guide. When prompted to mount the file system, select Skip.
15.1. Adding Swap Space
Sometimes it is necessary to add more swap space after installation. For example, you may upgrade the amount of RAM in your system from 1 GB to 2 GB, but there is only 2 GB of swap space. It might be advantageous to increase the amount of swap space to 4 GB if you perform memory-intense operations or run applications that require a large amount of memory.
You have three options: create a new swap partition, create a new swap file, or extend swap on an existing LVM2 logical volume. It is recommended that you extend an existing logical volume.
15.1.1. Extending Swap on an LVM2 Logical Volume
By default, Red Hat Enterprise Linux 7 uses all available space during installation. If this is the case with your system, then you must first add a new physical volume to the volume group used by the swap space.
After adding additional storage to the swap space's volume group, it is now possible to extend it. To do so, perform the following procedure (assuming /dev/VolGroup00/LogVol01 is the volume you want to extend by 2 GB):
Procedure 15.1. Extending Swap on an LVM2 Logical Volume
Disable swapping for the associated logical volume:
# swapoff -v /dev/VolGroup00/LogVol01
Resize the LVM2 logical volume by 2 GB:
# lvresize /dev/VolGroup00/LogVol01 -L +2G
Format the new swap space:
# mkswap /dev/VolGroup00/LogVol01
Enable the extended logical volume:
# swapon -v /dev/VolGroup00/LogVol01
To test if the swap logical volume was successfully extended and activated, inspect active swap space:
$ cat /proc/swaps $ free -h
15.1.2. Creating an LVM2 Logical Volume for Swap
To add a swap volume group 2 GB in size, assuming /dev/VolGroup00/LogVol02 is the swap volume you want to add:
Create the LVM2 logical volume of size 2 GB:
# lvcreate VolGroup00 -n LogVol02 -L 2GFormat the new swap space:
# mkswap /dev/VolGroup00/LogVol02Add the following entry to the /etc/fstab file:
/dev/VolGroup00/LogVol02 swap swap defaults 0 0Regenerate mount units so that your system registers the new configuration:
# systemctl daemon-reloadActivate swap on the logical volume:
# swapon -v /dev/VolGroup00/LogVol02To test if the swap logical volume was successfully created and activated, inspect active swap space:
$ cat /proc/swaps $ free -h
15.1.3. Creating a Swap File
To add a swap file:
Procedure 15.2. Add a Swap File
Determine the size of the new swap file in megabytes and multiply by 1024 to determine the number of blocks. For example, the block size of a 64 MB swap file is 65536.
Create an empty file:
# dd if=/dev/zero of=/swapfile bs=1024 count=65536Replace count with the value equal to the desired block size.
Set up the swap file with the command:
# mkswap /swapfileChange the security of the swap file so it is not world readable.
# chmod 0600 /swapfileTo enable the swap file at boot time, edit /etc/fstab as root to include the following entry:
/swapfile swap swap defaults 0 0
The next time the system boots, it activates the new swap file.
Regenerate mount units so that your system registers the new /etc/fstab configuration:
# systemctl daemon-reloadTo activate the swap file immediately:
# swapon /swapfileTo test if the new swap file was successfully created and activated, inspect active swap space:
$ cat /proc/swaps $ free -h
15.2. Removing Swap Space
Sometimes it can be prudent to reduce swap space after installation. For example, you have downgraded the amount of RAM in your system from 1 GB to 512 MB, but there is 2 GB of swap space still assigned. It might be advantageous to reduce the amount of swap space to 1 GB, since the larger 2 GB could be wasting disk space.
You have three options: remove an entire LVM2 logical volume used for swap, remove a swap file, or reduce swap space on an existing LVM2 logical volume.
15.2.1. Reducing Swap on an LVM2 Logical Volume
To reduce an LVM2 swap logical volume (assuming /dev/VolGroup00/LogVol01 is the volume you want to reduce):
Procedure 15.3. Reducing an LVM2 Swap Logical Volume
Disable swapping for the associated logical volume:
# swapoff -v /dev/VolGroup00/LogVol01Reduce the LVM2 logical volume by 512 MB:
# lvreduce /dev/VolGroup00/LogVol01 -L -512MFormat the new swap space:
# mkswap /dev/VolGroup00/LogVol01Activate swap on the logical volume:
# swapon -v /dev/VolGroup00/LogVol01To test if the swap logical volume was successfully reduced, inspect active swap space:
$ cat /proc/swaps $ free -h
15.2.2. Removing an LVM2 Logical Volume for Swap
To remove a swap volume group (assuming /dev/VolGroup00/LogVol02 is the swap volume you want to remove):
Procedure 15.4. Remove a Swap Volume Group
Disable swapping for the associated logical volume:
# swapoff -v /dev/VolGroup00/LogVol02Remove the LVM2 logical volume:
# lvremove /dev/VolGroup00/LogVol02Remove the following associated entry from the /etc/fstab file:
/dev/VolGroup00/LogVol02 swap swap defaults 0 0Regenerate mount units so that your system registers the new configuration:
# systemctl daemon-reloadRemove all references to the removed swap storage from the /etc/default/grub file:
# vi /etc/default/grubRebuild the grub configuration:
on BIOS-based machines, run:
# grub2-mkconfig -o /boot/grub2/grub.cfgon UEFI-based machines, run:
# grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg
To test if the logical volume was successfully removed, inspect active swap space:
$ cat /proc/swaps $ free -h
15.2.3. Removing a Swap File
To remove a swap file:
Procedure 15.5. Remove a Swap File
At a shell prompt, execute the following command to disable the swap file (where /swapfile is the swap file):
# swapoff -v /swapfileRemove its entry from the /etc/fstab file.
Regenerate mount units so that your system registers the new configuration:
# systemctl daemon-reloadRemove the actual file:
# rm /swapfile
15.3. Moving Swap Space
To move swap space from one location to another:
Chapter 16. System Storage Manager (SSM)
System Storage Manager (SSM) provides a command line interface to manage storage in various technologies. Storage systems are becoming increasingly complicated through the use of Device Mappers (DM), Logical Volume Managers (LVM), and Multiple Devices (MD). This creates a system that is not user friendly and makes it easier for errors and problems to arise. SSM alleviates this by creating a unified user interface. This interface allows users to run complicated systems in a simple manner. For example, to create and mount a new file system without SSM, there are five commands that must be used. With SSM only one is needed.
This chapter explains how SSM interacts with various back ends and some common use cases.
16.1. SSM Back Ends
SSM uses a core abstraction layer in ssmlib/main.py which complies with the device, pool, and volume abstraction, ignoring the specifics of the underlying technology. Back ends can be registered in ssmlib/main.py to handle specific storage technology methods, such as create, snapshot, or to remove volumes and pools.
There are already several back ends registered in SSM. The following sections provide basic information on them as well as definitions on how they handle pools, volumes, snapshots, and devices.
16.1.1. Btrfs Back End
Btrfs is available as a Technology Preview feature in Red Hat Enterprise Linux 7 but has been deprecated since the Red Hat Enterprise Linux 7.4 release. It will be removed in a future major release of Red Hat Enterprise Linux.
For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.4 Release Notes.
Btrfs, a file system with many advanced features, is used as a volume management back end in SSM. Pools, volumes, and snapshots can be created with the Btrfs back end.
16.1.1.1. Btrfs Pool
The Btrfs file system itself is the pool. It can be extended by adding more devices or shrunk by removing devices. SSM creates a Btrfs file system when a Btrfs pool is created. This means that every new Btrfs pool has one volume of the same name as the pool which cannot be removed without removing the entire pool. The default Btrfs pool name is btrfs_pool.
The name of the pool is used as the file system label. If there is already an existing Btrfs file system in the system without a label, the Btrfs pool will generate a name for internal use in the format of btrfs_device_base_name.
16.1.1.2. Btrfs Volume
Volumes created after the first volume in a pool are the same as sub-volumes. SSM temporarily mounts the Btrfs file system if it is unmounted in order to create a sub-volume.
The name of a volume is used as the subvolume path in the Btrfs file system. For example, a subvolume displays as /dev/lvm_pool/lvol001. Every object in this path must exist in order for the volume to be created. Volumes can also be referenced with its mount point.
16.1.1.3. Btrfs Snapshot
Snapshots can be taken of any Btrfs volume in the system with SSM. Be aware that Btrfs does not distinguish between subvolumes and snapshots. While this means that SSM cannot recognize the Btrfs snapshot destination, it will try to recognize special name formats. If the name specified when creating the snapshot does the specific pattern, the snapshot is not be recognized and instead be listed as a regular Btrfs volume.
16.1.1.4. Btrfs Device
Btrfs does not require any special device to be created on.
16.1.2. LVM Back End
Pools, volumes, and snapshots can be created with LVM. The following definitions are from an LVM point of view.
16.1.2.1. LVM Pool
LVM pool is the same as an LVM volume group. This means that grouping devices and new logical volumes can be created out of the LVM pool. The default LVM pool name is lvm_pool.
16.1.2.2. LVM Volume
An LVM volume is the same as an ordinary logical volume.
16.1.2.3. LVM Snapshot
When a snapshot is created from the LVM volume a new snapshot volume is created which can then be handled just like any other LVM volume. Unlike Btrfs, LVM is able to distinguish snapshots from regular volumes so there is no need for a snapshot name to match a particular pattern.
16.1.2.4. LVM Device
SSM makes the need for an LVM back end to be created on a physical device transparent for the user.
16.1.3. Crypt Back End
The crypt back end in SSM uses cryptsetup and dm-crypt target to manage encrypted volumes. Crypt back ends can be used as a regular back end for creating encrypted volumes on top of regular block devices (or on other volumes such as LVM or MD volumes), or to create encrypted LVM volumes in a single steps.
Only volumes can be created with a crypt back end; pooling is not supported and it does not require special devices.
The following sections define volumes and snapshots from the crypt point of view.
16.1.3.1. Crypt Volume
Crypt volumes are created by dm-crypt and represent the data on the original encrypted device in an unencrypted form. It does not support RAID or any device concatenation.
Two modes, or extensions, are supported: luks and plain. Luks is used by default. For more information on the extensions, see man cryptsetup.
16.1.3.2. Crypt Snapshot
While the crypt back end does not support snapshotting, if the encrypted volume is created on top of an LVM volume, the volume itself can be snapshotted. The snapshot can then be opened by using cryptsetup.
16.1.4. Multiple Devices (MD) Back End
MD back end is currently limited to only gathering the information about MD volumes in the system.
16.2. Common SSM Tasks
The following sections describe common SSM tasks.
16.2.1. Installing SSM
To install SSM use the following command:
# yum install system-storage-managerThere are several back ends that are enabled only if the supporting packages are installed:
The LVM back end requires the lvm2 package.
The Btrfs back end requires the btrfs-progs package.
The Crypt back end requires the device-mapper and cryptsetup packages.
16.2.2. Displaying Information about All Detected Devices
Displaying information about all detected devices, pools, volumes, and snapshots is done with the list command. The ssm list command with no options display the following output:
# ssm list ---------------------------------------------------------- Device Free Used Total Pool Mount point ---------------------------------------------------------- /dev/sda 2.00 GB PARTITIONED /dev/sda1 47.83 MB /test /dev/vda 15.00 GB PARTITIONED /dev/vda1 500.00 MB /boot /dev/vda2 0.00 KB 14.51 GB 14.51 GB rhel ---------------------------------------------------------- ------------------------------------------------ Pool Type Devices Free Used Total ------------------------------------------------ rhel lvm 1 0.00 KB 14.51 GB 14.51 GB ------------------------------------------------ --------------------------------------------------------------------------------- Volume Pool Volume size FS FS size Free Type Mount point --------------------------------------------------------------------------------- /dev/rhel/root rhel 13.53 GB xfs 13.52 GB 9.64 GB linear / /dev/rhel/swap rhel 1000.00 MB linear /dev/sda1 47.83 MB xfs 44.50 MB 44.41 MB part /test /dev/vda1 500.00 MB xfs 496.67 MB 403.56 MB part /boot ---------------------------------------------------------------------------------This display can be further narrowed down by using arguments to specify what should be displayed. The list of available options can be found with the ssm list --help command.
Depending on the argument given, SSM may not display everything.
Running the devices or dev argument omits some devices. CDRoms and DM/MD devices, for example, are intentionally hidden as they are listed as volumes.
Some back ends do not support snapshots and cannot distinguish between a snapshot and a regular volume. Running the snapshot argument on one of these back ends cause SSM to attempt to recognize the volume name in order to identify a snapshot. If the SSM regular expression does not match the snapshot pattern then the snapshot is not be recognized.
With the exception of the main Btrfs volume (the file system itself), any unmounted Btrfs volumes are not shown.
16.2.3. Creating a New Pool, Logical Volume, and File System
In this section, a new pool is be created with a default name which have the devices /dev/vdb and /dev/vdc, a logical volume of 1G, and an XFS file system.
The command to create this scenario is ssm create --fs xfs -s 1G /dev/vdb /dev/vdc. The following options are used:
The --fs option specifies the required file system type. Current supported file system types are:
ext3
ext4
xfs
btrfs
The -s specifies the size of the logical volume. The following suffixes are supported to define units:
K or k for kilobytes
M or m for megabytes
G or g for gigabytes
T or t for terabytes
P or p for petabytes
E or e for exabytes
The two listed devices, /dev/vdb and /dev/vdc, are the two devices you wish to create.
There are two other options for the ssm command that may be useful. The first is the -p pool command. This specifies the pool the volume is to be created on. If it does not yet exist, then SSM creates it. This was omitted in the given example which caused SSM to use the default name lvm_pool. However, to use a specific name to fit in with any existing naming conventions, the -p option should be used.
The second useful option is the -n name command. This names the newly created logical volume. As with the -p, this is needed in order to use a specific name to fit in with any existing naming conventions.
An example of these two options being used follows:
# ssm create --fs xfs -p new_pool -n XFS_Volume /dev/vdd Volume group "new_pool" successfully created Logical volume "XFS_Volume" createdSSM has now created two physical volumes, a pool, and a logical volume with the ease of only one command.
16.2.4. Checking a File System's Consistency
The ssm check command checks the file system consistency on the volume. It is possible to specify multiple volumes to check. If there is no file system on the volume, then the volume is skipped.
To check all devices in the volume lvol001, run the command ssm check /dev/lvm_pool/lvol001.
# ssm check /dev/lvm_pool/lvol001 Checking xfs file system on '/dev/mapper/lvm_pool-lvol001'. Phase 1 - find and verify superblock... Phase 2 - using internal log - scan filesystem freespace and inode maps... - found root inode chunk Phase 3 - for each AG... - scan (but don't clear) agi unlinked lists... - process known inodes and perform inode discovery... - agno = 0 - agno = 1 - agno = 2 - agno = 3 - agno = 4 - agno = 5 - agno = 6 - process newly discovered inodes... Phase 4 - check for duplicate blocks... - setting up duplicate extent list... - check for inodes claiming duplicate blocks... - agno = 0 - agno = 1 - agno = 2 - agno = 3 - agno = 4 - agno = 5 - agno = 6 No modify flag set, skipping phase 5 Phase 6 - check inode connectivity... - traversing filesystem ... - traversal finished ... - moving disconnected inodes to lost+found ... Phase 7 - verify link counts... No modify flag set, skipping filesystem flush and exiting.16.2.5. Increasing a Volume's Size
The ssm resize command changes the size of the specified volume and file system. If there is no file system then only the volume itself will be resized.
For this example, we currently have one logical volume on /dev/vdb that is 900MB called lvol001.
# ssm list ----------------------------------------------------------------- Device Free Used Total Pool Mount point ----------------------------------------------------------------- /dev/vda 15.00 GB PARTITIONED /dev/vda1 500.00 MB /boot /dev/vda2 0.00 KB 14.51 GB 14.51 GB rhel /dev/vdb 120.00 MB 900.00 MB 1.00 GB lvm_pool /dev/vdc 1.00 GB ----------------------------------------------------------------- --------------------------------------------------------- Pool Type Devices Free Used Total --------------------------------------------------------- lvm_pool lvm 1 120.00 MB 900.00 MB 1020.00 MB rhel lvm 1 0.00 KB 14.51 GB 14.51 GB --------------------------------------------------------- -------------------------------------------------------------------------------------------- Volume Pool Volume size FS FS size Free Type Mount point -------------------------------------------------------------------------------------------- /dev/rhel/root rhel 13.53 GB xfs 13.52 GB 9.64 GB linear / /dev/rhel/swap rhel 1000.00 MB linear /dev/lvm_pool/lvol001 lvm_pool 900.00 MB xfs 896.67 MB 896.54 MB linear /dev/vda1 500.00 MB xfs 496.67 MB 403.56 MB part /boot --------------------------------------------------------------------------------------------The logical volume needs to be increased by another 500MB. To do so we will need to add an extra device to the pool:
~]# ssm resize -s +500M /dev/lvm_pool/lvol001 /dev/vdc Physical volume "/dev/vdc" successfully created Volume group "lvm_pool" successfully extended Phase 1 - find and verify superblock... Phase 2 - using internal log - scan filesystem freespace and inode maps... - found root inode chunk Phase 3 - for each AG... - scan (but don't clear) agi unlinked lists... - process known inodes and perform inode discovery... - agno = 0 - agno = 1 - agno = 2 - agno = 3 - process newly discovered inodes... Phase 4 - check for duplicate blocks... - setting up duplicate extent list... - check for inodes claiming duplicate blocks... - agno = 0 - agno = 1 - agno = 2 - agno = 3 No modify flag set, skipping phase 5 Phase 6 - check inode connectivity... - traversing filesystem ... - traversal finished ... - moving disconnected inodes to lost+found ... Phase 7 - verify link counts... No modify flag set, skipping filesystem flush and exiting. Extending logical volume lvol001 to 1.37 GiB Logical volume lvol001 successfully resized meta-data=/dev/mapper/lvm_pool-lvol001 isize=256 agcount=4, agsize=57600 blks = sectsz=512 attr=2, projid32bit=1 = crc=0 data = bsize=4096 blocks=230400, imaxpct=25 = sunit=0 swidth=0 blks naming =version 2 bsize=4096 ascii-ci=0 ftype=0 log =internal bsize=4096 blocks=853, version=2 = sectsz=512 sunit=0 blks, lazy-count=1 realtime =none extsz=4096 blocks=0, rtextents=0 data blocks changed from 230400 to 358400SSM runs a check on the device and then extends the volume by the specified amount. This can be verified with the ssm list command.
# ssm list ------------------------------------------------------------------ Device Free Used Total Pool Mount point ------------------------------------------------------------------ /dev/vda 15.00 GB PARTITIONED /dev/vda1 500.00 MB /boot /dev/vda2 0.00 KB 14.51 GB 14.51 GB rhel /dev/vdb 0.00 KB 1020.00 MB 1.00 GB lvm_pool /dev/vdc 640.00 MB 380.00 MB 1.00 GB lvm_pool ------------------------------------------------------------------ ------------------------------------------------------ Pool Type Devices Free Used Total ------------------------------------------------------ lvm_pool lvm 2 640.00 MB 1.37 GB 1.99 GB rhel lvm 1 0.00 KB 14.51 GB 14.51 GB ------------------------------------------------------ ---------------------------------------------------------------------------------------------- Volume Pool Volume size FS FS size Free Type Mount point ---------------------------------------------------------------------------------------------- /dev/rhel/root rhel 13.53 GB xfs 13.52 GB 9.64 GB linear / /dev/rhel/swap rhel 1000.00 MB linear /dev/lvm_pool/lvol001 lvm_pool 1.37 GB xfs 1.36 GB 1.36 GB linear /dev/vda1 500.00 MB xfs 496.67 MB 403.56 MB part /boot ----------------------------------------------------------------------------------------------It is only possible to decrease an LVM volume's size; it is not supported with other volume types. This is done by using a - instead of a +. For example, to decrease the size of an LVM volume by 50M the command would be:
# ssm resize -s-50M /dev/lvm_pool/lvol002 Rounding size to boundary between physical extents: 972.00 MiB WARNING: Reducing active logical volume to 972.00 MiB THIS MAY DESTROY YOUR DATA (filesystem etc.) Do you really want to reduce lvol002? [y/n]: y Reducing logical volume lvol002 to 972.00 MiB Logical volume lvol002 successfully resizedWithout either the + or -, the value is taken as absolute.
16.2.6. Snapshot
To take a snapshot of an existing volume, use the ssm snapshot command.
This operation fails if the back end that the volume belongs to does not support snapshotting.
To create a snapshot of the lvol001, use the following command:
# ssm snapshot /dev/lvm_pool/lvol001 Logical volume "snap20150519T130900" createdTo verify this, use the ssm list, and note the extra snapshot section.
# ssm list ---------------------------------------------------------------- Device Free Used Total Pool Mount point ---------------------------------------------------------------- /dev/vda 15.00 GB PARTITIONED /dev/vda1 500.00 MB /boot /dev/vda2 0.00 KB 14.51 GB 14.51 GB rhel /dev/vdb 0.00 KB 1020.00 MB 1.00 GB lvm_pool /dev/vdc 1.00 GB ---------------------------------------------------------------- -------------------------------------------------------- Pool Type Devices Free Used Total -------------------------------------------------------- lvm_pool lvm 1 0.00 KB 1020.00 MB 1020.00 MB rhel lvm 1 0.00 KB 14.51 GB 14.51 GB -------------------------------------------------------- ---------------------------------------------------------------------------------------------- Volume Pool Volume size FS FS size Free Type Mount point ---------------------------------------------------------------------------------------------- /dev/rhel/root rhel 13.53 GB xfs 13.52 GB 9.64 GB linear / /dev/rhel/swap rhel 1000.00 MB linear /dev/lvm_pool/lvol001 lvm_pool 900.00 MB xfs 896.67 MB 896.54 MB linear /dev/vda1 500.00 MB xfs 496.67 MB 403.56 MB part /boot ---------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------- Snapshot Origin Pool Volume size Size Type ---------------------------------------------------------------------------------- /dev/lvm_pool/snap20150519T130900 lvol001 lvm_pool 120.00 MB 0.00 KB linear ----------------------------------------------------------------------------------16.2.7. Removing an Item
The ssm remove is used to remove an item, either a device, pool, or volume.
If a device is being used by a pool when removed, it will fail. This can be forced using the -f argument.
If the volume is mounted when removed, it will fail. Unlike the device, it cannot be forced with the -f argument.
To remove the lvm_pool and everything within it use the following command:
# ssm remove lvm_pool Do you really want to remove volume group "lvm_pool" containing 2 logical volumes? [y/n]: y Do you really want to remove active logical volume snap20150519T130900? [y/n]: y Logical volume "snap20150519T130900" successfully removed Do you really want to remove active logical volume lvol001? [y/n]: y Logical volume "lvol001" successfully removed Volume group "lvm_pool" successfully removed16.3. SSM Resources
For more information on SSM, see the following resources:
The man ssm page provides good descriptions and examples, as well as details on all of the commands and options too specific to be documented here.
Local documentation for SSM is stored in the doc/ directory.
Chapter 17. Disk Quotas
Disk space can be restricted by implementing disk quotas which alert a system administrator before a user consumes too much disk space or a partition becomes full.
Disk quotas can be configured for individual users as well as user groups. This makes it possible to manage the space allocated for user-specific files (such as email) separately from the space allocated to the projects a user works on (assuming the projects are given their own groups).
In addition, quotas can be set not just to control the number of disk blocks consumed but to control the number of inodes (data structures that contain information about files in UNIX file systems). Because inodes are used to contain file-related information, this allows control over the number of files that can be created.
The quota RPM must be installed to implement disk quotas.
This chapter is for all file systems, however some file systems have their own quota management tools. See the corresponding description for the applicable file systems.
Btrfs does not have disk quotas so is not covered.
17.1. Configuring Disk Quotas
To implement disk quotas, use the following steps:
Enable quotas per file system by modifying the /etc/fstab file.
Remount the file system(s).
Create the quota database files and generate the disk usage table.
Assign quota policies.
Each of these steps is discussed in detail in the following sections.
17.1.1. Enabling Quotas
Procedure 17.1. Enabling Quotas
Log in as root.
Edit the /etc/fstab file.
Add either the usrquota or grpquota or both options to the file systems that require quotas.
Example 17.1. Edit /etc/fstab
For example, to use the text editor vim type the following:
# vim /etc/fstabExample 17.2. Add Quotas
/dev/VolGroup00/LogVol00 / ext3 defaults 1 1 LABEL=/boot /boot ext3 defaults 1 2 none /dev/pts devpts gid=5,mode=620 0 0 none /dev/shm tmpfs defaults 0 0 none /proc proc defaults 0 0 none /sys sysfs defaults 0 0 /dev/VolGroup00/LogVol02 /home ext3 defaults,usrquota,grpquota 1 2 /dev/VolGroup00/LogVol01 swap swap defaults 0 0 . . .
In this example, the /home file system has both user and group quotas enabled.
The following examples assume that a separate /home partition was created during the installation of Red Hat Enterprise Linux. The root (/) partition can be used for setting quota policies in the /etc/fstab file.
17.1.2. Remounting the File Systems
After adding either the usrquota or grpquota or both options, remount each file system whose fstab entry has been modified. If the file system is not in use by any process, use one of the following methods:
Run the umount command followed by the mount command to remount the file system. See the man page for both umount and mount for the specific syntax for mounting and unmounting various file system types.
Run the mount -o remount file-system command (where file-system is the name of the file system) to remount the file system. For example, to remount the /home file system, run the mount -o remount /home command.
If the file system is currently in use, the easiest method for remounting the file system is to reboot the system.
17.1.3. Creating the Quota Database Files
After each quota-enabled file system is remounted run the quotacheck command.
The quotacheck command examines quota-enabled file systems and builds a table of the current disk usage per file system. The table is then used to update the operating system's copy of disk usage. In addition, the file system's disk quota files are updated.
The quotacheck command has no effect on XFS as the table of disk usage is completed automatically at mount time. See the man page xfs_quota(8) for more information.
Procedure 17.2. Creating the Quota Database Files
Create the quota files on the file system using the following command:
# quotacheck -cug /file systemGenerate the table of current disk usage per file system using the following command:
# quotacheck -avug
Following are the options used to create quota files:
c
Specifies that the quota files should be created for each file system with quotas enable.
uChecks for user quotas.
gChecks for group quotas. If only -g is specified, only the group quota file is created.
If neither the -u or -g options are specified, only the user quota file is created.
The following options are used to generate the table of current disk usage:
a
Check all quota-enabled, locally-mounted file systems
vDisplay verbose status information as the quota check proceeds
uCheck user disk quota information
gCheck group disk quota information
After quotacheck has finished running, the quota files corresponding to the enabled quotas (either user or group or both) are populated with data for each quota-enabled locally-mounted file system such as /home.
17.1.4. Assigning Quotas per User
The last step is assigning the disk quotas with the edquota command.
Prerequisite
User must exist prior to setting the user quota.
Procedure 17.3. Assigning Quotas per User
To assign the quota for a user, use the following command:
# edquota usernameReplace username with the user to which you want to assign the quotas.
To verify that the quota for the user has been set, use the following command:
# quota username
Example 17.3. Assigning Quotas to a user
For example, if a quota is enabled in /etc/fstab for the /home partition (/dev/VolGroup00/LogVol02 in the following example) and the command edquota testuser is executed, the following is shown in the editor configured as the default for the system:
Disk quotas for user testuser (uid 501): Filesystem blocks soft hard inodes soft hard /dev/VolGroup00/LogVol02 440436 0 0 37418 0 0The text editor defined by the EDITOR environment variable is used by edquota. To change the editor, set the EDITOR environment variable in your ~/.bash_profile file to the full path of the editor of your choice.
The first column is the name of the file system that has a quota enabled for it. The second column shows how many blocks the user is currently using. The next two columns are used to set soft and hard block limits for the user on the file system. The inodes column shows how many inodes the user is currently using. The last two columns are used to set the soft and hard inode limits for the user on the file system.
The hard block limit is the absolute maximum amount of disk space that a user or group can use. Once this limit is reached, no further disk space can be used.
The soft block limit defines the maximum amount of disk space that can be used. However, unlike the hard limit, the soft limit can be exceeded for a certain amount of time. That time is known as the grace period. The grace period can be expressed in seconds, minutes, hours, days, weeks, or months.
If any of the values are set to 0, that limit is not set. In the text editor, change the desired limits.
Example 17.4. Change Desired Limits
For example:
Disk quotas for user testuser (uid 501): Filesystem blocks soft hard inodes soft hard /dev/VolGroup00/LogVol02 440436 500000 550000 37418 0 0To verify that the quota for the user has been set, use the command:
# quota testuser Disk quotas for user username (uid 501): Filesystem blocks quota limit grace files quota limit grace /dev/sdb 1000* 1000 1000 0 0 017.1.5. Assigning Quotas per Group
Quotas can also be assigned on a per-group basis.
Prerequisite
Group must exist prior to setting the group quota.
Procedure 17.4. Assigning Quotas per Group
To set a group quota, use the following command:
# edquota -g groupnameTo verify that the group quota is set, use the following command:
# quota -g groupname
Example 17.5. Assigning quotas to group
For example, to set a group quota for the devel group, use the command:
# edquota -g develThis command displays the existing quota for the group in the text editor:
Disk quotas for group devel (gid 505): Filesystem blocks soft hard inodes soft hard /dev/VolGroup00/LogVol02 440400 0 0 37418 0 0Modify the limits, then save the file.
To verify that the group quota has been set, use the command:
# quota -g devel17.1.6. Setting the Grace Period for Soft Limits
If a given quota has soft limits, you can edit the grace period (i.e. the amount of time a soft limit can be exceeded) with the following command:
# edquota -tThis command works on quotas for inodes or blocks, for either users or groups.
While other edquota commands operate on quotas for a particular user or group, the -t option operates on every file system with quotas enabled.
17.2. Managing Disk Quotas
If quotas are implemented, they need some maintenance mostly in the form of watching to see if the quotas are exceeded and making sure the quotas are accurate.
If users repeatedly exceed their quotas or consistently reach their soft limits, a system administrator has a few choices to make depending on what type of users they are and how much disk space impacts their work. The administrator can either help the user determine how to use less disk space or increase the user's disk quota.
17.2.1. Enabling and Disabling
It is possible to disable quotas without setting them to 0. To turn all user and group quotas off, use the following command:
# quotaoff -vaugIf neither the -u or -g options are specified, only the user quotas are disabled. If only -g is specified, only group quotas are disabled. The -v switch causes verbose status information to display as the command executes.
To enable user and group quotas again, use the following command:
# quotaonTo enable user and group quotas for all file systems, use the following command:
# quotaon -vaugIf neither the -u or -g options are specified, only the user quotas are enabled. If only -g is specified, only group quotas are enabled.
To enable quotas for a specific file system, such as /home, use the following command:
# quotaon -vug /homeThe quotaon command is not always needed for XFS because it is performed automatically at mount time. Refer to the man page quotaon(8) for more information.
17.2.2. Reporting on Disk Quotas
Creating a disk usage report entails running the repquota utility.
Example 17.6. Output of the repquota Command
For example, the command repquota /home produces this output:
*** Report for user quotas on device /dev/mapper/VolGroup00-LogVol02 Block grace time: 7days; Inode grace time: 7days Block limits File limits User used soft hard grace used soft hard grace ---------------------------------------------------------------------- root -- 36 0 0 4 0 0 kristin -- 540 0 0 125 0 0 testuser -- 440400 500000 550000 37418 0 0To view the disk usage report for all (option -a) quota-enabled file systems, use the command:
# repquota -aWhile the report is easy to read, a few points should be explained. The -- displayed after each user is a quick way to determine whether the block or inode limits have been exceeded. If either soft limit is exceeded, a + appears in place of the corresponding -; the first - represents the block limit, and the second represents the inode limit.
The grace columns are normally blank. If a soft limit has been exceeded, the column contains a time specification equal to the amount of time remaining on the grace period. If the grace period has expired, none appears in its place.
17.2.3. Keeping Quotas Accurate
When a file system fails to unmount cleanly, for example due to a system crash, it is necessary to run the following command:
# quotacheckHowever, quotacheck can be run on a regular basis, even if the system has not crashed. Safe methods for periodically running quotacheck include:
Ensuring quotacheck runs on next reboot
This method works best for (busy) multiuser systems which are periodically rebooted.
Save a shell script into the /etc/cron.daily/ or /etc/cron.weekly/ directory or schedule one using the following command:
# crontab -eThe crontab -e command contains the touch /forcequotacheck command. This creates an empty forcequotacheck file in the root directory, which the system init script looks for at boot time. If it is found, the init script runs quotacheck. Afterward, the init script removes the /forcequotacheck file; thus, scheduling this file to be created periodically with cron ensures that quotacheck is run during the next reboot.
For more information about cron, see man cron.
Running quotacheck in single user modeAn alternative way to safely run quotacheck is to boot the system into single-user mode to prevent the possibility of data corruption in quota files and run the following commands:
# quotaoff -vug /file_system# quotacheck -vug /file_system# quotaon -vug /file_system Running quotacheck on a running systemIf necessary, it is possible to run quotacheck on a machine during a time when no users are logged in, and thus have no open files on the file system being checked. Run the command quotacheck -vug file_system; this command will fail if quotacheck cannot remount the given file_system as read-only. Note that, following the check, the file system will be remounted read-write.
Running quotacheck on a live file system mounted read-write is not recommended due to the possibility of quota file corruption.
See man cron for more information about configuring cron.
17.3. Disk Quota References
For more information on disk quotas, refer to the man pages of the following commands:
quotacheck
edquota
repquota
quota
quotaon
quotaoff
Chapter 18. Redundant Array of Independent Disks (RAID)
The basic idea behind RAID is to combine multiple small, inexpensive disk drives into an array to accomplish performance or redundancy goals not attainable with one large and expensive drive. This array of drives appears to the computer as a single logical storage unit or drive.
RAID allows information to be spread across several disks. RAID uses techniques such as disk striping (RAID Level 0), disk mirroring (RAID Level 1), and disk striping with parity (RAID Level 5) to achieve redundancy, lower latency, increased bandwidth, and maximized ability to recover from hard disk crashes.
RAID distributes data across each drive in the array by breaking it down into consistently-sized chunks (commonly 256K or 512k, although other values are acceptable). Each chunk is then written to a hard drive in the RAID array according to the RAID level employed. When the data is read, the process is reversed, giving the illusion that the multiple drives in the array are actually one large drive.
System Administrators and others who manage large amounts of data would benefit from using RAID technology. Primary reasons to deploy RAID include:
Enhances speed
Increases storage capacity using a single virtual disk
Minimizes data loss from disk failure
18.1. RAID Types
There are three possible RAID approaches: Firmware RAID, Hardware RAID, and Software RAID.
Firmware RAID
Firmware RAID, also known as ATARAID, is a type of software RAID where the RAID sets can be configured using a firmware-based menu. The firmware used by this type of RAID also hooks into the BIOS, allowing you to boot from its RAID sets. Different vendors use different on-disk metadata formats to mark the RAID set members. The Intel Matrix RAID is a good example of a firmware RAID system.
Hardware RAID
The hardware-based array manages the RAID subsystem independently from the host. It presents a single disk per RAID array to the host.
A Hardware RAID device may be internal or external to the system, with internal devices commonly consisting of a specialized controller card that handles the RAID tasks transparently to the operating system and with external devices commonly connecting to the system via SCSI, Fibre Channel, iSCSI, InfiniBand, or other high speed network interconnect and presenting logical volumes to the system.
RAID controller cards function like a SCSI controller to the operating system, and handle all the actual drive communications. The user plugs the drives into the RAID controller (just like a normal SCSI controller) and then adds them to the RAID controllers configuration. The operating system will not be able to tell the difference.
Software RAID
Software RAID implements the various RAID levels in the kernel disk (block device) code. It offers the cheapest possible solution, as expensive disk controller cards or hot-swap chassis [2] are not required. Software RAID also works with cheaper IDE disks as well as SCSI disks. With today's faster CPUs, Software RAID also generally outperforms Hardware RAID.
The Linux kernel contains a multi-disk (MD) driver that allows the RAID solution to be completely hardware independent. The performance of a software-based array depends on the server CPU performance and load.
Key features of the Linux software RAID stack:
Multithreaded design
Portability of arrays between Linux machines without reconstruction
Backgrounded array reconstruction using idle system resources
Hot-swappable drive support
Automatic CPU detection to take advantage of certain CPU features such as streaming SIMD support
Automatic correction of bad sectors on disks in an array
Regular consistency checks of RAID data to ensure the health of the array
Proactive monitoring of arrays with email alerts sent to a designated email address on important events
Write-intent bitmaps which drastically increase the speed of resync events by allowing the kernel to know precisely which portions of a disk need to be resynced instead of having to resync the entire array
Resync checkpointing so that if you reboot your computer during a resync, at startup the resync will pick up where it left off and not start all over again
The ability to change parameters of the array after installation. For example, you can grow a 4-disk RAID5 array to a 5-disk RAID5 array when you have a new disk to add. This grow operation is done live and does not require you to reinstall on the new array.
18.2. RAID Levels and Linear Support
RAID supports various configurations, including levels 0, 1, 4, 5, 6, 10, and linear. These RAID types are defined as follows:
Level 0
RAID level 0, often called "striping," is a performance-oriented striped data mapping technique. This means the data being written to the array is broken down into strips and written across the member disks of the array, allowing high I/O performance at low inherent cost but provides no redundancy.
Many RAID level 0 implementations will only stripe the data across the member devices up to the size of the smallest device in the array. This means that if you have multiple devices with slightly different sizes, each device will get treated as though it is the same size as the smallest drive. Therefore, the common storage capacity of a level 0 array is equal to the capacity of the smallest member disk in a Hardware RAID or the capacity of smallest member partition in a Software RAID multiplied by the number of disks or partitions in the array.
Level 1RAID level 1, or "mirroring," has been used longer than any other form of RAID. Level 1 provides redundancy by writing identical data to each member disk of the array, leaving a "mirrored" copy on each disk. Mirroring remains popular due to its simplicity and high level of data availability. Level 1 operates with two or more disks, and provides very good data reliability and improves performance for read-intensive applications but at a relatively high cost. [3]
The storage capacity of the level 1 array is equal to the capacity of the smallest mirrored hard disk in a Hardware RAID or the smallest mirrored partition in a Software RAID. Level 1 redundancy is the highest possible among all RAID types, with the array being able to operate with only a single disk present.
Level 4Level 4 uses parity [4] concentrated on a single disk drive to protect data. Because the dedicated parity disk represents an inherent bottleneck on all write transactions to the RAID array, level 4 is seldom used without accompanying technologies such as write-back caching, or in specific circumstances where the system administrator is intentionally designing the software RAID device with this bottleneck in mind (such as an array that will have little to no write transactions once the array is populated with data). RAID level 4 is so rarely used that it is not available as an option in Anaconda. However, it could be created manually by the user if truly needed.
The storage capacity of Hardware RAID level 4 is equal to the capacity of the smallest member partition multiplied by the number of partitions minus one. Performance of a RAID level 4 array will always be asymmetrical, meaning reads will outperform writes. This is because writes consume extra CPU and main memory bandwidth when generating parity, and then also consume extra bus bandwidth when writing the actual data to disks because you are writing not only the data, but also the parity. Reads need only read the data and not the parity unless the array is in a degraded state. As a result, reads generate less traffic to the drives and across the busses of the computer for the same amount of data transfer under normal operating conditions.
Level 5This is the most common type of RAID. By distributing parity across all of an array's member disk drives, RAID level 5 eliminates the write bottleneck inherent in level 4. The only performance bottleneck is the parity calculation process itself. With modern CPUs and Software RAID, that is usually not a bottleneck at all since modern CPUs can generate parity very fast. However, if you have a sufficiently large number of member devices in a software RAID5 array such that the combined aggregate data transfer speed across all devices is high enough, then this bottleneck can start to come into play.
As with level 4, level 5 has asymmetrical performance, with reads substantially outperforming writes. The storage capacity of RAID level 5 is calculated the same way as with level 4.
Level 6This is a common level of RAID when data redundancy and preservation, and not performance, are the paramount concerns, but where the space inefficiency of level 1 is not acceptable. Level 6 uses a complex parity scheme to be able to recover from the loss of any two drives in the array. This complex parity scheme creates a significantly higher CPU burden on software RAID devices and also imposes an increased burden during write transactions. As such, level 6 is considerably more asymmetrical in performance than levels 4 and 5.
The total capacity of a RAID level 6 array is calculated similarly to RAID level 5 and 4, except that you must subtract 2 devices (instead of 1) from the device count for the extra parity storage space.
Level 10This RAID level attempts to combine the performance advantages of level 0 with the redundancy of level 1. It also helps to alleviate some of the space wasted in level 1 arrays with more than 2 devices. With level 10, it is possible to create a 3-drive array configured to store only 2 copies of each piece of data, which then allows the overall array size to be 1.5 times the size of the smallest devices instead of only equal to the smallest device (like it would be with a 3-device, level 1 array).
The number of options available when creating level 10 arrays as well as the complexity of selecting the right options for a specific use case make it impractical to create during installation. It is possible to create one manually using the command line mdadm tool. For more information on the options and their respective performance trade-offs, see man md.
Linear RAIDLinear RAID is a grouping of drives to create a larger virtual drive. In linear RAID, the chunks are allocated sequentially from one member drive, going to the next drive only when the first is completely filled. This grouping provides no performance benefit, as it is unlikely that any I/O operations split between member drives. Linear RAID also offers no redundancy and decreases reliability; if any one member drive fails, the entire array cannot be used. The capacity is the total of all member disks.
18.3. Linux RAID Subsystems
RAID in Linux is composed of the following subsystems:
Linux Hardware RAID Controller Drivers
Hardware RAID controllers have no specific RAID subsystem in Linux. Because they use special RAID chipsets, hardware RAID controllers come with their own drivers; these drivers allow the system to detect the RAID sets as regular disks.
mdraid
The mdraid subsystem was designed as a software RAID solution for Linux; it is also the preferred solution for software RAID under Linux. This subsystem uses its own metadata format, generally referred to as native mdraid metadata.
mdraid also supports other metadata formats, known as external metadata. Red Hat Enterprise Linux 7 uses mdraid with external metadata to access ISW / IMSM (Intel firmware RAID) sets. mdraid sets are configured and controlled through the mdadm utility.
dmraid
The dmraid tool is used on a wide variety of firmware RAID implementations. dmraid also supports Intel firmware RAID, although Red Hat Enterprise Linux 7 uses mdraid to access Intel firmware RAID sets.
dmraid has been deprecated since the Red Hat Enterprise Linux 7.5 release. It will be removed in a future major release of Red Hat Enterprise Linux. For more information, see Deprecated Functionality in the Red Hat Enterprise Linux 7.5 Release Notes.
18.4. RAID Support in the Anaconda Installer
The Anaconda installer automatically detects any hardware and firmware RAID sets on a system, making them available for installation. Anaconda also supports software RAID using mdraid, and can recognize existing mdraid sets.
Anaconda provides utilities for creating RAID sets during installation; however, these utilities only allow partitions (as opposed to entire disks) to be members of new sets. To use an entire disk for a set, create a partition on it spanning the entire disk, and use that partition as the RAID set member.
When the root file system uses a RAID set, Anaconda adds special kernel command-line options to the bootloader configuration telling the initrd which RAID set(s) to activate before searching for the root file system.
For instructions on configuring RAID during installation, see the Red Hat Enterprise Linux 7 Installation Guide.
18.5. Converting Root Disk to RAID1 after Installation
On the PowerPC (PPC) architecture, take the following additional steps:
Copy the contents of the PowerPC Reference Platform (PReP) boot partition from /dev/sda1 to /dev/sdb1:
# dd if=/dev/sda1 of=/dev/sdb1Update the Prep and boot flag on the first partition on both disks:
$ parted /dev/sda set 1 prep on $ parted /dev/sda set 1 boot on $ parted /dev/sdb set 1 prep on $ parted /dev/sdb set 1 boot on
Running the grub2-install /dev/sda command does not work on a PowerPC machine and returns an error, but the system boots as expected.
18.6. Configuring RAID Sets
Most RAID sets are configured during creation, typically through the firmware menu or from the installer. In some cases, you may need to create or modify RAID sets after installing the system, preferably without having to reboot the machine and enter the firmware menu to do so.
Some hardware RAID controllers allow you to configure RAID sets on-the-fly or even define completely new sets after adding extra disks. This requires the use of driver-specific utilities, as there is no standard API for this. For more information, see your hardware RAID controller's driver documentation for information on this.
mdadm
The mdadm command-line tool is used to manage software RAID in Linux, i.e. mdraid. For information on the different mdadm modes and options, see man mdadm. The man page also contains useful examples for common operations like creating, monitoring, and assembling software RAID arrays.
dmraid
As the name suggests, dmraid is used to manage device-mapper RAID sets. The dmraid tool finds ATARAID devices using multiple metadata format handlers, each supporting various formats. For a complete list of supported formats, run dmraid -l.
As mentioned earlier in Section 18.3, “Linux RAID Subsystems”, the dmraid tool cannot configure RAID sets after creation. For more information about using dmraid, see man dmraid.
18.7. Creating Advanced RAID Devices
In some cases, you may wish to install the operating system on an array that can't be created after the installation completes. Usually, this means setting up the /boot or root file system arrays on a complex RAID device; in such cases, you may need to use array options that are not supported by Anaconda. To work around this, perform the following procedure:
Procedure 18.1. Creating Advanced RAID Devices
Insert the install disk.
During the initial boot up, select instead of or . When the system fully boots into Rescue mode, the user will be presented with a command line terminal.
From this terminal, use parted to create RAID partitions on the target hard drives. Then, use mdadm to manually create raid arrays from those partitions using any and all settings and options available. For more information on how to do these, see Chapter 13, Partitions, man parted, and man mdadm.
Once the arrays are created, you can optionally create file systems on the arrays as well.
Reboot the computer and this time select or to install as normal. As Anaconda searches the disks in the system, it will find the pre-existing RAID devices.
When asked about how to use the disks in the system, select and click Next. In the device listing, the pre-existing MD RAID devices will be listed.
Select a RAID device, click Edit and configure its mount point and (optionally) the type of file system it should use (if you did not create one earlier) then click Done. Anaconda will perform the install to this pre-existing RAID device, preserving the custom options you selected when you created it in Rescue Mode.
The limited Rescue Mode of the installer does not include man pages. Both the man mdadm and man md contain useful information for creating custom RAID arrays, and may be needed throughout the workaround. As such, it can be helpful to either have access to a machine with these man pages present, or to print them out prior to booting into Rescue Mode and creating your custom arrays.
Chapter 19. Using the mount Command
On Linux, UNIX, and similar operating systems, file systems on different partitions and removable devices (CDs, DVDs, or USB flash drives for example) can be attached to a certain point (the mount point) in the directory tree, and then detached again. To attach or detach a file system, use the mount or umount command respectively. This chapter describes the basic use of these commands, as well as some advanced topics, such as moving a mount point or creating shared subtrees.
19.1. Listing Currently Mounted File Systems
To display all currently attached file systems, use the following command with no additional arguments:
$ mountThis command displays the list of known mount points. Each line provides important information about the device name, the file system type, the directory in which it is mounted, and relevant mount options in the following form:
device on directory type type (options)
The findmnt utility, which allows users to list mounted file systems in a tree-like form, is also available from Red Hat Enterprise Linux 6.1. To display all currently attached file systems, run the findmnt command with no additional arguments:
$ findmnt19.1.1. Specifying the File System Type
By default, the output of the mount command includes various virtual file systems such as sysfs and tmpfs. To display only the devices with a certain file system type, provide the -t option:
$ mount -t typeSimilarly, to display only the devices with a certain file system using the findmnt command:
$ findmnt -t typeExample 19.1. Listing Currently Mounted ext4 File Systems
Usually, both / and /boot partitions are formatted to use ext4. To display only the mount points that use this file system, use the following command:
$ mount -t ext4 /dev/sda2 on / type ext4 (rw) /dev/sda1 on /boot type ext4 (rw)To list such mount points using the findmnt command, type:
$ findmnt -t ext4 TARGET SOURCE FSTYPE OPTIONS / /dev/sda2 ext4 rw,realtime,seclabel,barrier=1,data=ordered /boot /dev/sda1 ext4 rw,realtime,seclabel,barrier=1,data=ordered19.2. Mounting a File System
To attach a certain file system, use the mount command in the following form:
$ mount [option…] device directoryThe device can be identified by:
a full path to a block device: for example, /dev/sda3
a universally unique identifier (
UUID): for example, UUID=34795a28-ca6d-4fd8-a347-73671d0c19cb
a volume label: for example, LABEL=home
Note that while a file system is mounted, the original content of the directory is not accessible.
Linux does not prevent a user from mounting a file system to a directory with a file system already attached to it. To determine whether a particular directory serves as a mount point, run the findmnt utility with the directory as its argument and verify the exit code:
findmnt directory; echo $?If no file system is attached to the directory, the given command returns 1.
When you run the mount command without all required information, that is without the device name, the target directory, or the file system type, the mount reads the contents of the /etc/fstab file to check if the given file system is listed. The /etc/fstab file contains a list of device names and the directories in which the selected file systems are set to be mounted as well as the file system type and mount options. Therefore, when mounting a file system that is specified in /etc/fstab, you can choose one of the following options:
mount [option…] directory mount [option…] deviceTo determine the UUID and—if the device uses it—the label of a particular device, use the blkid command in the following form:
blkid deviceFor example, to display information about /dev/sda3:
# blkid /dev/sda3 /dev/sda3: LABEL="home" UUID="34795a28-ca6d-4fd8-a347-73671d0c19cb" TYPE="ext3"19.2.1. Specifying the File System Type
In most cases, mount detects the file system automatically. However, there are certain file systems, such as NFS (Network File System) or CIFS (Common Internet File System), that are not recognized, and need to be specified manually. To specify the file system type, use the mount command in the following form:
$ mount -t type device directoryTable 19.1. Common File System Types
ext2 | The ext2 file system. |
ext3 | The ext3 file system. |
ext4 | The ext4 file system. |
btrfs | The btrfs file system. |
xfs | The xfs file system. |
iso9660 | The ISO 9660 file system. It is commonly used by optical media, typically CDs. |
nfs | The NFS file system. It is commonly used to access files over the network. |
nfs4 | The NFSv4 file system. It is commonly used to access files over the network. |
udf | The UDF file system. It is commonly used by optical media, typically DVDs. |
vfat | The FAT file system. It is commonly used on machines that are running the Windows operating system, and on certain digital media such as USB flash drives or floppy disks. |
Example 19.2. Mounting a USB Flash Drive
Older USB flash drives often use the FAT file system. Assuming that such drive uses the /dev/sdc1 device and that the /media/flashdisk/ directory exists, mount it to this directory by typing the following at a shell prompt as root:
~]# mount -t vfat /dev/sdc1 /media/flashdisk19.2.2. Specifying the Mount Options
To specify additional mount options, use the command in the following form:
mount -o options device directoryWhen supplying multiple options, do not insert a space after a comma, or mount interprets incorrectly the values following spaces as additional parameters.
Table 19.2. Common Mount Options
async | Allows the asynchronous input/output operations on the file system. |
auto | Allows the file system to be mounted automatically using the mount -a command. |
defaults | Provides an alias for async,auto,dev,exec,nouser,rw,suid. |
exec | Allows the execution of binary files on the particular file system. |
loop | Mounts an image as a loop device. |
noauto | Default behavior disallows the automatic mount of the file system using the mount -a command. |
noexec | Disallows the execution of binary files on the particular file system. |
nouser | Disallows an ordinary user (that is, other than root) to mount and unmount the file system. |
remount | Remounts the file system in case it is already mounted. |
ro | Mounts the file system for reading only. |
rw | Mounts the file system for both reading and writing. |
user | Allows an ordinary user (that is, other than root) to mount and unmount the file system. |
Example 19.3. Mounting an ISO Image
An ISO image (or a disk image in general) can be mounted by using the loop device. Assuming that the ISO image of the Fedora 14 installation disc is present in the current working directory and that the /media/cdrom/ directory exists, mount the image to this directory by running the following command:
# mount -o ro,loop Fedora-14-x86_64-Live-Desktop.iso /media/cdromNote that ISO 9660 is by design a read-only file system.
19.2.3. Sharing Mounts
Occasionally, certain system administration tasks require access to the same file system from more than one place in the directory tree (for example, when preparing a chroot environment). This is possible, and Linux allows you to mount the same file system to as many directories as necessary. Additionally, the mount command implements the --bind option that provides a means for duplicating certain mounts. Its usage is as follows:
$ mount --bind old_directory new_directoryAlthough this command allows a user to access the file system from both places, it does not apply on the file systems that are mounted within the original directory. To include these mounts as well, use the following command:
$ mount --rbind old_directory new_directoryAdditionally, to provide as much flexibility as possible, Red Hat Enterprise Linux 7 implements the functionality known as shared subtrees. This feature allows the use of the following four mount types:
Shared Mount
A shared mount allows the creation of an exact replica of a given mount point. When a mount point is marked as a shared mount, any mount within the original mount point is reflected in it, and vice versa. To change the type of a mount point to a shared mount, type the following at a shell prompt:
$ mount --make-shared mount_pointAlternatively, to change the mount type for the selected mount point and all mount points under it:
$ mount --make-rshared mount_point Slave MountA slave mount allows the creation of a limited duplicate of a given mount point. When a mount point is marked as a slave mount, any mount within the original mount point is reflected in it, but no mount within a slave mount is reflected in its original. To change the type of a mount point to a slave mount, type the following at a shell prompt:
mount --make-slave mount_pointAlternatively, it is possible to change the mount type for the selected mount point and all mount points under it by typing:
mount --make-rslave mount_pointExample 19.5. Creating a Slave Mount Point
This example shows how to get the content of the /media/ directory to appear in /mnt/ as well, but without any mounts in the /mnt/ directory to be reflected in /media/. As root, first mark the /media/ directory as shared:
~]# mount --bind /media /media ~]# mount --make-shared /mediaThen create its duplicate in /mnt/, but mark it as "slave":
~]# mount --bind /media /mnt ~]# mount --make-slave /mntNow verify that a mount within /media/ also appears in /mnt/. For example, if the CD-ROM drive contains non-empty media and the /media/cdrom/ directory exists, run the following commands:
~]# mount /dev/cdrom /media/cdrom ~]# ls /media/cdrom EFI GPL isolinux LiveOS ~]# ls /mnt/cdrom EFI GPL isolinux LiveOSAlso verify that file systems mounted in the /mnt/ directory are not reflected in /media/. For instance, if a non-empty USB flash drive that uses the /dev/sdc1 device is plugged in and the /mnt/flashdisk/ directory is present, type:
~]# mount /dev/sdc1 /mnt/flashdisk ~]# ls /media/flashdisk ~]# ls /mnt/flashdisk en-US publican.cfg Private MountA private mount is the default type of mount, and unlike a shared or slave mount, it does not receive or forward any propagation events. To explicitly mark a mount point as a private mount, type the following at a shell prompt:
mount --make-private mount_pointAlternatively, it is possible to change the mount type for the selected mount point and all mount points under it:
mount --make-rprivate mount_pointExample 19.6. Creating a Private Mount Point
Taking into account the scenario in Example 19.4, “Creating a Shared Mount Point”, assume that a shared mount point has been previously created by using the following commands as root:
~]# mount --bind /media /media ~]# mount --make-shared /media ~]# mount --bind /media /mntTo mark the /mnt/ directory as private, type:
~]# mount --make-private /mntIt is now possible to verify that none of the mounts within /media/ appears in /mnt/. For example, if the CD-ROM drives contains non-empty media and the /media/cdrom/ directory exists, run the following commands:
~]# mount /dev/cdrom /media/cdrom ~]# ls /media/cdrom EFI GPL isolinux LiveOS ~]# ls /mnt/cdrom ~]#It is also possible to verify that file systems mounted in the /mnt/ directory are not reflected in /media/. For instance, if a non-empty USB flash drive that uses the /dev/sdc1 device is plugged in and the /mnt/flashdisk/ directory is present, type:
~]# mount /dev/sdc1 /mnt/flashdisk ~]# ls /media/flashdisk ~]# ls /mnt/flashdisk en-US publican.cfg Unbindable MountIn order to prevent a given mount point from being duplicated whatsoever, an unbindable mount is used. To change the type of a mount point to an unbindable mount, type the following at a shell prompt:
mount --make-unbindable mount_pointAlternatively, it is possible to change the mount type for the selected mount point and all mount points under it:
mount --make-runbindable mount_pointExample 19.7. Creating an Unbindable Mount Point
To prevent the /media/ directory from being shared, as root:
# mount --bind /media /media # mount --make-unbindable /mediaThis way, any subsequent attempt to make a duplicate of this mount fails with an error:
# mount --bind /media /mnt mount: wrong fs type, bad option, bad superblock on /media, missing codepage or helper program, or other error In some cases useful info is found in syslog - try dmesg | tail or so19.2.4. Moving a Mount Point
To change the directory in which a file system is mounted, use the following command:
# mount --move old_directory new_directoryExample 19.8. Moving an Existing NFS Mount Point
An NFS storage contains user directories and is already mounted in /mnt/userdirs/. As root, move this mount point to /home by using the following command:
# mount --move /mnt/userdirs /homeTo verify the mount point has been moved, list the content of both directories:
# ls /mnt/userdirs # ls /home jill joe19.2.5. Setting Read-only Permissions for root
Sometimes, you need to mount the root file system with read-only permissions. Example use cases include enhancing security or ensuring data integrity after an unexpected system power-off.
19.2.5.1. Configuring root to Mount with Read-only Permissions on Boot
In the /etc/sysconfig/readonly-root file, change READONLY to yes:
# Set to 'yes' to mount the file systems as read-only. READONLY=yes [output truncated]Change defaults to ro in the root entry (/) in the /etc/fstab file:
/dev/mapper/luks-c376919e... / ext4 ro,x-systemd.device-timeout=0 1 1Add ro to the GRUB_CMDLINE_LINUX directive in the /etc/default/grub file and ensure that it does not contain rw:
GRUB_CMDLINE_LINUX="crashkernel=auto rd.lvm.lv=rhel/root rd.lvm.lv=rhel/swap rhgb quiet ro"Recreate the GRUB2 configuration file:
# grub2-mkconfig -o /boot/grub2/grub.cfgIf you need to add files and directories to be mounted with write permissions in the tmpfs file system, create a text file in the /etc/rwtab.d/ directory and put the configuration there. For example, to mount /etc/example/file with write permissions, add this line to the /etc/rwtab.d/example file:
files /etc/example/fileChanges made to files and directories in tmpfs do not persist across boots.
Reboot the system.
19.2.5.2. Remounting root Instantly
If root (/) was mounted with read-only permissions on system boot, you can remount it with write permissions:
# mount -o remount,rw /This can be particularly useful when / is incorrectly mounted with read-only permissions.
To remount / with read-only permissions again, run:
# mount -o remount,ro /19.2.5.3. Files and Directories That Retain Write Permissions
For the system to function properly, some files and directories need to retain write permissions. With root in read-only mode, they are mounted in RAM in the tmpfs temporary file system. The default set of such files and directories is read from the /etc/rwtab file, which contains:
dirs /var/cache/man dirs /var/gdm [output truncated] empty /tmp empty /var/cache/foomatic [output truncated] files /etc/adjtime files /etc/ntp.conf [output truncated]Entries in the /etc/rwtab file follow this format:
how the file or directory is copied to tmpfs path to the file or directoryA file or directory can be copied to tmpfs in the following three ways:
empty path: An empty path is copied to tmpfs. Example: empty /tmp
dirs path: A directory tree is copied to tmpfs, empty. Example: dirs /var/run
files path: A file or a directory tree is copied to tmpfs intact. Example: files /etc/resolv.conf
The same format applies when adding custom paths to /etc/rwtab.d/.
19.3. Unmounting a File System
To detach a previously mounted file system, use either of the following variants of the umount command:
$ umount directory $ umount deviceWhen a file system is in use (for example, when a process is reading a file on this file system, or when it is used by the kernel), running the umount command fails with an error. To determine which processes are accessing the file system, use the fuser command in the following form:
$ fuser -m directoryFor example, to list the processes that are accessing a file system mounted to the /media/cdrom/ directory:
$ fuser -m /media/cdrom /media/cdrom: 1793 2013 2022 2435 10532c 10672cExample 19.9. Unmounting a CD
To unmount a CD that was previously mounted to the /media/cdrom/ directory, use the following command:
$ umount /media/cdrom19.4. mount Command References
The following resources provide an in-depth documentation on the subject.
Manual Page Documentation
man 8 mount: The manual page for the mount command that provides a full documentation on its usage.
man 8 umount: The manual page for the umount command that provides a full documentation on its usage.
man 8 findmnt: The manual page for the findmnt command that provides a full documentation on its usage.
man 5 fstab: The manual page providing a thorough description of the /etc/fstab file format.
Useful Websites
Shared subtrees — An LWN article covering the concept of shared subtrees.
Chapter 20. The volume_key Function
The volume_key function provides two tools, libvolume_key and volume_key. libvolume_key is a library for manipulating storage volume encryption keys and storing them separately from volumes. volume_key is an associated command line tool used to extract keys and passphrases in order to restore access to an encrypted hard drive.
This is useful for when the primary user forgets their keys and passwords, after an employee leaves abruptly, or in order to extract data after a hardware or software failure corrupts the header of the encrypted volume. In a corporate setting, the IT help desk can use volume_key to back up the encryption keys before handing over the computer to the end user.
Currently, volume_key only supports the LUKS volume encryption format.
20.1. volume_key Commands
The format for volume_key is:
volume_key [OPTION]... OPERANDThe operands and mode of operation for volume_key are determined by specifying one of the following options:
--save
This command expects the operand volume [packet]. If a packet is provided then volume_key will extract the keys and passphrases from it. If packet is not provided, then volume_key will extract the keys and passphrases from the volume, prompting the user where necessary. These keys and passphrases will then be stored in one or more output packets.
--restoreThis command expects the operands volume packet. It then opens the volume and uses the keys and passphrases in the packet to make the volume accessible again, prompting the user where necessary, such as allowing the user to enter a new passphrase, for example.
--setup-volumeThis command expects the operands volume packet name. It then opens the volume and uses the keys and passphrases in the packet to set up the volume for use of the decrypted data as name.
Name is the name of a dm-crypt volume. This operation makes the decrypted volume available as /dev/mapper/name.
This operation does not permanently alter the volume by adding a new passphrase, for example. The user can access and modify the decrypted volume, modifying volume in the process.
--reencrypt, --secrets, and --dumpThese three commands perform similar functions with varying output methods. They each require the operand packet, and each opens the packet, decrypting it where necessary. --reencrypt then stores the information in one or more new output packets. --secrets outputs the keys and passphrases contained in the packet. --dump outputs the content of the packet, though the keys and passphrases are not output by default. This can be changed by appending --with-secrets to the command. It is also possible to only dump the unencrypted parts of the packet, if any, by using the --unencrypted command. This does not require any passphrase or private key access.
Each of these can be appended with the following options:
-o, --output packet
This command writes the default key or passphrase to the packet. The default key or passphrase depends on the volume format. Ensure it is one that is unlikely to expire, and will allow --restore to restore access to the volume.
--output-format formatThis command uses the specified format for all output packets. Currently, format can be one of the following:
asymmetric: uses CMS to encrypt the whole packet, and requires a certificate
asymmetric_wrap_secret_only: wraps only the secret, or keys and passphrases, and requires a certificate
passphrase: uses GPG to encrypt the whole packet, and requires a passphrase
This command generates a random alphanumeric passphrase, adds it to the volume (without affecting other passphrases), and then stores this random passphrase into the packet.
20.2. Using volume_key as an Individual User
As an individual user, volume_key can be used to save encryption keys by using the following procedure.
For all examples in this file, /path/to/volume is a LUKS device, not the plaintext device contained within. blkid -s type /path/to/volume should report type="crypto_LUKS".
Procedure 20.1. Using volume_key Stand-alone
Run:
volume_key --save /path/to/volume -o escrow-packetA prompt will then appear requiring an escrow packet passphrase to protect the key.
Save the generated escrow-packet file, ensuring that the passphrase is not forgotten.
If the volume passphrase is forgotten, use the saved escrow packet to restore access to the data.
Procedure 20.2. Restore Access to Data with Escrow Packet
Boot the system in an environment where volume_key can be run and the escrow packet is available (a rescue mode, for example).
Run:
volume_key --restore /path/to/volume escrow-packetA prompt will appear for the escrow packet passphrase that was used when creating the escrow packet, and for the new passphrase for the volume.
Mount the volume using the chosen passphrase.
To free up the passphrase slot in the LUKS header of the encrypted volume, remove the old, forgotten passphrase by using the command cryptsetup luksKillSlot.
20.3. Using volume_key in a Larger Organization
In a larger organization, using a single password known by every system administrator and keeping track of a separate password for each system is impractical and a security risk. To counter this, volume_key can use asymmetric cryptography to minimize the number of people who know the password required to access encrypted data on any computer.
This section will cover the procedures required for preparation before saving encryption keys, how to save encryption keys, restoring access to a volume, and setting up emergency passphrases.
20.3.1. Preparation for Saving Encryption Keys
In order to begin saving encryption keys, some preparation is required.
Procedure 20.3. Preparation
Create an X509 certificate/private pair.
Designate trusted users who are trusted not to compromise the private key. These users will be able to decrypt the escrow packets.
Choose which systems will be used to decrypt the escrow packets. On these systems, set up an NSS database that contains the private key.
If the private key was not created in an NSS database, follow these steps:
Store the certificate and private key in an PKCS#12 file.
Run:
certutil -d /the/nss/directory -NAt this point it is possible to choose an NSS database password. Each NSS database can have a different password so the designated users do not need to share a single password if a separate NSS database is used by each user.
Run:
pk12util -d /the/nss/directory -i the-pkcs12-file
Distribute the certificate to anyone installing systems or saving keys on existing systems.
For saved private keys, prepare storage that allows them to be looked up by machine and volume. For example, this can be a simple directory with one subdirectory per machine, or a database used for other system management tasks as well.
20.3.2. Saving Encryption Keys
For all examples in this file, /path/to/volume is a LUKS device, not the plaintext device contained within; blkid -s type /path/to/volume should report type="crypto_LUKS".
Procedure 20.4. Saving Encryption Keys
Run:
volume_key --save /path/to/volume -c /path/to/cert escrow-packetSave the generated escrow-packet file in the prepared storage, associating it with the system and the volume.
These steps can be performed manually, or scripted as part of system installation.
20.3.3. Restoring Access to a Volume
Procedure 20.5. Restoring Access to a Volume
Get the escrow packet for the volume from the packet storage and send it to one of the designated users for decryption.
The designated user runs:
volume_key --reencrypt -d /the/nss/directory escrow-packet-in -o escrow-packet-outAfter providing the NSS database password, the designated user chooses a passphrase for encrypting escrow-packet-out. This passphrase can be different every time and only protects the encryption keys while they are moved from the designated user to the target system.
Obtain the escrow-packet-out file and the passphrase from the designated user.
Boot the target system in an environment that can run volume_key and have the escrow-packet-out file available, such as in a rescue mode.
Run:
volume_key --restore /path/to/volume escrow-packet-outA prompt will appear for the packet passphrase chosen by the designated user, and for a new passphrase for the volume.
Mount the volume using the chosen volume passphrase.
It is possible to remove the old passphrase that was forgotten by using cryptsetup luksKillSlot, for example, to free up the passphrase slot in the LUKS header of the encrypted volume. This is done with the command cryptsetup luksKillSlot device key-slot. For more information and examples see cryptsetup --help.
20.3.4. Setting up Emergency Passphrases
In some circumstances (such as traveling for business) it is impractical for system administrators to work directly with the affected systems, but users still need access to their data. In this case, volume_key can work with passphrases as well as encryption keys.
During the system installation, run:
volume_key --save /path/to/volume -c /path/to/ert --create-random-passphrase passphrase-packetThis generates a random passphrase, adds it to the specified volume, and stores it to passphrase-packet. It is also possible to combine the --create-random-passphrase and -o options to generate both packets at the same time.
If a user forgets the password, the designated user runs:
volume_key --secrets -d /your/nss/directory passphrase-packetThis shows the random passphrase. Give this passphrase to the end user.
20.4. volume_key References
More information on volume_key can be found:
in the readme file located at /usr/share/doc/volume_key-*/README
on volume_key's manpage using man volume_key
Chapter 21. Solid-State Disk Deployment Guidelines
Solid-state disks (SSD) are storage devices that use NAND flash chips to persistently store data. This sets them apart from previous generations of disks, which store data in rotating, magnetic platters. In an SSD, the access time for data across the full Logical Block Address (LBA) range is constant; whereas with older disks that use rotating media, access patterns that span large address ranges incur seek costs. As such, SSD devices have better latency and throughput.
Performance degrades as the number of used blocks approaches the disk capacity. The degree of performance impact varies greatly by vendor. However, all devices experience some degradation.
To address the degradation issue, the host system (for example, the Linux kernel) may use discard requests to inform the storage that a given range of blocks is no longer in use. An SSD can use this information to free up space internally, using the free blocks for wear-leveling. Discards will only be issued if the storage advertises support in terms of its storage protocol (be it ATA or SCSI). Discard requests are issued to the storage using the negotiated discard command specific to the storage protocol (TRIM command for ATA, and WRITE SAME with UNMAP set, or UNMAP command for SCSI).
Enabling discard support is most useful when the following points are true:
Free space is still available on the file system.
Most logical blocks on the underlying storage device have already been written to.
Not all solid-state devices in the market have discard support. To determine if your solid-state device has discard support, check for /sys/block/sda/queue/discard_granularity, which is the size of internal allocation unit of device.
Deployment Considerations
Because of the internal layout and operation of SSDs, it is best to partition devices on an internal erase block boundary. Partitioning utilities in Red Hat Enterprise Linux 7 chooses sane defaults if the SSD exports topology information. However, if the device does not export topology information, Red Hat recommends that the first partition should be created at a 1MB boundary.
SSD has various types of TRIM mechanism depending on the vendors choice. The early versions of disks improved the performance by compromising possible data leakage after the read command.
Following are the types of TRIM mechanism:
Non-deterministic TRIM
Deterministic TRIM (DRAT)
Deterministic Read Zero after TRIM (RZAT)
The first two types of TRIM mechanism can cause data leakage as the read command to the LBA after a TRIM returns different or same data. RZAT returns zero after the read command and Red Hat recommends this TRIM mechanism to avoid data leakage. It is affected only in SSD. Choose the disk which supports RZAT mechanism.
Type of TRIM mechanism used depends on hardware implementation. To find the type of TRIM mechanism on ATA, use the hdparm command. See the following example to find the type of TRIM mechanism:
# hdparm -I /dev/sda | grep TRIM Data Set Management TRIM supported (limit 8 block) Deterministic read data after TRIMFor more information, see man hdparm.
The Logical Volume Manager (LVM), the device-mapper (DM) targets, and MD (software raid) targets that LVM uses support discards. The only DM targets that do not support discards are dm-snapshot, dm-crypt, and dm-raid45. Discard support for the dm-mirror was added in Red Hat Enterprise Linux 6.1 and as of 7.0 MD supports discards.
Using RAID level 5 over SSD results in low performance if SSDs do not handle discard correctly. You can set discard in the raid456.conf file, or in the GRUB2 configuration. For instructions, see the following procedures.
Procedure 21.1. Setting discard in raid456.conf
The devices_handle_discard_safely module parameter is set in the raid456 module. To enable discard in the raid456.conf file:
Verify that your hardware supports discards:
# cat /sys/block/disk-name/queue/discard_zeroes_dataIf the returned value is 1, discards are supported. If the command returns 0, the RAID code has to zero the disk out, which takes more time.
Create the /etc/modprobe.d/raid456.conf file, and include the following line:
options raid456 devices_handle_discard_safely=YUse the dracut -f command to rebuild the initial ramdisk (initrd).
Reboot the system for the changes to take effect.
Procedure 21.2. Setting discard in the GRUB2 Configuration
The devices_handle_discard_safely module parameter is set in the raid456 module. To enable discard in the GRUB2 configuration:
Verify that your hardware supports discards:
# cat /sys/block/disk-name/queue/discard_zeroes_dataIf the returned value is 1, discards are supported. If the command returns 0, the RAID code has to zero the disk out, which takes more time.
Add the following line to the /etc/default/grub file:
raid456.devices_handle_discard_safely=YThe location of the GRUB2 configuration file is different on systems with the BIOS firmware and on systems with UEFI. Use one of the following commands to recreate the GRUB2 configuration file.
On a system with the BIOS firmware, use:
# grub2-mkconfig -o /boot/grub2/grub.cfgOn a system with the UEFI firmware, use:
# grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg
Reboot the system for the changes to take effect.
In Red Hat Enterprise Linux 7, discard is fully supported by the ext4 and XFS file systems only.
In Red Hat Enterprise Linux 6.3 and earlier, only the ext4 file system fully supports discard. Starting with Red Hat Enterprise Linux 6.4, both ext4 and XFS file systems fully support discard. To enable discard commands on a device, use the discard option of the mount command. For example, to mount /dev/sda2 to /mnt with discard enabled, use:
# mount -t ext4 -o discard /dev/sda2 /mntBy default, ext4 does not issue the discard command to, primarily, avoid problems on devices which might not properly implement discard. The Linux swap code issues discard commands to discard-enabled devices, and there is no option to control this behavior.
Performance Tuning Considerations
For information on performance tuning considerations regarding solid-state disks, see the Solid-State Disks section in the Red Hat Enterprise Linux 7 Performance Tuning Guide.
Chapter 22. Write Barriers
A write barrier is a kernel mechanism used to ensure that file system metadata is correctly written and ordered on persistent storage, even when storage devices with volatile write caches lose power. File systems with write barriers enabled ensures that data transmitted via fsync() is persistent throughout a power loss.
Enabling write barriers incurs a substantial performance penalty for some applications. Specifically, applications that use fsync() heavily or create and delete many small files will likely run much slower.
22.1. Importance of Write Barriers
File systems safely update metadata, ensuring consistency. Journalled file systems bundle metadata updates into transactions and send them to persistent storage in the following manner:
The file system sends the body of the transaction to the storage device.
The file system sends a commit block.
If the transaction and its corresponding commit block are written to disk, the file system assumes that the transaction will survive any power failure.
However, file system integrity during power failure becomes more complex for storage devices with extra caches. Storage target devices like local S-ATA or SAS drives may have write caches ranging from 32MB to 64MB in size (with modern drives). Hardware RAID controllers often contain internal write caches. Further, high end arrays, like those from NetApp, IBM, Hitachi and EMC (among others), also have large caches.
Storage devices with write caches report I/O as "complete" when the data is in cache; if the cache loses power, it loses its data as well. Worse, as the cache de-stages to persistent storage, it may change the original metadata ordering. When this occurs, the commit block may be present on disk without having the complete, associated transaction in place. As a result, the journal may replay these uninitialized transaction blocks into the file system during post-power-loss recovery; this will cause data inconsistency and corruption.
How Write Barriers Work
Write barriers are implemented in the Linux kernel via storage write cache flushes before and after the I/O, which is order-critical. After the transaction is written, the storage cache is flushed, the commit block is written, and the cache is flushed again. This ensures that:
The disk contains all the data.
No re-ordering has occurred.
With barriers enabled, an fsync() call also issues a storage cache flush. This guarantees that file data is persistent on disk even if power loss occurs shortly after fsync() returns.
22.2. Enabling and Disabling Write Barriers
To mitigate the risk of data corruption during power loss, some storage devices use battery-backed write caches. Generally, high-end arrays and some hardware controllers use battery-backed write caches. However, because the cache's volatility is not visible to the kernel, Red Hat Enterprise Linux 7 enables write barriers by default on all supported journaling file systems.
Write caches are designed to increase I/O performance. However, enabling write barriers means constantly flushing these caches, which can significantly reduce performance.
For devices with non-volatile, battery-backed write caches and those with write-caching disabled, you can safely disable write barriers at mount time using the -o nobarrier option for mount. However, some devices do not support write barriers; such devices log an error message to /var/log/messages. For more information, see Table 22.1, “Write Barrier Error Messages per File System”.
Table 22.1. Write Barrier Error Messages per File System
ext3/ext4 | JBD: barrier-based sync failed on device - disabling barriers |
XFS | Filesystem device - Disabling barriers, trial barrier write failed |
btrfs | btrfs: disabling barriers on dev device |
22.3. Write Barrier Considerations
Some system configurations do not need write barriers to protect data. In most cases, other methods are preferable to write barriers, since enabling write barriers causes a significant performance penalty.
Disabling Write Caches
One way to alternatively avoid data integrity issues is to ensure that no write caches lose data on power failures. When possible, the best way to configure this is to disable the write cache. On a simple server or desktop with one or more SATA drives (off a local SATA controller Intel AHCI part), you can disable the write cache on the target SATA drives with the following command:
# hdparm -W0 /device/Battery-Backed Write Caches
Write barriers are also unnecessary whenever the system uses hardware RAID controllers with battery-backed write cache. If the system is equipped with such controllers and if its component drives have write caches disabled, the controller acts as a write-through cache; this informs the kernel that the write cache data survives a power loss.
Most controllers use vendor-specific tools to query and manipulate target drives. For example, the LSI Megaraid SAS controller uses a battery-backed write cache; this type of controller requires the MegaCli64 tool to manage target drives. To show the state of all back-end drives for LSI Megaraid SAS, use:
# MegaCli64 -LDGetProp -DskCache -LAll -aALLTo disable the write cache of all back-end drives for LSI Megaraid SAS, use:
# MegaCli64 -LDSetProp -DisDskCache -Lall -aALLHardware RAID cards recharge their batteries while the system is operational. If a system is powered off for an extended period of time, the batteries will lose their charge, leaving stored data vulnerable during a power failure.
High-End Arrays
High-end arrays have various ways of protecting data in the event of a power failure. As such, there is no need to verify the state of the internal drives in external RAID storage.
NFS
NFS clients do not need to enable write barriers, since data integrity is handled by the NFS server side. As such, NFS servers should be configured to ensure data persistence throughout a power loss (whether through write barriers or other means).
Chapter 23. Storage I/O Alignment and Size
Recent enhancements to the SCSI and ATA standards allow storage devices to indicate their preferred (and in some cases, required) I/O alignment and I/O size. This information is particularly useful with newer disk drives that increase the physical sector size from 512 bytes to 4k bytes. This information may also be beneficial for RAID devices, where the chunk size and stripe size may impact performance.
The Linux I/O stack has been enhanced to process vendor-provided I/O alignment and I/O size information, allowing storage management tools (parted, lvm, mkfs.*, and the like) to optimize data placement and access. If a legacy device does not export I/O alignment and size data, then storage management tools in Red Hat Enterprise Linux 7 will conservatively align I/O on a 4k (or larger power of 2) boundary. This will ensure that 4k-sector devices operate correctly even if they do not indicate any required/preferred I/O alignment and size.
For information on determining the information that the operating system obtained from the device, see the Section 23.2, “Userspace Access”. This data is subsequently used by the storage management tools to determine data placement.
The IO scheduler has changed for Red Hat Enterprise Linux 7. Default IO Scheduler is now Deadline, except for SATA drives. CFQ is the default IO scheduler for SATA drives. For faster storage, Deadline outperforms CFQ and when it is used there is a performance increase without the need of special tuning.
If default is not right for some disks (for example, SAS rotational disks), then change the IO scheduler to CFQ. This instance will depend on the workload.
23.1. Parameters for Storage Access
The operating system uses the following information to determine I/O alignment and size:
physical_block_size
Smallest internal unit on which the device can operate
logical_block_sizeUsed externally to address a location on the device
alignment_offsetThe number of bytes that the beginning of the Linux block device (partition/MD/LVM device) is offset from the underlying physical alignment
minimum_io_sizeThe device’s preferred minimum unit for random I/O
optimal_io_sizeThe device’s preferred unit for streaming I/O
For example, certain 4K sector devices may use a 4K physical_block_size internally but expose a more granular 512-byte logical_block_size to Linux. This discrepancy introduces potential for misaligned I/O. To address this, the Red Hat Enterprise Linux 7 I/O stack will attempt to start all data areas on a naturally-aligned boundary (physical_block_size) by making sure it accounts for any alignment_offset if the beginning of the block device is offset from the underlying physical alignment.
Storage vendors can also supply I/O hints about the preferred minimum unit for random I/O (minimum_io_size) and streaming I/O (optimal_io_size) of a device. For example, minimum_io_size and optimal_io_size may correspond to a RAID device's chunk size and stripe size respectively.
23.2. Userspace Access
Always take care to use properly aligned and sized I/O. This is especially important for Direct I/O access. Direct I/O should be aligned on a logical_block_size boundary, and in multiples of the logical_block_size.
With native 4K devices (i.e. logical_block_size is 4K) it is now critical that applications perform direct I/O in multiples of the device's logical_block_size. This means that applications will fail with native 4k devices that perform 512-byte aligned I/O rather than 4k-aligned I/O.
To avoid this, an application should consult the I/O parameters of a device to ensure it is using the proper I/O alignment and size. As mentioned earlier, I/O parameters are exposed through the both sysfs and block device ioctl interfaces.
For more information, see man libblkid. This man page is provided by the libblkid-devel package.
sysfs Interface
/sys/block/disk/alignment_offset
or
/sys/block/disk/partition/alignment_offset
The file location depends on whether the disk is a physical disk (be that a local disk, local RAID, or a multipath LUN) or a virtual disk. The first file location is applicable to physical disks while the second file location is applicable to virtual disks. The reason for this is because virtio-blk will always report an alignment value for the partition. Physical disks may or may not report an alignment value.
/sys/block/disk/queue/physical_block_size
/sys/block/disk/queue/logical_block_size
/sys/block/disk/queue/minimum_io_size
/sys/block/disk/queue/optimal_io_size
The kernel will still export these sysfs attributes for "legacy" devices that do not provide I/O parameters information, for example:
Example 23.1. sysfs Interface
alignment_offset: 0 physical_block_size: 512 logical_block_size: 512 minimum_io_size: 512 optimal_io_size: 0
Block Device ioctls
BLKALIGNOFF: alignment_offset
BLKPBSZGET: physical_block_size
BLKSSZGET: logical_block_size
BLKIOMIN: minimum_io_size
BLKIOOPT: optimal_io_size
23.3. I/O Standards
This section describes I/O standards used by ATA and SCSI devices.
ATA
ATA devices must report appropriate information via the IDENTIFY DEVICE command. ATA devices only report I/O parameters for physical_block_size, logical_block_size, and alignment_offset. The additional I/O hints are outside the scope of the ATA Command Set.
SCSI
I/O parameters support in Red Hat Enterprise Linux 7 requires at least version 3 of the SCSI Primary Commands (SPC-3) protocol. The kernel will only send an extended inquiry (which gains access to the BLOCK LIMITS VPD page) and READ CAPACITY(16) command to devices which claim compliance with SPC-3.
The READ CAPACITY(16) command provides the block sizes and alignment offset:
LOGICAL BLOCK LENGTH IN BYTES is used to derive /sys/block/disk/queue/physical_block_size
LOGICAL BLOCKS PER PHYSICAL BLOCK EXPONENT is used to derive /sys/block/disk/queue/logical_block_size
LOWEST ALIGNED LOGICAL BLOCK ADDRESS is used to derive:
/sys/block/disk/alignment_offset
/sys/block/disk/partition/alignment_offset
The BLOCK LIMITS VPD page (0xb0) provides the I/O hints. It also uses OPTIMAL TRANSFER LENGTH GRANULARITY and OPTIMAL TRANSFER LENGTH to derive:
/sys/block/disk/queue/minimum_io_size
/sys/block/disk/queue/optimal_io_size
The sg3_utils package provides the sg_inq utility, which can be used to access the BLOCK LIMITS VPD page. To do so, run:
# sg_inq -p 0xb0 disk23.4. Stacking I/O Parameters
All layers of the Linux I/O stack have been engineered to propagate the various I/O parameters up the stack. When a layer consumes an attribute or aggregates many devices, the layer must expose appropriate I/O parameters so that upper-layer devices or tools will have an accurate view of the storage as it transformed. Some practical examples are:
Only one layer in the I/O stack should adjust for a non-zero alignment_offset; once a layer adjusts accordingly, it will export a device with an alignment_offset of zero.
A striped Device Mapper (DM) device created with LVM must export a minimum_io_size and optimal_io_size relative to the stripe count (number of disks) and user-provided chunk size.
In Red Hat Enterprise Linux 7, Device Mapper and Software Raid (MD) device drivers can be used to arbitrarily combine devices with different I/O parameters. The kernel's block layer will attempt to reasonably combine the I/O parameters of the individual devices. The kernel will not prevent combining heterogeneous devices; however, be aware of the risks associated with doing so.
For instance, a 512-byte device and a 4K device may be combined into a single logical DM device, which would have a logical_block_size of 4K. File systems layered on such a hybrid device assume that 4K will be written atomically, but in reality it will span 8 logical block addresses when issued to the 512-byte device. Using a 4K logical_block_size for the higher-level DM device increases potential for a partial write to the 512-byte device if there is a system crash.
If combining the I/O parameters of multiple devices results in a conflict, the block layer may issue a warning that the device is susceptible to partial writes and/or is misaligned.
23.5. Logical Volume Manager
LVM provides userspace tools that are used to manage the kernel's DM devices. LVM will shift the start of the data area (that a given DM device will use) to account for a non-zero alignment_offset associated with any device managed by LVM. This means logical volumes will be properly aligned (alignment_offset=0).
By default, LVM will adjust for any alignment_offset, but this behavior can be disabled by setting data_alignment_offset_detection to 0 in /etc/lvm/lvm.conf. Disabling this is not recommended.
LVM will also detect the I/O hints for a device. The start of a device's data area will be a multiple of the minimum_io_size or optimal_io_size exposed in sysfs. LVM will use the minimum_io_size if optimal_io_size is undefined (i.e. 0).
By default, LVM will automatically determine these I/O hints, but this behavior can be disabled by setting data_alignment_detection to 0 in /etc/lvm/lvm.conf. Disabling this is not recommended.
Chapter 24. Setting up a Remote Diskless System
To set up a basic remote diskless system booted over PXE, you need the following packages:
tftp-server
xinetd
dhcp
syslinux
dracut-network
After installing the dracut-network package, add the following line to /etc/dracut.conf:
add_dracutmodules+="nfs"
Remote diskless system booting requires both a tftp service (provided by tftp-server) and a DHCP service (provided by dhcp). The tftp service is used to retrieve kernel image and initrd over the network via the PXE loader.
SELinux is only supported over NFSv4.2. To use SELinux, NFS must be explicitly enabled in /etc/sysconfig/nfs by adding the line:
RPCNFSDARGS="-V 4.2"
Then, in /var/lib/tftpboot/pxelinux.cfg/default, change root=nfs:server-ip:/exported/root/directory to root=nfs:server-ip:/exported/root/directory,vers=4.2.
Finally, reboot the NFS server.
The following sections outline the necessary procedures for deploying remote diskless systems in a network environment.
Some RPM packages have started using file capabilities (such as setcap and getcap). However, NFS does not currently support these so attempting to install or update any packages that use file capabilities will fail.
24.1. Configuring a tftp Service for Diskless Clients
Prerequisites
Procedure
To configure tftp, perform the following steps:
Procedure 24.1. To Configure tftp
Enable PXE booting over the network:
# systemctl enable --now tftpThe tftp root directory (chroot) is located in /var/lib/tftpboot. Copy /usr/share/syslinux/pxelinux.0 to /var/lib/tftpboot/:
# cp /usr/share/syslinux/pxelinux.0 /var/lib/tftpboot/Create a pxelinux.cfg directory inside the tftp root directory:
# mkdir -p /var/lib/tftpboot/pxelinux.cfg/Configure firewall rules to allow tftp traffic.
As tftp supports TCP wrappers, you can configure host access to tftp in the /etc/hosts.allow configuration file. For more information on configuring TCP wrappers and the /etc/hosts.allow configuration file, see the Red Hat Enterprise Linux 7 Security Guide. The hosts_access(5) also provides information about /etc/hosts.allow.
Next Steps
24.2. Configuring DHCP for Diskless Clients
Prerequisites
Procedure
After configuring a tftp server, you need to set up a DHCP service on the same host machine. For instructions on setting up a DHCP server, see the Configuring a DHCP Server.
Enable PXE booting on the DHCP server by adding the following configuration to /etc/dhcp/dhcpd.conf:
allow booting; allow bootp; class "pxeclients" { match if substring(option vendor-class-identifier, 0, 9) = "PXEClient"; next-server server-ip; filename "pxelinux.0"; }Replace server-ip with the IP address of the host machine on which the tftp and DHCP services reside.
When libvirt virtual machines are used as the diskless client, libvirt provides the DHCP service and the stand alone DHCP server is not used. In this situation, network booting must be enabled with the bootp file='filename' option in the libvirt network configuration, virsh net-edit.
Next Steps
24.3. Configuring an Exported File System for Diskless Clients
Prerequisites
Procedure
The root directory of the exported file system (used by diskless clients in the network) is shared via NFS. Configure the NFS service to export the root directory by adding it to /etc/exports. For instructions on how to do so, see the Section 8.6.1, “The /etc/exports Configuration File”.
To accommodate completely diskless clients, the root directory should contain a complete Red Hat Enterprise Linux installation. You can either clone an existing installation or install a new base system:
To synchronize with a running system, use the rsync utility:
# rsync -a -e ssh --exclude='/proc/*' --exclude='/sys/*' \ hostname.com:/exported-root-directoryReplace hostname.com with the hostname of the running system with which to synchronize via rsync.
Replace exported-root-directory with the path to the exported file system.
To install Red Hat Enterprise Linux to the exported location, use the yum utility with the --installroot option:
# yum install @Base kernel dracut-network nfs-utils \ --installroot=exported-root-directory --releasever=/
The file system to be exported still needs to be configured further before it can be used by diskless clients. To do this, perform the following procedure:
Procedure 24.2. Configure File System
Select the kernel that diskless clients should use (vmlinuz-kernel-version) and copy it to the tftp boot directory:
# cp /boot/vmlinuz-kernel-version /var/lib/tftpboot/Create the initrd (that is, initramfs-kernel-version.img) with NFS support:
# dracut --add nfs initramfs-kernel-version.img kernel-versionChange the initrd's file permissions to 644 using the following command:
# chmod 644 initramfs-kernel-version.imgIf the initrd's file permissions are not changed, the pxelinux.0 boot loader will fail with a "file not found" error.
Copy the resulting initramfs-kernel-version.img into the tftp boot directory as well.
Edit the default boot configuration to use the initrd and kernel in the /var/lib/tftpboot/ directory. This configuration should instruct the diskless client's root to mount the exported file system (/exported/root/directory) as read-write. Add the following configuration in the /var/lib/tftpboot/pxelinux.cfg/default file:
default rhel7 label rhel7 kernel vmlinuz-kernel-version append initrd=initramfs-kernel-version.img root=nfs:server-ip:/exported/root/directory rwReplace server-ip with the IP address of the host machine on which the tftp and DHCP services reside.
The NFS share is now ready for exporting to diskless clients. These clients can boot over the network via PXE.
Chapter 25. Online Storage Management
It is often desirable to add, remove or re-size storage devices while the operating system is running, and without rebooting. This chapter outlines the procedures that may be used to reconfigure storage devices on Red Hat Enterprise Linux 7 host systems while the system is running. It covers iSCSI and Fibre Channel storage interconnects; other interconnect types may be added it the future.
This chapter focuses on adding, removing, modifying, and monitoring storage devices. It does not discuss the Fibre Channel or iSCSI protocols in detail. For more information about these protocols, refer to other documentation.
This chapter makes reference to various sysfs objects. Red Hat advises that the sysfs object names and directory structure are subject to change in major Red Hat Enterprise Linux releases. This is because the upstream Linux kernel does not provide a stable internal API. For guidelines on how to reference sysfs objects in a transportable way, refer to the document /usr/share/doc/kernel-doc-version/Documentation/sysfs-rules.txt in the kernel source tree for guidelines.
Online storage reconfiguration must be done carefully. System failures or interruptions during the process can lead to unexpected results. Red Hat advises that you reduce system load to the maximum extent possible during the change operations. This will reduce the chance of I/O errors, out-of-memory errors, or similar errors occurring in the midst of a configuration change. The following sections provide more specific guidelines regarding this.
In addition, Red Hat recommends that you back up all data before reconfiguring online storage.
25.1. Target Setup
Red Hat Enterprise Linux 7 uses the targetcli shell as a front end for viewing, editing, and saving the configuration of the Linux-IO Target without the need to manipulate the kernel target's configuration files directly. The targetcli tool is a command-line interface that allows an administrator to export local storage resources, which are backed by either files, volumes, local SCSI devices, or RAM disks, to remote systems. The targetcli tool has a tree-based layout, includes built-in tab completion, and provides full auto-complete support and inline documentation.
The hierarchy of targetcli does not always match the kernel interface exactly because targetcli is simplified where possible.
To ensure that the changes made in targetcli are persistent, start and enable the target service:
# systemctl start target # systemctl enable target25.1.1. Installing and Running targetcli
To install targetcli, use:
# yum install targetcliStart the target service:
# systemctl start targetConfigure target to start at boot time:
# systemctl enable targetOpen port 3260 in the firewall and reload the firewall configuration:
# firewall-cmd --permanent --add-port=3260/tcp Success # firewall-cmd --reload SuccessUse the targetcli command, and then use the ls command for the layout of the tree interface:
# targetcli : /> ls o- /........................................[...] o- backstores.............................[...] | o- block.................[Storage Objects: 0] | o- fileio................[Storage Objects: 0] | o- pscsi.................[Storage Objects: 0] | o- ramdisk...............[Storage Ojbects: 0] o- iscsi...........................[Targets: 0] o- loopback........................[Targets: 0]In Red Hat Enterprise Linux 7.0, using the targetcli command from Bash, for example, targetcli iscsi/ create, does not work and does not return an error. Starting with Red Hat Enterprise Linux 7.1, an error status code is provided to make using targetcli with shell scripts more useful.
25.1.2. Creating a Backstore
Backstores enable support for different methods of storing an exported LUN's data on the local machine. Creating a storage object defines the resources the backstore uses.
In Red Hat Enterprise Linux 6, the term 'backing-store' is used to refer to the mappings created. However, to avoid confusion between the various ways 'backstores' can be used, in Red Hat Enterprise Linux 7 the term 'storage objects' refers to the mappings created and 'backstores' is used to describe the different types of backing devices.
The backstore devices that LIO supports are:
FILEIO (Linux file-backed storage)
FILEIO storage objects can support either write_back or write_thru operation. The write_back enables the local file system cache. This improves performance but increases the risk of data loss. It is recommended to use write_back=false to disable write_back in favor of write_thru.
To create a fileio storage object, run the command /backstores/fileio create file_name file_location file_size write_back=false. For example:
/> /backstores/fileio create file1 /tmp/disk1.img 200M write_back=false Created fileio file1 with size 209715200 BLOCK (Linux BLOCK devices)The block driver allows the use of any block device that appears in the /sys/block to be used with LIO. This includes physical devices (for example, HDDs, SSDs, CDs, DVDs) and logical devices (for example, software or hardware RAID volumes, or LVM volumes).
BLOCK backstores usually provide the best performance.
To create a BLOCK backstore using any block device, use the following command:
# fdisk /dev/vdb Welcome to fdisk (util-linux 2.23.2). Changes will remain in memory only, until you decide to write them. Be careful before using the write command. Device does not contain a recognized partition table Building a new DOS disklabel with disk identifier 0x39dc48fb. Command (m for help): n Partition type: p primary (0 primary, 0 extended, 4 free) e extended Select (default p): *Enter* Using default response p Partition number (1-4, default 1): *Enter* First sector (2048-2097151, default 2048): *Enter* Using default value 2048 Last sector, +sectors or +size{K,M,G} (2048-2097151, default 2097151): +250M Partition 1 of type Linux and of size 250 MiB is set Command (m for help): w The partition table has been altered! Calling ioctl() to re-read partition table. Syncing disks. /> /backstores/block create name=block_backend dev=/dev/vdb Generating a wwn serial. Created block storage object block_backend using /dev/vdb.You can also create a BLOCK backstore on a logical volume.
PSCSI (Linux pass-through SCSI devices)Any storage object that supports direct pass-through of SCSI commands without SCSI emulation, and with an underlying SCSI device that appears with lsscsi in /proc/scsi/scsi (such as a SAS hard drive) can be configured as a backstore. SCSI-3 and higher is supported with this subsystem.
PSCSI should only be used by advanced users. Advanced SCSI commands such as for Aysmmetric Logical Unit Assignment (ALUAs) or Persistent Reservations (for example, those used by VMware ESX, and vSphere) are usually not implemented in the device firmware and can cause malfunctions or crashes. When in doubt, use BLOCK for production setups instead.
To create a PSCSI backstore for a physical SCSI device, a TYPE_ROM device using /dev/sr0 in this example, use:
/> backstores/pscsi/ create name=pscsi_backend dev=/dev/sr0 Generating a wwn serial. Created pscsi storage object pscsi_backend using /dev/sr0 Memory Copy RAM disk (Linux RAMDISK_MCP)Memory Copy RAM disks (ramdisk) provide RAM disks with full SCSI emulation and separate memory mappings using memory copy for initiators. This provides capability for multi-sessions and is particularly useful for fast, volatile mass storage for production purposes.
To create a 1GB RAM disk backstore, use the following command:
/> backstores/ramdisk/ create name=rd_backend size=1GB Generating a wwn serial. Created rd_mcp ramdisk rd_backend with size 1GB.25.1.3. Creating an iSCSI Target
To create an iSCSI target:
Procedure 25.1. Creating an iSCSI target
Run targetcli.
Move into the iSCSI configuration path:
/> iscsi/The cd command is also accepted to change directories, as well as simply listing the path to move into.
Create an iSCSI target using a default target name.
/iscsi> create Created target iqn.2003-01.org.linux-iscsi.hostname.x8664:sn.78b473f296ff Created TPG1Or create an iSCSI target using a specified name.
/iscsi > create iqn.2006-04.com.example:444 Created target iqn.2006-04.com.example:444 Created TPG1Verify that the newly created target is visible when targets are listed with ls.
/iscsi > ls o- iscsi.......................................[1 Target] o- iqn.2006-04.com.example:444................[1 TPG] o- tpg1...........................[enabled, auth] o- acls...............................[0 ACL] o- luns...............................[0 LUN] o- portals.........................[0 Portal]
As of Red Hat Enterprise Linux 7.1, whenever a target is created, a default portal is also created.
25.1.4. Configuring an iSCSI Portal
To configure an iSCSI portal, an iSCSI target must first be created and associated with a TPG. For instructions on how to do this, refer to Section 25.1.3, “Creating an iSCSI Target”.
As of Red Hat Enterprise Linux 7.1 when an iSCSI target is created, a default portal is created as well. This portal is set to listen on all IP addresses with the default port number (that is, 0.0.0.0:3260). To remove this and add only specified portals, use /iscsi/iqn-name/tpg1/portals delete ip_address=0.0.0.0 ip_port=3260 then create a new portal with the required information.
Procedure 25.2. Creating an iSCSI Portal
Move into the TPG.
/iscsi> iqn.2006-04.example:444/tpg1/There are two ways to create a portal: create a default portal, or create a portal specifying what IP address to listen to.
Creating a default portal uses the default iSCSI port 3260 and allows the target to listen on all IP addresses on that port.
/iscsi/iqn.20...mple:444/tpg1> portals/ create Using default IP port 3260 Binding to INADDR_Any (0.0.0.0) Created network portal 0.0.0.0:3260To create a portal specifying what IP address to listen to, use the following command.
/iscsi/iqn.20...mple:444/tpg1> portals/ create 192.168.122.137 Using default IP port 3260 Created network portal 192.168.122.137:3260Verify that the newly created portal is visible with the ls command.
/iscsi/iqn.20...mple:444/tpg1> ls o- tpg.................................. [enambled, auth] o- acls ......................................[0 ACL] o- luns ......................................[0 LUN] o- portals ................................[1 Portal] o- 192.168.122.137:3260......................[OK]
25.1.5. Configuring LUNs
Procedure 25.3. Configuring LUNs
Create LUNs of already created storage objects.
/iscsi/iqn.20...mple:444/tpg1> luns/ create /backstores/ramdisk/rd_backend Created LUN 0. /iscsi/iqn.20...mple:444/tpg1> luns/ create /backstores/block/block_backend Created LUN 1. /iscsi/iqn.20...mple:444/tpg1> luns/ create /backstores/fileio/file1 Created LUN 2.Show the changes.
/iscsi/iqn.20...mple:444/tpg1> ls o- tpg.................................. [enambled, auth] o- acls ......................................[0 ACL] o- luns .....................................[3 LUNs] | o- lun0.........................[ramdisk/ramdisk1] | o- lun1.................[block/block1 (/dev/vdb1)] | o- lun2...................[fileio/file1 (/foo.img)] o- portals ................................[1 Portal] o- 192.168.122.137:3260......................[OK]Be aware that the default LUN name starts at 0, as opposed to 1 as was the case when using tgtd in Red Hat Enterprise Linux 6.
By default, LUNs are created with read-write permissions. In the event that a new LUN is added after ACLs have been created that LUN will be automatically mapped to all available ACLs. This can cause a security risk. Use the following procedure to create a LUN as read-only.
Procedure 25.4. Create a Read-only LUN
To create a LUN with read-only permissions, first use the following command:
/> set global auto_add_mapped_luns=false Parameter auto_add_mapped_luns is now 'false'.This prevents the auto mapping of LUNs to existing ACLs allowing the manual mapping of LUNs.
Next, manually create the LUN with the command iscsi/target_iqn_name/tpg1/acls/initiator_iqn_name/ create mapped_lun=next_sequential_LUN_number tpg_lun_or_backstore=backstore write_protect=1.
/> iscsi/iqn.2015-06.com.redhat:target/tpg1/acls/iqn.2015-06.com.redhat:initiator/ create mapped_lun=1 tpg_lun_or_backstore=/backstores/block/block2 write_protect=1 Created LUN 1. Created Mapped LUN 1. /> ls o- / ...................................................... [...] o- backstores ........................................... [...] <snip> o- iscsi ......................................... [Targets: 1] | o- iqn.2015-06.com.redhat:target .................. [TPGs: 1] | o- tpg1 ............................ [no-gen-acls, no-auth] | o- acls ....................................... [ACLs: 2] | | o- iqn.2015-06.com.redhat:initiator .. [Mapped LUNs: 2] | | | o- mapped_lun0 .............. [lun0 block/disk1 (rw)] | | | o- mapped_lun1 .............. [lun1 block/disk2 (ro)] | o- luns ....................................... [LUNs: 2] | | o- lun0 ...................... [block/disk1 (/dev/vdb)] | | o- lun1 ...................... [block/disk2 (/dev/vdc)] <snip>The mapped_lun1 line now has (ro) at the end (unlike mapped_lun0's (rw)) stating that it is read-only.
25.1.6. Configuring ACLs
Create an ACL for each initiator that will be connecting. This enforces authentication when that initiator connects, allowing only LUNs to be exposed to each initiator. Usually each initator has exclusive access to a LUN. Both targets and initiators have unique identifying names. The initiator's unique name must be known to configure ACLs. For open-iscsi initiators, this can be found in /etc/iscsi/initiatorname.iscsi.
Procedure 25.5. Configuring ACLs
Move into the acls directory.
/iscsi/iqn.20...mple:444/tpg1> acls/Create an ACL. Either use the initiator name found in /etc/iscsi/initiatorname.iscsi on the initiator, or if using a name that is easier to remember, refer to Section 25.2, “Creating an iSCSI Initiator” to ensure ACL matches the initiator. For example:
/iscsi/iqn.20...444/tpg1/acls> create iqn.2006-04.com.example.foo:888 Created Node ACL for iqn.2006-04.com.example.foo:888 Created mapped LUN 2. Created mapped LUN 1. Created mapped LUN 0.The given example's behavior depends on the setting used. In this case, the global setting auto_add_mapped_luns is used. This automatically maps LUNs to any created ACL.
You can set user-created ACLs within the TPG node on the target server:
/iscsi/iqn.20...scsi:444/tpg1> set attribute generate_node_acls=1Show the changes.
/iscsi/iqn.20...444/tpg1/acls> ls o- acls .................................................[1 ACL] o- iqn.2006-04.com.example.foo:888 ....[3 Mapped LUNs, auth] o- mapped_lun0 .............[lun0 ramdisk/ramdisk1 (rw)] o- mapped_lun1 .................[lun1 block/block1 (rw)] o- mapped_lun2 .................[lun2 fileio/file1 (rw)]
25.1.7. Configuring Fibre Channel over Ethernet (FCoE) Target
Procedure 25.6. Configure FCoE target
Setting up an FCoE target requires the installation of the targetcli package, along with its dependencies. Refer to Section 25.1, “Target Setup” for more information on targetcli basics and set up.
Create an FCoE target instance on an FCoE interface.
/> tcm_fc/ create 00:11:22:33:44:55:66:77If FCoE interfaces are present on the system, tab-completing after create will list available interfaces. If not, ensure fcoeadm -i shows active interfaces.
Map a backstore to the target instance.
Example 25.1. Example of Mapping a Backstore to the Target Instance
/> tcm_fc/00:11:22:33:44:55:66:77/> luns/ create /backstores/fileio/example2
Allow access to the LUN from an FCoE initiator.
/> acls/ create 00:99:88:77:66:55:44:33The LUN should now be accessible to that initiator.
To make the changes persistent across reboots, use the saveconfig command and type yes when prompted. If this is not done the configuration will be lost after rebooting.
Exit targetcli by typing exit or entering ctrl+D.
25.1.8. Removing Objects with targetcli
To remove an backstore use the command:
/> /backstores/backstore-type/backstore-nameTo remove parts of an iSCSI target, such as an ACL, use the following command:
/> /iscsi/iqn-name/tpg/acls/ delete iqn-nameTo remove the entire target, including all ACLs, LUNs, and portals, use the following command:
/> /iscsi delete iqn-name25.1.9. targetcli References
For more information on targetcli, refer to the following resources:
man targetcli
The targetcli man page. It includes an example walk through.
The Linux SCSI Target WikiScreencast by Andy GroverThis was uploaded on February 28, 2012. As such, the service name has changed from targetcli to target.
25.2. Creating an iSCSI Initiator
In Red Hat Enterprise Linux 7, the iSCSI service is lazily started by default: the service starts after running the iscsiadm command.
Procedure 25.7. Creating an iSCSI Initiator
Install iscsi-initiator-utils:
# yum install iscsi-initiator-utils -yIf the ACL was given a custom name in Section 25.1.6, “Configuring ACLs”, modify the /etc/iscsi/initiatorname.iscsi file accordingly. For example:
# cat /etc/iscsi/initiatorname.iscsi InitiatorName=iqn.2006-04.com.example.node1 # vi /etc/iscsi/initiatorname.iscsiDiscover the target:
# iscsiadm -m discovery -t st -p target-ip-address 10.64.24.179:3260,1 iqn.2006-04.com.example:3260Log in to the target with the target IQN you discovered in step 3:
# iscsiadm -m node -T iqn.2006-04.com.example:3260 -l Logging in to [iface: default, target: iqn.2006-04.com.example:3260, portal: 10.64.24.179,3260] (multiple) Login to [iface: default, target: iqn.2006-04.com.example:3260, portal: 10.64.24.179,3260] successful.This procedure can be followed for any number of initators connected to the same LUN so long as their specific initiator names are added to the ACL as described in Section 25.1.6, “Configuring ACLs”.
Find the iSCSI disk name and create a file system on this iSCSI disk:
# grep "Attached SCSI" /var/log/messages # mkfs.ext4 /dev/disk_nameReplace disk_name with the iSCSI disk name displayed in /var/log/messages.
Mount the file system:
# mkdir /mount/point # mount /dev/disk_name /mount/pointReplace /mount/point with the mount point of the partition.
Edit the /etc/fstab to mount the file system automatically when the system boots:
# vim /etc/fstab /dev/disk_name /mount/point ext4 _netdev 0 0Replace disk_name with the iSCSI disk name.
Log off from the target:
# iscsiadm -m node -T iqn.2006-04.com.example:3260 -u
25.3. Setting up the Challenge-Handshake Authentication Protocol
The CHAP allows the user to protect the target with a password. The initiator must be aware of this password to be able to connect to the target.
Procedure 25.8. Setting up the CHAP for target
Set attribute authentication:
/iscsi/iqn.20...mple:444/tpg1> set attribute authentication=1 Parameter authentication is now '1'.Set userid and password:
/iscsi/iqn.20...mple:444/tpg1> set auth userid=redhat Parameter userid is now 'redhat'. /iscsi/iqn.20...mple:444/tpg1> set auth password=redhat_passwd Parameter password is now 'redhat_passwd'.
Procedure 25.9. Setting up the CHAP for initiator
Edit the iscsid.conf file:
Enable the CHAP authentication in the iscsid.conf file:
# vi /etc/iscsi/iscsid.conf node.session.auth.authmethod = CHAPBy default, the node.session.auth.authmethod option is set to None.
Add target user name and password in the iscsid.conf file:
node.session.auth.username = redhat node.session.auth.password = redhat_passwd
Restart the iscsid service:
# systemctl restart iscsid.service
For more information, see the targetcli and iscsiadm man pages.
25.4. Fibre Channel
This section discusses the Fibre Channel API, native Red Hat Enterprise Linux 7 Fibre Channel drivers, and the Fibre Channel capabilities of these drivers.
25.4.1. Fibre Channel API
Following is a list of /sys/class/ directories that contain files used to provide the userspace API. In each item, host numbers are designated by H, bus numbers are B, targets are T, logical unit numbers (LUNs) are L, and remote port numbers are R.
If your system is using multipath software, Red Hat recommends that you consult your hardware vendor before changing any of the values described in this section.
Transport: /sys/class/fc_transport/targetH:B:T/
port_id — 24-bit port ID/address
node_name — 64-bit node name
port_name — 64-bit port name
port_id
node_name
port_name
dev_loss_tmo: controls when the scsi device gets removed from the system. After dev_loss_tmo triggers, the scsi device is removed.
In multipath.conf, you can set dev_loss_tmo to infinity, which sets its value to 2,147,483,647 seconds, or 68 years, and is the maximum dev_loss_tmo value.
In Red Hat Enterprise Linux 7, if you do not set the fast_io_fail_tmo option, dev_loss_tmo is capped to 600 seconds. By default, fast_io_fail_tmo is set to 5 seconds in Red Hat Enterprise Linux 7 if the multipathd service is running; otherwise, it is set to off.
fast_io_fail_tmo: specifies the number of seconds to wait before it marks a link as "bad". Once a link is marked bad, existing running I/O or any new I/O on its corresponding path fails.
If I/O is in a blocked queue, it will not be failed until dev_loss_tmo expires and the queue is unblocked.
If fast_io_fail_tmo is set to any value except off, dev_loss_tmo is uncapped. If fast_io_fail_tmo is set to off, no I/O fails until the device is removed from the system. If fast_io_fail_tmo is set to a number, I/O fails immediately when the fast_io_fail_tmo timeout triggers.
port_id
issue_lip: instructs the driver to rediscover remote ports.
25.4.2. Native Fibre Channel Drivers and Capabilities
Red Hat Enterprise Linux 7 ships with the following native Fibre Channel drivers:
lpfc
qla2xxx
zfcp
bfa
The qla2xxx driver runs in initiator mode by default. To use qla2xxx with Linux-IO, enable Fibre Channel target mode with the corresponding qlini_mode module parameter.
First, make sure that the firmware package for your qla device, such as ql2200-firmware or similar, is installed.
To enable target mode, add the following parameter to the /usr/lib/modprobe.d/qla2xxx.conf qla2xxx module configuration file:
options qla2xxx qlini_mode=disabledThen, use the dracut -f command to rebuild the initial ramdisk (initrd), and reboot the system for the changes to take effect.
Table 25.1, “Fibre Channel API Capabilities” describes the different Fibre Channel API capabilities of each native Red Hat Enterprise Linux 7 driver. X denotes support for the capability.
Table 25.1. Fibre Channel API Capabilities
Transport port_id | X | X | X | X |
Transport node_name | X | X | X | X |
Transport port_name | X | X | X | X |
Remote Port dev_loss_tmo | X | X | X | X |
Remote Port fast_io_fail_tmo | X | X [a] | X [b] | X |
Host port_id | X | X | X | X |
Host issue_lip | X | X | X | |
[a] Supported as of Red Hat Enterprise Linux 5.4 [b] Supported as of Red Hat Enterprise Linux 6.0 |
25.5. Configuring a Fibre Channel over Ethernet Interface
Setting up and deploying a Fibre Channel over Ethernet (FCoE) interface requires two packages:
fcoe-utils
lldpad
Once these packages are installed, perform the following procedure to enable FCoE over a virtual LAN (VLAN):
Procedure 25.10. Configuring an Ethernet Interface to Use FCoE
To configure a new VLAN, make a copy of an existing network script, for example /etc/fcoe/cfg-eth0, and change the name to the Ethernet device that supports FCoE. This provides you with a default file to configure. Given that the FCoE device is ethX, run:
# cp /etc/fcoe/cfg-ethx /etc/fcoe/cfg-ethXModify the contents of cfg-ethX as needed. Notably, set DCB_REQUIRED to no for networking interfaces that implement a hardware Data Center Bridging Exchange (DCBX) protocol client.
If you want the device to automatically load during boot time, set ONBOOT=yes in the corresponding /etc/sysconfig/network-scripts/ifcfg-ethX file. For example, if the FCoE device is eth2, edit /etc/sysconfig/network-scripts/ifcfg-eth2 accordingly.
Start the data center bridging daemon (dcbd) by running:
# systemctl start lldpadFor networking interfaces that implement a hardware DCBX client, skip this step.
For interfaces that require a software DCBX client, enable data center bridging on the Ethernet interface by running:
# dcbtool sc ethX dcb onThen, enable FCoE on the Ethernet interface by running:
# dcbtool sc ethX app:fcoe e:1Note that these commands only work if the dcbd settings for the Ethernet interface were not changed.
Load the FCoE device now using:
# ip link set dev ethX upStart FCoE using:
# systemctl start fcoeThe FCoE device appears soon if all other settings on the fabric are correct. To view configured FCoE devices, run:
# fcoeadm -i
After correctly configuring the Ethernet interface to use FCoE, Red Hat recommends that you set FCoE and the lldpad service to run at startup. To do so, use the systemctl utility:
# systemctl enable lldpad# systemctl enable fcoeRunning the # systemctl stop fcoe command stops the daemon, but does not reset the configuration of FCoE interfaces. To do so, run the # systemctl -s SIGHUP kill fcoe command.
As of Red Hat Enterprise Linux 7, Network Manager has the ability to query and set the DCB settings of a DCB capable Ethernet interface.
25.6. Configuring an FCoE Interface to Automatically Mount at Boot
The instructions in this section are available in /usr/share/doc/fcoe-utils-version/README as of Red Hat Enterprise Linux 6.1. Refer to that document for any possible changes throughout minor releases.
You can mount newly discovered disks via udev rules, autofs, and other similar methods. Sometimes, however, a specific service might require the FCoE disk to be mounted at boot-time. In such cases, the FCoE disk should be mounted as soon as the fcoe service runs and before the initiation of any service that requires the FCoE disk.
To configure an FCoE disk to automatically mount at boot, add proper FCoE mounting code to the startup script for the fcoe service. The fcoe startup script is /lib/systemd/system/fcoe.service.
The FCoE mounting code is different per system configuration, whether you are using a simple formatted FCoE disk, LVM, or multipathed device node.
Example 25.2. FCoE Mounting Code
The following is a sample FCoE mounting code for mounting file systems specified via wild cards in /etc/fstab:
mount_fcoe_disks_from_fstab() { local timeout=20 local done=1 local fcoe_disks=($(egrep 'by-path\/fc-.*_netdev' /etc/fstab | cut -d ' ' -f1)) test -z $fcoe_disks && return 0 echo -n "Waiting for fcoe disks . " while [ $timeout -gt 0 ]; do for disk in ${fcoe_disks[*]}; do if ! test -b $disk; then done=0 break fi done test $done -eq 1 && break; sleep 1 echo -n ". " done=1 let timeout-- done if test $timeout -eq 0; then echo "timeout!" else echo "done!" fi # mount any newly discovered disk mount -a 2>/dev/null }The mount_fcoe_disks_from_fstab function should be invoked after the fcoe service script starts the fcoemon daemon. This will mount FCoE disks specified by the following paths in /etc/fstab:
/dev/disk/by-path/fc-0xXX:0xXX /mnt/fcoe-disk1 ext3 defaults,_netdev 0 0 /dev/disk/by-path/fc-0xYY:0xYY /mnt/fcoe-disk2 ext3 defaults,_netdev 0 0Entries with fc- and _netdev sub-strings enable the mount_fcoe_disks_from_fstab function to identify FCoE disk mount entries. For more information on /etc/fstab entries, refer to man 5 fstab.
The fcoe service does not implement a timeout for FCoE disk discovery. As such, the FCoE mounting code should implement its own timeout period.
25.7. iSCSI
This section describes the iSCSI API and the iscsiadm utility. Before using the iscsiadm utility, install the iscsi-initiator-utils package first by running yum install iscsi-initiator-utils.
In Red Hat Enterprise Linux 7, the iSCSI service is lazily started by default. If root is not on an iSCSI device or there are no nodes marked with node.startup = automatic then the iSCSI service will not start until an iscsiadm command is run that requires iscsid or the iscsi kernel modules to be started. For example, running the discovery command iscsiadm -m discovery -t st -p ip:port will cause iscsiadm to start the iSCSI service.
To force the iscsid daemon to run and iSCSI kernel modules to load, run systemctl start iscsid.service.
25.7.1. iSCSI API
To get information about running sessions, run:
# iscsiadm -m session -P 3This command displays the session/device state, session ID (sid), some negotiated parameters, and the SCSI devices accessible through the session.
For shorter output (for example, to display only the sid-to-node mapping), run:
# iscsiadm -m session -P 0or
# iscsiadm -m sessionThese commands print the list of running sessions with the format:
driver [sid] target_ip:port,target_portal_group_tag proper_target_nameExample 25.3. Output of the iscsisadm -m session Command
For example:
# iscsiadm -m session tcp [2] 10.15.84.19:3260,2 iqn.1992-08.com.netapp:sn.33615311 tcp [3] 10.15.85.19:3260,3 iqn.1992-08.com.netapp:sn.33615311For more information about the iSCSI API, refer to /usr/share/doc/iscsi-initiator-utils-version/README.
25.8. Persistent Naming
Red Hat Enterprise Linux provides a number of ways to identify storage devices. It is important to use the correct option to identify each device when used in order to avoid inadvertently accessing the wrong device, particularly when installing to or reformatting drives.
25.8.1. Major and Minor Numbers of Storage Devices
Storage devices managed by the sd driver are identified internally by a collection of major device numbers and their associated minor numbers. The major device numbers used for this purpose are not in a contiguous range. Each storage device is represented by a major number and a range of minor numbers, which are used to identify either the entire device or a partition within the device. There is a direct association between the major and minor numbers allocated to a device and numbers in the form of sd<letter(s)>[number(s)]. Whenever the sd driver detects a new device, an available major number and minor number range is allocated. Whenever a device is removed from the operating system, the major number and minor number range is freed for later reuse.
The major and minor number range and associated sd names are allocated for each device when it is detected. This means that the association between the major and minor number range and associated sd names can change if the order of device detection changes. Although this is unusual with some hardware configurations (for example, with an internal SCSI controller and disks that have their SCSI target ID assigned by their physical location within a chassis), it can nevertheless occur. Examples of situations where this can happen are as follows:
A disk may fail to power up or respond to the SCSI controller. This will result in it not being detected by the normal device probe. The disk will not be accessible to the system and subsequent devices will have their major and minor number range, including the associated sd names shifted down. For example, if a disk normally referred to as sdb is not detected, a disk that is normally referred to as sdc would instead appear as sdb.
A SCSI controller (host bus adapter, or HBA) may fail to initialize, causing all disks connected to that HBA to not be detected. Any disks connected to subsequently probed HBAs would be assigned different major and minor number ranges, and different associated sd names.
The order of driver initialization could change if different types of HBAs are present in the system. This would cause the disks connected to those HBAs to be detected in a different order. This can also occur if HBAs are moved to different PCI slots on the system.
Disks connected to the system with Fibre Channel, iSCSI, or FCoE adapters might be inaccessible at the time the storage devices are probed, due to a storage array or intervening switch being powered off, for example. This could occur when a system reboots after a power failure, if the storage array takes longer to come online than the system take to boot. Although some Fibre Channel drivers support a mechanism to specify a persistent SCSI target ID to WWPN mapping, this will not cause the major and minor number ranges, and the associated sd names to be reserved, it will only provide consistent SCSI target ID numbers.
These reasons make it undesirable to use the major and minor number range or the associated sd names when referring to devices, such as in the /etc/fstab file. There is the possibility that the wrong device will be mounted and data corruption could result.
Occasionally, however, it is still necessary to refer to the sd names even when another mechanism is used (such as when errors are reported by a device). This is because the Linux kernel uses sd names (and also SCSI host/channel/target/LUN tuples) in kernel messages regarding the device.
25.8.2. World Wide Identifier (WWID)
The World Wide Identifier (WWID) can be used in reliably identifying devices. It is a persistent, system-independent ID that the SCSI Standard requires from all SCSI devices. The WWID identifier is guaranteed to be unique for every storage device, and independent of the path that is used to access the device.
This identifier can be obtained by issuing a SCSI Inquiry to retrieve the Device Identification Vital Product Data (page 0x83) or Unit Serial Number (page 0x80). The mappings from these WWIDs to the current /dev/sd names can be seen in the symlinks maintained in the /dev/disk/by-id/ directory.
Example 25.4. WWID
For example, a device with a page 0x83 identifier would have:
scsi-3600508b400105e210000900000490000 -> ../../sdaOr, a device with a page 0x80 identifier would have:
scsi-SSEAGATE_ST373453LW_3HW1RHM6 -> ../../sdaRed Hat Enterprise Linux automatically maintains the proper mapping from the WWID-based device name to a current /dev/sd name on that system. Applications can use the /dev/disk/by-id/ name to reference the data on the disk, even if the path to the device changes, and even when accessing the device from different systems.
If there are multiple paths from a system to a device, DM Multipath uses the WWID to detect this. DM Multipath then presents a single "pseudo-device" in the /dev/mapper/wwid directory, such as /dev/mapper/3600508b400105df70000e00000ac0000.
The command multipath -l shows the mapping to the non-persistent identifiers: Host:Channel:Target:LUN, /dev/sd name, and the major:minor number.
3600508b400105df70000e00000ac0000 dm-2 vendor,product [size=20G][features=1 queue_if_no_path][hwhandler=0][rw] \_ round-robin 0 [prio=0][active] \_ 5:0:1:1 sdc 8:32 [active][undef] \_ 6:0:1:1 sdg 8:96 [active][undef] \_ round-robin 0 [prio=0][enabled] \_ 5:0:0:1 sdb 8:16 [active][undef] \_ 6:0:0:1 sdf 8:80 [active][undef]DM Multipath automatically maintains the proper mapping of each WWID-based device name to its corresponding /dev/sd name on the system. These names are persistent across path changes, and they are consistent when accessing the device from different systems.
When the user_friendly_names feature (of DM Multipath) is used, the WWID is mapped to a name of the form /dev/mapper/mpathn. By default, this mapping is maintained in the file /etc/multipath/bindings. These mpathn names are persistent as long as that file is maintained.
In addition to these persistent names provided by the system, you can also use udev rules to implement persistent names of your own, mapped to the WWID of the storage.
25.8.3. Device Names Managed by the udev Mechanism in /dev/disk/by-*
The udev mechanism consists of three major components:
The kernel
Generates events that are sent to user space when devices are added, removed, or changed.
The udevd serviceReceives the events.
The udev rulesSpecifies the action to take when the udev service receives the kernel events.
This mechanism is used for all types of devices in Linux, not just for storage devices. In the case of storage devices, Red Hat Enterprise Linux contains udev rules that create symbolic links in the /dev/disk/ directory allowing storage devices to be referred to by their contents, a unique identifier, their serial number, or the hardware path used to access the device.
/dev/disk/by-label/
Entries in this directory provide a symbolic name that refers to the storage device by a label in the contents (that is, the data) stored on the device. The blkid utility is used to read data from the device and determine a name (that is, a label) for the device. For example:
/dev/disk/by-label/BootThe information is obtained from the contents (that is, the data) on the device so if the contents are copied to another device, the label will remain the same.
The label can also be used to refer to the device in /etc/fstab using the following syntax:
LABEL=Boot/dev/disk/by-uuid/Entries in this directory provide a symbolic name that refers to the storage device by a unique identifier in the contents (that is, the data) stored on the device. The blkid utility is used to read data from the device and obtain a unique identifier (that is, the UUID) for the device. For example:
UUID=3e6be9de-8139-11d1-9106-a43f08d823a6/dev/disk/by-id/Entries in this directory provide a symbolic name that refers to the storage device by a unique identifier (different from all other storage devices). The identifier is a property of the device but is not stored in the contents (that is, the data) on the devices. For example:
/dev/disk/by-id/scsi-3600508e000000000ce506dc50ab0ad05 /dev/disk/by-id/wwn-0x600508e000000000ce506dc50ab0ad05The id is obtained from the world-wide ID of the device, or the device serial number. The /dev/disk/by-id/ entries may also include a partition number. For example:
/dev/disk/by-id/scsi-3600508e000000000ce506dc50ab0ad05-part1/dev/disk/by-id/wwn-0x600508e000000000ce506dc50ab0ad05-part1/dev/disk/by-path/Entries in this directory provide a symbolic name that refers to the storage device by the hardware path used to access the device, beginning with a reference to the storage controller in the PCI hierarchy, and including the SCSI host, channel, target, and LUN numbers and, optionally, the partition number. Although these names are preferable to using major and minor numbers or sd names, caution must be used to ensure that the target numbers do not change in a Fibre Channel SAN environment (for example, through the use of persistent binding) and that the use of the names is updated if a host adapter is moved to a different PCI slot. In addition, there is the possibility that the SCSI host numbers could change if a HBA fails to probe, if drivers are loaded in a different order, or if a new HBA is installed on the system. An example of by-path listing is:
/dev/disk/by-path/pci-0000:03:00.0-scsi-0:1:0:0The /dev/disk/by-path/ entries may also include a partition number, such as:
/dev/disk/by-path/pci-0000:03:00.0-scsi-0:1:0:0-part125.8.3.1. Limitations of the udev Device Naming Convention
The following are some limitations of the udev naming convention.
It is possible that the device may not be accessible at the time the query is performed because the udev mechanism may rely on the ability to query the storage device when the udev rules are processed for a udev event. This is more likely to occur with Fibre Channel, iSCSI or FCoE storage devices when the device is not located in the server chassis.
The kernel may also send udev events at any time, causing the rules to be processed and possibly causing the /dev/disk/by-*/ links to be removed if the device is not accessible.
There can be a delay between when the udev event is generated and when it is processed, such as when a large number of devices are detected and the user-space udevd service takes some amount of time to process the rules for each one). This could cause a delay between when the kernel detects the device and when the /dev/disk/by-*/ names are available.
External programs such as blkid invoked by the rules may open the device for a brief period of time, making the device inaccessible for other uses.
25.8.3.2. Modifying Persistent Naming Attributes
Although udev naming attributes are persistent, in that they do not change on their own across system reboots, some are also configurable. You can set custom values for the following persistent naming attributes:
UUID: file system UUID
LABEL: file system label
Because the UUID and LABEL attributes are related to the file system, the tool you need to use depends on the file system on that partition.
To change the UUID or LABEL attributes of an XFS file system, unmount the file system and then use the xfs_admin utility to change the attribute:
# umount /dev/device # xfs_admin [-U new_uuid] [-L new_label] /dev/device # udevadm settleTo change the UUID or LABEL attributes of an ext4, ext3, or ext2 file system, use the tune2fs utility:
# tune2fs [-U new_uuid] [-L new_label] /dev/device # udevadm settle
Replace new_uuid with the UUID you want to set; for example, 1cdfbc07-1c90-4984-b5ec-f61943f5ea50. Replace new_label with a label; for example, backup_data.
Changing udev attributes happens in the background and might take a long time. The udevadm settle command waits until the change is fully registered, which ensures that your next command will be able to utilize the new attribute correctly.
You should also use the command after creating new devices; for example, after using the parted tool to create a partition with a custom PARTUUID or PARTLABEL attribute, or after creating a new file system.
25.9. Removing a Storage Device
Before removing access to the storage device itself, it is advisable to back up data from the device first. Afterwards, flush I/O and remove all operating system references to the device (as described below). If the device uses multipathing, then do this for the multipath "pseudo device" (Section 25.8.2, “World Wide Identifier (WWID)”) and each of the identifiers that represent a path to the device. If you are only removing a path to a multipath device, and other paths will remain, then the procedure is simpler, as described in Section 25.11, “Adding a Storage Device or Path”.
Removal of a storage device is not recommended when the system is under memory pressure, since the I/O flush will add to the load. To determine the level of memory pressure, run the command vmstat 1 100; device removal is not recommended if:
Free memory is less than 5% of the total memory in more than 10 samples per 100 (the command free can also be used to display the total memory).
Swapping is active (non-zero si and so columns in the vmstat output).
The general procedure for removing all access to a device is as follows:
Procedure 25.11. Ensuring a Clean Device Removal
Close all users of the device and backup device data as needed.
Use umount to unmount any file systems that mounted the device.
Remove the device from any md and LVM volume using it. If the device is a member of an LVM Volume group, then it may be necessary to move data off the device using the pvmove command, then use the vgreduce command to remove the physical volume, and (optionally) pvremove to remove the LVM metadata from the disk.
Run multipath -l command to find the list of devices which are configured as multipath device. If the device is configured as a multipath device, run multipath -f device command to flush any outstanding I/O and to remove the multipath device.
Flush any outstanding I/O to the used paths. This is important for raw devices, where there is no umount or vgreduce operation to cause an I/O flush. You need to do this step only if:
the device is not configured as a multipath device, or
the device is configured as a multipath device and I/O has been issued directly to its individual paths at some point in the past.
Use the following command to flush any outstanding I/O:
# blockdev --flushbufs deviceRemove any reference to the device's path-based name, like /dev/sd, /dev/disk/by-path or the major:minor number, in applications, scripts, or utilities on the system. This is important in ensuring that different devices added in the future will not be mistaken for the current device.
Finally, remove each path to the device from the SCSI subsystem. To do so, use the command echo 1 > /sys/block/device-name/device/delete where device-name may be sde, for example.
Another variation of this operation is echo 1 > /sys/class/scsi_device/h:c:t:l/device/delete, where h is the HBA number, c is the channel on the HBA, t is the SCSI target ID, and l is the LUN.
The older form of these commands, echo "scsi remove-single-device 0 0 0 0" > /proc/scsi/scsi, is deprecated.
You can determine the device-name, HBA number, HBA channel, SCSI target ID and LUN for a device from various commands, such as lsscsi, scsi_id, multipath -l, and ls -l /dev/disk/by-*.
Other procedures, such as the physical removal of the device, followed by a rescan of the SCSI bus (as described in Section 25.12, “Scanning Storage Interconnects”) to cause the operating system state to be updated to reflect the change, are not recommended. This will cause delays due to I/O timeouts, and devices may be removed unexpectedly. If it is necessary to perform a rescan of an interconnect, it must be done while I/O is paused, as described in Section 25.12, “Scanning Storage Interconnects”.
25.10. Removing a Path to a Storage Device
If you are removing a path to a device that uses multipathing (without affecting other paths to the device), then the general procedure is as follows:
Procedure 25.12. Removing a Path to a Storage Device
Remove any reference to the device's path-based name, like /dev/sd or /dev/disk/by-path or the major:minor number, in applications, scripts, or utilities on the system. This is important in ensuring that different devices added in the future will not be mistaken for the current device.
Take the path offline using echo offline > /sys/block/sda/device/state.
This will cause any subsequent I/O sent to the device on this path to be failed immediately. Device-mapper-multipath will continue to use the remaining paths to the device.
Remove the path from the SCSI subsystem. To do so, use the command echo 1 > /sys/block/device-name/device/delete where device-name may be sde, for example (as described in Procedure 25.11, “Ensuring a Clean Device Removal”).
After performing Procedure 25.12, “Removing a Path to a Storage Device”, the path can be safely removed from the running system. It is not necessary to stop I/O while this is done, as device-mapper-multipath will re-route I/O to remaining paths according to the configured path grouping and failover policies.
Other procedures, such as the physical removal of the cable, followed by a rescan of the SCSI bus to cause the operating system state to be updated to reflect the change, are not recommended. This will cause delays due to I/O timeouts, and devices may be removed unexpectedly. If it is necessary to perform a rescan of an interconnect, it must be done while I/O is paused, as described in Section 25.12, “Scanning Storage Interconnects”.
25.11. Adding a Storage Device or Path
When adding a device, be aware that the path-based device name (/dev/sd name, major:minor number, and /dev/disk/by-path name, for example) the system assigns to the new device may have been previously in use by a device that has since been removed. As such, ensure that all old references to the path-based device name have been removed. Otherwise, the new device may be mistaken for the old device.
Procedure 25.13. Add a Storage Device or Path
The first step in adding a storage device or path is to physically enable access to the new storage device, or a new path to an existing device. This is done using vendor-specific commands at the Fibre Channel or iSCSI storage server. When doing so, note the LUN value for the new storage that will be presented to your host. If the storage server is Fibre Channel, also take note of the World Wide Node Name (WWNN) of the storage server, and determine whether there is a single WWNN for all ports on the storage server. If this is not the case, note the World Wide Port Name (WWPN) for each port that will be used to access the new LUN.
Next, make the operating system aware of the new storage device, or path to an existing device. The recommended command to use is:
$ echo "c t l" > /sys/class/scsi_host/hosth/scanIn the previous command, h is the HBA number, c is the channel on the HBA, t is the SCSI target ID, and l is the LUN.
The older form of this command, echo "scsi add-single-device 0 0 0 0" > /proc/scsi/scsi, is deprecated.
In some Fibre Channel hardware, a newly created LUN on the RAID array may not be visible to the operating system until a Loop Initialization Protocol (LIP) operation is performed. Refer to Section 25.12, “Scanning Storage Interconnects” for instructions on how to do this.
It will be necessary to stop I/O while this operation is executed if an LIP is required.
If a new LUN has been added on the RAID array but is still not being configured by the operating system, confirm the list of LUNs being exported by the array using the sg_luns command, part of the sg3_utils package. This will issue the SCSI REPORT LUNS command to the RAID array and return a list of LUNs that are present.
For Fibre Channel storage servers that implement a single WWNN for all ports, you can determine the correct h,c,and t values (i.e. HBA number, HBA channel, and SCSI target ID) by searching for the WWNN in sysfs.
Example 25.5. Determine Correct h, c, and t Values
For example, if the WWNN of the storage server is 0x5006016090203181, use:
$ grep 5006016090203181 /sys/class/fc_transport/*/node_nameThis should display output similar to the following:
/sys/class/fc_transport/target5:0:2/node_name:0x5006016090203181 /sys/class/fc_transport/target5:0:3/node_name:0x5006016090203181 /sys/class/fc_transport/target6:0:2/node_name:0x5006016090203181 /sys/class/fc_transport/target6:0:3/node_name:0x5006016090203181This indicates there are four Fibre Channel routes to this target (two single-channel HBAs, each leading to two storage ports). Assuming a LUN value is 56, then the following command will configure the first path:
$ echo "0 2 56" > /sys/class/scsi_host/host5/scanThis must be done for each path to the new device.
For Fibre Channel storage servers that do not implement a single WWNN for all ports, you can determine the correct HBA number, HBA channel, and SCSI target ID by searching for each of the WWPNs in sysfs.
Another way to determine the HBA number, HBA channel, and SCSI target ID is to refer to another device that is already configured on the same path as the new device. This can be done with various commands, such as lsscsi, scsi_id, multipath -l, and ls -l /dev/disk/by-*. This information, plus the LUN number of the new device, can be used as shown above to probe and configure that path to the new device.
After adding all the SCSI paths to the device, execute the multipath command, and check to see that the device has been properly configured. At this point, the device can be added to md, LVM, mkfs, or mount, for example.
If the steps above are followed, then a device can safely be added to a running system. It is not necessary to stop I/O to other devices while this is done. Other procedures involving a rescan (or a reset) of the SCSI bus, which cause the operating system to update its state to reflect the current device connectivity, are not recommended while storage I/O is in progress.
25.12. Scanning Storage Interconnects
Certain commands allow you to reset, scan, or both reset and scan one or more interconnects, which potentially adds and removes multiple devices in one operation. This type of scan can be disruptive, as it can cause delays while I/O operations time out, and remove devices unexpectedly. Red Hat recommends using interconnect scanning only when necessary. Observe the following restrictions when scanning storage interconnects:
All I/O on the effected interconnects must be paused and flushed before executing the procedure, and the results of the scan checked before I/O is resumed.
As with removing a device, interconnect scanning is not recommended when the system is under memory pressure. To determine the level of memory pressure, run the vmstat 1 100 command. Interconnect scanning is not recommended if free memory is less than 5% of the total memory in more than 10 samples per 100. Also, interconnect scanning is not recommended if swapping is active (non-zero si and so columns in the vmstat output). The free command can also display the total memory.
The following commands can be used to scan storage interconnects:
echo "1" > /sys/class/fc_host/hostN/issue_lip
(Replace N with the host number.)
This operation performs a Loop Initialization Protocol (LIP), scans the interconnect, and causes the SCSI layer to be updated to reflect the devices currently on the bus. Essentially, an LIP is a bus reset, and causes device addition and removal. This procedure is necessary to configure a new SCSI target on a Fibre Channel interconnect.
Note that issue_lip is an asynchronous operation. The command can complete before the entire scan has completed. You must monitor /var/log/messages to determine when issue_lip finishes.
The lpfc, qla2xxx, and bnx2fc drivers support issue_lip. For more information about the API capabilities supported by each driver in Red Hat Enterprise Linux, see Table 25.1, “Fibre Channel API Capabilities”.
/usr/bin/rescan-scsi-bus.shThe /usr/bin/rescan-scsi-bus.sh script was introduced in Red Hat Enterprise Linux 5.4. By default, this script scans all the SCSI buses on the system, and updates the SCSI layer to reflect new devices on the bus. The script provides additional options to allow device removal, and the issuing of LIPs. For more information about this script, including known issues, see Section 25.18, “Adding/Removing a Logical Unit Through rescan-scsi-bus.sh”.
echo "- - -" > /sys/class/scsi_host/hosth/scanThis is the same command as described in Section 25.11, “Adding a Storage Device or Path” to add a storage device or path. In this case, however, the channel number, SCSI target ID, and LUN values are replaced by wildcards. Any combination of identifiers and wildcards is allowed, so you can make the command as specific or broad as needed. This procedure adds LUNs, but does not remove them.
modprobe --remove driver-name, modprobe driver-nameRunning the modprobe --remove driver-name command followed by the modprobe driver-name command completely re-initializes the state of all interconnects controlled by the driver. Despite being rather extreme, using the described commands can be appropriate in certain situations. The commands can be used, for example, to restart the driver with a different module parameter value.
25.13. iSCSI Discovery Configuration
The default iSCSI configuration file is /etc/iscsi/iscsid.conf. This file contains iSCSI settings used by iscsid and iscsiadm.
During target discovery, the iscsiadm tool uses the settings in /etc/iscsi/iscsid.conf to create two types of records:
Node records in /var/lib/iscsi/nodes
When logging into a target, iscsiadm uses the settings in this file.
Discovery records in /var/lib/iscsi/discovery_typeWhen performing discovery to the same destination, iscsiadm uses the settings in this file.
Before using different settings for discovery, delete the current discovery records (i.e. /var/lib/iscsi/discovery_type) first. To do this, use the following command: [5]
# iscsiadm -m discovery -t discovery_type -p target_IP:port -o deleteHere, discovery_type can be either sendtargets, isns, or fw.
For details on different types of discovery, refer to the DISCOVERY TYPES section of the iscsiadm(8) man page.
There are two ways to reconfigure discovery record settings:
Edit the /etc/iscsi/iscsid.conf file directly prior to performing a discovery. Discovery settings use the prefix discovery; to view them, run:
# iscsiadm -m discovery -t discovery_type -p target_IP:portAlternatively, iscsiadm can also be used to directly change discovery record settings, as in:
# iscsiadm -m discovery -t discovery_type -p target_IP:port -o update -n setting -v %valueRefer to the iscsiadm(8) man page for more information on available setting options and valid value options for each.
After configuring discovery settings, any subsequent attempts to discover new targets will use the new settings. Refer to Section 25.15, “Scanning iSCSI Interconnects” for details on how to scan for new iSCSI targets.
For more information on configuring iSCSI target discovery, refer to the man pages of iscsiadm and iscsid. The /etc/iscsi/iscsid.conf file also contains examples on proper configuration syntax.
25.14. Configuring iSCSI Offload and Interface Binding
This chapter describes how to set up iSCSI interfaces in order to bind a session to a NIC port when using software iSCSI. It also describes how to set up interfaces for use with network devices that support offloading.
The network subsystem can be configured to determine the path/NIC that iSCSI interfaces should use for binding. For example, if portals and NICs are set up on different subnets, then it is not necessary to manually configure iSCSI interfaces for binding.
Before attempting to configure an iSCSI interface for binding, run the following command first:
$ ping -I ethX target_IPIf ping fails, then you will not be able to bind a session to a NIC. If this is the case, check the network settings first.
25.14.1. Viewing Available iface Configurations
iSCSI offload and interface binding is supported for the following iSCSI initiator implementations:
Software iSCSI
This stack allocates an iSCSI host instance (that is, scsi_host) per session, with a single connection per session. As a result, /sys/class_scsi_host and /proc/scsi will report a scsi_host for each connection/session you are logged into.
Offload iSCSIThis stack allocates a scsi_host for each PCI device. As such, each port on a host bus adapter will show up as a different PCI device, with a different scsi_host per HBA port.
To manage both types of initiator implementations, iscsiadm uses the iface structure. With this structure, an iface configuration must be entered in /var/lib/iscsi/ifaces for each HBA port, software iSCSI, or network device (ethX) used to bind sessions.
To view available iface configurations, run iscsiadm -m iface. This will display iface information in the following format:
iface_name transport_name,hardware_address,ip_address,net_ifacename,initiator_nameRefer to the following table for an explanation of each value/setting.
Table 25.2. iface Settings
iface_name | iface configuration name. |
transport_name | Name of driver |
hardware_address | MAC address |
ip_address | IP address to use for this port |
net_iface_name | Name used for the vlan or alias binding of a software iSCSI session. For iSCSI offloads, net_iface_name will be <empty> because this value is not persistent across reboots. |
initiator_name | This setting is used to override a default name for the initiator, which is defined in /etc/iscsi/initiatorname.iscsi |
Example 25.6. Sample Output of the iscsiadm -m iface Command
The following is a sample output of the iscsiadm -m iface command:
iface0 qla4xxx,00:c0:dd:08:63:e8,20.15.0.7,default,iqn.2005-06.com.redhat:madmax iface1 qla4xxx,00:c0:dd:08:63:ea,20.15.0.9,default,iqn.2005-06.com.redhat:madmaxFor software iSCSI, each iface configuration must have a unique name (with less than 65 characters). The iface_name for network devices that support offloading appears in the format transport_name.hardware_name.
Example 25.7. iscsiadm -m iface Output with a Chelsio Network Card
For example, the sample output of iscsiadm -m iface on a system using a Chelsio network card might appear as:
default tcp,<empty>,<empty>,<empty>,<empty> iser iser,<empty>,<empty>,<empty>,<empty> cxgb3i.00:07:43:05:97:07 cxgb3i,00:07:43:05:97:07,<empty>,<empty>,<empty>It is also possible to display the settings of a specific iface configuration in a more friendly way. To do so, use the option -I iface_name. This will display the settings in the following format:
iface.setting = valueExample 25.8. Using iface Settings with a Chelsio Converged Network Adapter
Using the previous example, the iface settings of the same Chelsio converged network adapter (i.e. iscsiadm -m iface -I cxgb3i.00:07:43:05:97:07) would appear as:
# BEGIN RECORD 2.0-871 iface.iscsi_ifacename = cxgb3i.00:07:43:05:97:07 iface.net_ifacename = <empty> iface.ipaddress = <empty> iface.hwaddress = 00:07:43:05:97:07 iface.transport_name = cxgb3i iface.initiatorname = <empty> # END RECORD25.14.2. Configuring an iface for Software iSCSI
As mentioned earlier, an iface configuration is required for each network object that will be used to bind a session.
Before
To create an iface configuration for software iSCSI, run the following command:
# iscsiadm -m iface -I iface_name --op=newThis will create a new empty iface configuration with a specified iface_name. If an existing iface configuration already has the same iface_name, then it will be overwritten with a new, empty one.
To configure a specific setting of an iface configuration, use the following command:
# iscsiadm -m iface -I iface_name --op=update -n iface.setting -v hw_addressExample 25.9. Set MAC Address of iface0
For example, to set the MAC address (hardware_address) of iface0 to 00:0F:1F:92:6B:BF, run:
# iscsiadm -m iface -I iface0 --op=update -n iface.hwaddress -v 00:0F:1F:92:6B:BFDo not use default or iser as iface names. Both strings are special values used by iscsiadm for backward compatibility. Any manually-created iface configurations named default or iser will disable backwards compatibility.
25.14.3. Configuring an iface for iSCSI Offload
By default, iscsiadm creates an iface configuration for each port. To view available iface configurations, use the same command for doing so in software iSCSI: iscsiadm -m iface.
Before using the iface of a network card for iSCSI offload, first set the iface.ipaddress value of the offload interface to the initiator IP address that the interface should use:
For devices that use the be2iscsi driver, the IP address is configured in the BIOS setup screen.
For all other devices, to configure the IP address of the iface, use:
# iscsiadm -m iface -I iface_name -o update -n iface.ipaddress -v initiator_ip_address
Example 25.10. Set the iface IP Address of a Chelsio Card
For example, to set the iface IP address to 20.15.0.66 when using a card with the iface name of cxgb3i.00:07:43:05:97:07, use:
# iscsiadm -m iface -I cxgb3i.00:07:43:05:97:07 -o update -n iface.ipaddress -v 20.15.0.66
25.14.4. Binding/Unbinding an iface to a Portal
Whenever iscsiadm is used to scan for interconnects, it will first check the iface.transport settings of each iface configuration in /var/lib/iscsi/ifaces. The iscsiadm utility will then bind discovered portals to any iface whose iface.transport is tcp.
This behavior was implemented for compatibility reasons. To override this, use the -I iface_name to specify which portal to bind to an iface, as in:
# iscsiadm -m discovery -t st -p target_IP:port -I iface_name -P 1 [5]By default, the iscsiadm utility will not automatically bind any portals to iface configurations that use offloading. This is because such iface configurations will not have iface.transport set to tcp. As such, the iface configurations need to be manually bound to discovered portals.
It is also possible to prevent a portal from binding to any existing iface. To do so, use default as the iface_name, as in:
# iscsiadm -m discovery -t st -p IP:port -I default -P 1To remove the binding between a target and iface, use:
# iscsiadm -m node -targetname proper_target_name -I iface0 --op=delete[6]To delete all bindings for a specific iface, use:
# iscsiadm -m node -I iface_name --op=deleteTo delete bindings for a specific portal (e.g. for Equalogic targets), use:
# iscsiadm -m node -p IP:port -I iface_name --op=deleteIf there are no iface configurations defined in /var/lib/iscsi/iface and the -I option is not used, iscsiadm will allow the network subsystem to decide which device a specific portal should use.
25.15. Scanning iSCSI Interconnects
For iSCSI, if the targets send an iSCSI async event indicating new storage is added, then the scan is done automatically.
However, if the targets do not send an iSCSI async event, you need to manually scan them using the iscsiadm utility. Before doing so, however, you need to first retrieve the proper --targetname and the --portal values. If your device model supports only a single logical unit and portal per target, use iscsiadm to issue a sendtargets command to the host, as in:
# iscsiadm -m discovery -t sendtargets -p target_IP:port [5]The output will appear in the following format:
target_IP:port,target_portal_group_tag proper_target_nameExample 25.11. Using iscsiadm to issue a sendtargets Command
For example, on a target with a proper_target_name of iqn.1992-08.com.netapp:sn.33615311 and a target_IP:port of 10.15.85.19:3260, the output may appear as:
10.15.84.19:3260,2 iqn.1992-08.com.netapp:sn.33615311 10.15.85.19:3260,3 iqn.1992-08.com.netapp:sn.33615311In this example, the target has two portals, each using target_ip:ports of 10.15.84.19:3260 and 10.15.85.19:3260.
To see which iface configuration will be used for each session, add the -P 1 option. This option will print also session information in tree format, as in:
Target: proper_target_name Portal: target_IP:port,target_portal_group_tag Iface Name: iface_nameExample 25.12. View iface Configuration
For example, with iscsiadm -m discovery -t sendtargets -p 10.15.85.19:3260 -P 1, the output may appear as:
Target: iqn.1992-08.com.netapp:sn.33615311 Portal: 10.15.84.19:3260,2 Iface Name: iface2 Portal: 10.15.85.19:3260,3 Iface Name: iface2This means that the target iqn.1992-08.com.netapp:sn.33615311 will use iface2 as its iface configuration.
With some device models a single target may have multiple logical units and portals. In this case, issue a sendtargets command to the host first to find new portals on the target. Then, rescan the existing sessions using:
# iscsiadm -m session --rescanYou can also rescan a specific session by specifying the session's SID value, as in:
# iscsiadm -m session -r SID --rescan[7]If your device supports multiple targets, you will need to issue a sendtargets command to the hosts to find new portals for each target. Rescan existing sessions to discover new logical units on existing sessions using the --rescan option.
The sendtargets command used to retrieve --targetname and --portal values overwrites the contents of the /var/lib/iscsi/nodes database. This database will then be repopulated using the settings in /etc/iscsi/iscsid.conf. However, this will not occur if a session is currently logged in and in use.
To safely add new targets/portals or delete old ones, use the -o new or -o delete options, respectively. For example, to add new targets/portals without overwriting /var/lib/iscsi/nodes, use the following command:
iscsiadm -m discovery -t st -p target_IP -o newTo delete /var/lib/iscsi/nodes entries that the target did not display during discovery, use:
iscsiadm -m discovery -t st -p target_IP -o deleteYou can also perform both tasks simultaneously, as in:
iscsiadm -m discovery -t st -p target_IP -o delete -o newThe sendtargets command will yield the following output:
ip:port,target_portal_group_tag proper_target_nameExample 25.13. Output of the sendtargets Command
For example, given a device with a single target, logical unit, and portal, with equallogic-iscsi1 as your target_name, the output should appear similar to the following:
10.16.41.155:3260,0 iqn.2001-05.com.equallogic:6-8a0900-ac3fe0101-63aff113e344a4a2-dl585-03-1Note that proper_target_name and ip:port,target_portal_group_tag are identical to the values of the same name in Section 25.7.1, “iSCSI API”.
At this point, you now have the proper --targetname and --portal values needed to manually scan for iSCSI devices. To do so, run the following command:
# iscsiadm --mode node --targetname proper_target_name --portal ip:port,target_portal_group_tag \ --login [8]Example 25.14. Full iscsiadm Command
Using our previous example (where proper_target_name is equallogic-iscsi1), the full command would be:
# iscsiadm --mode node --targetname \ iqn.2001-05.com.equallogic:6-8a0900-ac3fe0101-63aff113e344a4a2-dl585-03-1 \ --portal 10.16.41.155:3260,0 --login[8]25.16. Logging in to an iSCSI Target
As mentioned in Section 25.7, “iSCSI”, the iSCSI service must be running in order to discover or log into targets. To start the iSCSI service, run:
# systemctl start iscsiWhen this command is executed, the iSCSI init scripts will automatically log into targets where the node.startup setting is configured as automatic. This is the default value of node.startup for all targets.
To prevent automatic login to a target, set node.startup to manual. To do this, run the following command:
# iscsiadm -m node --targetname proper_target_name -p target_IP:port -o update -n node.startup -v manualDeleting the entire record will also prevent automatic login. To do this, run:
# iscsiadm -m node --targetname proper_target_name -p target_IP:port -o deleteTo automatically mount a file system from an iSCSI device on the network, add a partition entry for the mount in /etc/fstab with the _netdev option. For example, to automatically mount the iSCSI device sdb to /mount/iscsi during startup, add the following line to /etc/fstab:
/dev/sdb /mnt/iscsi ext3 _netdev 0 0To manually log in to an iSCSI target, use the following command:
# iscsiadm -m node --targetname proper_target_name -p target_IP:port -l25.17. Resizing an Online Logical Unit
In most cases, fully resizing an online logical unit involves two things: resizing the logical unit itself and reflecting the size change in the corresponding multipath device (if multipathing is enabled on the system).
To resize the online logical unit, start by modifying the logical unit size through the array management interface of your storage device. This procedure differs with each array; as such, consult your storage array vendor documentation for more information on this.
In order to resize an online file system, the file system must not reside on a partitioned device.
25.17.1. Resizing Fibre Channel Logical Units
After modifying the online logical unit size, re-scan the logical unit to ensure that the system detects the updated size. To do this for Fibre Channel logical units, use the following command:
$ echo 1 > /sys/block/sdX/device/rescanTo re-scan Fibre Channel logical units on a system that uses multipathing, execute the aforementioned command for each sd device (i.e. sd1, sd2, and so on) that represents a path for the multipathed logical unit. To determine which devices are paths for a multipath logical unit, use multipath -ll; then, find the entry that matches the logical unit being resized. It is advisable that you refer to the WWID of each entry to make it easier to find which one matches the logical unit being resized.
25.17.2. Resizing an iSCSI Logical Unit
After modifying the online logical unit size, re-scan the logical unit to ensure that the system detects the updated size. To do this for iSCSI devices, use the following command:
# iscsiadm -m node --targetname target_name -R [5]Replace target_name with the name of the target where the device is located.
You can also re-scan iSCSI logical units using the following command:
# iscsiadm -m node -R -I interfaceReplace interface with the corresponding interface name of the resized logical unit (for example, iface0). This command performs two operations:
It re-scans for new/modified logical units the same way that the command echo 1 > /sys/block/sdX/device/rescan does. Note that this command is the same one used for re-scanning Fibre Channel logical units.
25.17.3. Updating the Size of Your Multipath Device
If multipathing is enabled on your system, you will also need to reflect the change in logical unit size to the logical unit's corresponding multipath device (after resizing the logical unit). This can be done through multipathd. To do so, first ensure that multipathd is running using service multipathd status. Once you've verified that multipathd is operational, run the following command:
# multipathd -k"resize map multipath_device"The multipath_device variable is the corresponding multipath entry of your device in /dev/mapper. Depending on how multipathing is set up on your system, multipath_device can be either of two formats:
mpathX, where X is the corresponding entry of your device (for example, mpath0)
a WWID; for example, 3600508b400105e210000900000490000
To determine which multipath entry corresponds to your resized logical unit, run multipath -ll. This displays a list of all existing multipath entries in the system, along with the major and minor numbers of their corresponding devices.
Do not use multipathd -k"resize map multipath_device" if there are any commands queued to multipath_device. That is, do not use this command when the no_path_retry parameter (in /etc/multipath.conf) is set to "queue", and there are no active paths to the device.
25.17.4. Changing the Read/Write State of an Online Logical Unit
Certain storage devices provide the user with the ability to change the state of the device from Read/Write (R/W) to Read-Only (RO), and from RO to R/W. This is typically done through a management interface on the storage device. The operating system will not automatically update its view of the state of the device when a change is made. Follow the procedures described in this chapter to make the operating system aware of the change.
Run the following command, replacing XYZ with the desired device designator, to determine the operating system's current view of the R/W state of a device:
# blockdev --getro /dev/sdXYZThe following command is also available for Red Hat Enterprise Linux 7:
# cat /sys/block/sdXYZ/ro 1 = read-only 0 = read-writeWhen using multipath, refer to the ro or rw field in the second line of output from the multipath -ll command. For example:
36001438005deb4710000500000640000 dm-8 GZ,GZ500 [size=20G][features=0][hwhandler=0][ro] \_ round-robin 0 [prio=200][active] \_ 6:0:4:1 sdax 67:16 [active][ready] \_ 6:0:5:1 sday 67:32 [active][ready] \_ round-robin 0 [prio=40][enabled] \_ 6:0:6:1 sdaz 67:48 [active][ready] \_ 6:0:7:1 sdba 67:64 [active][ready]To change the R/W state, use the following procedure:
Procedure 25.14. Change the R/W State
To move the device from RO to R/W, see step 2.
To move the device from R/W to RO, ensure no further writes will be issued. Do this by stopping the application, or through the use of an appropriate, application-specific action.
Ensure that all outstanding write I/Os are complete with the following command:
# blockdev --flushbufs /dev/deviceReplace device with the desired designator; for a device mapper multipath, this is the entry for your device in dev/mapper. For example, /dev/mapper/mpath3.
Use the management interface of the storage device to change the state of the logical unit from R/W to RO, or from RO to R/W. The procedure for this differs with each array. Consult applicable storage array vendor documentation for more information.
Perform a re-scan of the device to update the operating system's view of the R/W state of the device. If using a device mapper multipath, perform this re-scan for each path to the device before issuing the command telling multipath to reload its device maps.
25.17.4.1. Rescanning Logical Units
# echo 1 > /sys/block/sdX/device/rescanTo re-scan logical units on a system that uses multipathing, execute the above command for each sd device that represents a path for the multipathed logical unit. For example, run the command on sd1, sd2 and all other sd devices. To determine which devices are paths for a multipath unit, use multipath -11, then find the entry that matches the logical unit to be changed.
Example 25.15. Use of the multipath -11 Command
For example, the multipath -11 above shows the path for the LUN with WWID 36001438005deb4710000500000640000. In this case, enter:
# echo 1 > /sys/block/sdax/device/rescan # echo 1 > /sys/block/sday/device/rescan # echo 1 > /sys/block/sdaz/device/rescan # echo 1 > /sys/block/sdba/device/rescan25.17.4.2. Updating the R/W State of a Multipath Device
If multipathing is enabled, after rescanning the logical unit, the change in its state will need to be reflected in the logical unit's corresponding multipath drive. Do this by reloading the multipath device maps with the following command:
# multipath -rThe multipath -11 command can then be used to confirm the change.
25.17.4.3. Documentation
25.18. Adding/Removing a Logical Unit Through rescan-scsi-bus.sh
The sg3_utils package provides the rescan-scsi-bus.sh script, which can automatically update the logical unit configuration of the host as needed (after a device has been added to the system). The rescan-scsi-bus.sh script can also perform an issue_lip on supported devices. For more information about how to use this script, refer to rescan-scsi-bus.sh --help.
To install the sg3_utils package, run yum install sg3_utils.
Known Issues with rescan-scsi-bus.sh
When using the rescan-scsi-bus.sh script, take note of the following known issues:
In order for rescan-scsi-bus.sh to work properly, LUN0 must be the first mapped logical unit. The rescan-scsi-bus.sh can only detect the first mapped logical unit if it is LUN0. The rescan-scsi-bus.sh will not be able to scan any other logical unit unless it detects the first mapped logical unit even if you use the --nooptscan option.
A race condition requires that rescan-scsi-bus.sh be run twice if logical units are mapped for the first time. During the first scan, rescan-scsi-bus.sh only adds LUN0; all other logical units are added in the second scan.
A bug in the rescan-scsi-bus.sh script incorrectly executes the functionality for recognizing a change in logical unit size when the --remove option is used.
The rescan-scsi-bus.sh script does not recognize ISCSI logical unit removals.
25.19. Modifying Link Loss Behavior
This section describes how to modify the link loss behavior of devices that use either Fibre Channel or iSCSI protocols.
25.19.1. Fibre Channel
If a driver implements the Transport dev_loss_tmo callback, access attempts to a device through a link will be blocked when a transport problem is detected. To verify if a device is blocked, run the following command:
$ cat /sys/block/device/device/stateThis command will return blocked if the device is blocked. If the device is operating normally, this command will return running.
Procedure 25.15. Determining the State of a Remote Port
To determine the state of a remote port, run the following command:
$ cat /sys/class/fc_remote_port/rport-H:B:R/port_stateThis command will return Blocked when the remote port (along with devices accessed through it) are blocked. If the remote port is operating normally, the command will return Online.
If the problem is not resolved within dev_loss_tmo seconds, the rport and devices will be unblocked and all I/O running on that device (along with any new I/O sent to that device) will be failed.
Procedure 25.16. Changing dev_loss_tmo
To change the dev_loss_tmo value, echo in the desired value to the file. For example, to set dev_loss_tmo to 30 seconds, run:
$ echo 30 > /sys/class/fc_remote_port/rport-H:B:R/dev_loss_tmo
When a link loss exceeds dev_loss_tmo, the scsi_device and sdN devices are removed. Typically, the Fibre Channel class will leave the device as is; i.e. /dev/sdx will remain /dev/sdx. This is because the target binding is saved by the Fibre Channel driver so when the target port returns, the SCSI addresses are recreated faithfully. However, this cannot be guaranteed; the sdx will be restored only if no additional change on in-storage box configuration of LUNs is made.
25.19.2. iSCSI Settings with dm-multipath
If dm-multipath is implemented, it is advisable to set iSCSI timers to immediately defer commands to the multipath layer. To configure this, nest the following line under device { in /etc/multipath.conf:
features "1 queue_if_no_path"This ensures that I/O errors are retried and queued if all paths are failed in the dm-multipath layer.
You may need to adjust iSCSI timers further to better monitor your SAN for problems. Available iSCSI timers you can configure are NOP-Out Interval/Timeouts and replacement_timeout, which are discussed in the following sections.
25.19.2.1. NOP-Out Interval/Timeout
To help monitor problems the SAN, the iSCSI layer sends a NOP-Out request to each target. If a NOP-Out request times out, the iSCSI layer responds by failing any running commands and instructing the SCSI layer to requeue those commands when possible.
When dm-multipath is being used, the SCSI layer will fail those running commands and defer them to the multipath layer. The multipath layer then retries those commands on another path. If dm-multipath is not being used, those commands are retried five times before failing altogether.
Intervals between NOP-Out requests are 5 seconds by default. To adjust this, open /etc/iscsi/iscsid.conf and edit the following line:
node.conn[0].timeo.noop_out_interval = [interval value]Once set, the iSCSI layer will send a NOP-Out request to each target every [interval value] seconds.
By default, NOP-Out requests time out in 5 seconds[9]. To adjust this, open /etc/iscsi/iscsid.conf and edit the following line:
node.conn[0].timeo.noop_out_timeout = [timeout value]This sets the iSCSI layer to timeout a NOP-Out request after [timeout value] seconds.
SCSI Error HandlerIf the SCSI Error Handler is running, running commands on a path will not be failed immediately when a NOP-Out request times out on that path. Instead, those commands will be failed after replacement_timeout seconds. For more information about replacement_timeout, refer to Section 25.19.2.2, “replacement_timeout”.
To verify if the SCSI Error Handler is running, run:
# iscsiadm -m session -P 325.19.2.2. replacement_timeout
replacement_timeout controls how long the iSCSI layer should wait for a timed-out path/session to reestablish itself before failing any commands on it. The default replacement_timeout value is 120 seconds.
To adjust replacement_timeout, open /etc/iscsi/iscsid.conf and edit the following line:
node.session.timeo.replacement_timeout = [replacement_timeout]The 1 queue_if_no_path option in /etc/multipath.conf sets iSCSI timers to immediately defer commands to the multipath layer (refer to Section 25.19.2, “iSCSI Settings with dm-multipath”). This setting prevents I/O errors from propagating to the application; because of this, you can set replacement_timeout to 15-20 seconds.
By configuring a lower replacement_timeout, I/O is quickly sent to a new path and executed (in the event of a NOP-Out timeout) while the iSCSI layer attempts to re-establish the failed path/session. If all paths time out, then the multipath and device mapper layer will internally queue I/O based on the settings in /etc/multipath.conf instead of /etc/iscsi/iscsid.conf.
Whether your considerations are failover speed or security, the recommended value for replacement_timeout will depend on other factors. These factors include the network, target, and system workload. As such, it is recommended that you thoroughly test any new configurations to replacements_timeout before applying it to a mission-critical system.
iSCSI and DM Multipath overridesThe recovery_tmo sysfs option controls the timeout for a particular iSCSI device. The following options globally override recovery_tmo values:
The replacement_timeout configuration option globally overrides the recovery_tmo value for all iSCSI devices.
For all iSCSI devices that are managed by DM Multipath, the fast_io_fail_tmo option in DM Multipath globally overrides the recovery_tmo value. The fast_io_fail_tmo option in DM Multipath also overrides the fast_io_fail_tmo option in Fibre Channel devices.
The DM Multipath fast_io_fail_tmo option takes precedence over replacement_timeout. Red Hat does not recommend using replacement_timeout to override recovery_tmo in devices managed by DM Multipath because DM Multipath always resets recovery_tmo when the multipathd service reloads.
25.19.3. iSCSI Root
When accessing the root partition directly through an iSCSI disk, the iSCSI timers should be set so that iSCSI layer has several chances to try to reestablish a path/session. In addition, commands should not be quickly re-queued to the SCSI layer. This is the opposite of what should be done when dm-multipath is implemented.
To start with, NOP-Outs should be disabled. You can do this by setting both NOP-Out interval and timeout to zero. To set this, open /etc/iscsi/iscsid.conf and edit as follows:
node.conn[0].timeo.noop_out_interval = 0 node.conn[0].timeo.noop_out_timeout = 0In line with this, replacement_timeout should be set to a high number. This will instruct the system to wait a long time for a path/session to reestablish itself. To adjust replacement_timeout, open /etc/iscsi/iscsid.conf and edit the following line:
node.session.timeo.replacement_timeout = replacement_timeoutAfter configuring /etc/iscsi/iscsid.conf, you must perform a re-discovery of the affected storage. This will allow the system to load and use any new values in /etc/iscsi/iscsid.conf. For more information on how to discover iSCSI devices, refer to Section 25.15, “Scanning iSCSI Interconnects”.
Configuring Timeouts for a Specific Session
You can also configure timeouts for a specific session and make them non-persistent (instead of using /etc/iscsi/iscsid.conf). To do so, run the following command (replace the variables accordingly):
# iscsiadm -m node -T target_name -p target_IP:port -o update -n node.session.timeo.replacement_timeout -v $timeout_valueThe configuration described here is recommended for iSCSI sessions involving root partition access. For iSCSI sessions involving access to other types of storage (namely, in systems that use dm-multipath), refer to Section 25.19.2, “iSCSI Settings with dm-multipath”.
25.20. Controlling the SCSI Command Timer and Device Status
The Linux SCSI layer sets a timer on each command. When this timer expires, the SCSI layer will quiesce the host bus adapter (HBA) and wait for all outstanding commands to either time out or complete. Afterwards, the SCSI layer will activate the driver's error handler.
When the error handler is triggered, it attempts the following operations in order (until one successfully executes):
Abort the command.
Reset the device.
Reset the bus.
Reset the host.
If all of these operations fail, the device will be set to the offline state. When this occurs, all I/O to that device will be failed, until the problem is corrected and the user sets the device to running.
The process is different, however, if a device uses the Fibre Channel protocol and the rport is blocked. In such cases, the drivers wait for several seconds for the rport to become online again before activating the error handler. This prevents devices from becoming offline due to temporary transport problems.
Device States
To display the state of a device, use:
$ cat /sys/block/device-name/device/stateTo set a device to the running state, use:
# echo running > /sys/block/device-name/device/stateCommand Timer
To control the command timer, modify the /sys/block/device-name/device/timeout file:
# echo value > /sys/block/device-name/device/timeoutReplace value in the command with the timeout value, in seconds, that you want to implement.
25.21. Troubleshooting Online Storage Configuration
This section provides solution to common problems users experience during online storage reconfiguration.
Logical unit removal status is not reflected on the host.
When a logical unit is deleted on a configured filer, the change is not reflected on the host. In such cases, lvm commands will hang indefinitely when dm-multipath is used, as the logical unit has now become stale.
To work around this, perform the following procedure:
Procedure 25.17. Working Around Stale Logical Units
Determine which mpath link entries in /etc/lvm/cache/.cache are specific to the stale logical unit. To do this, run the following command:
$ ls -l /dev/mpath | grep stale-logical-unitExample 25.16. Determine Specific mpath Link Entries
For example, if stale-logical-unit is 3600d0230003414f30000203a7bc41a00, the following results may appear:
lrwxrwxrwx 1 root root 7 Aug 2 10:33 /3600d0230003414f30000203a7bc41a00 -> ../dm-4 lrwxrwxrwx 1 root root 7 Aug 2 10:33 /3600d0230003414f30000203a7bc41a00p1 -> ../dm-5This means that 3600d0230003414f30000203a7bc41a00 is mapped to two mpath links: dm-4 and dm-5.
Next, open /etc/lvm/cache/.cache. Delete all lines containing stale-logical-unit and the mpath links that stale-logical-unit maps to.
Example 25.17. Delete Relevant Lines
Using the same example in the previous step, the lines you need to delete are:
/dev/dm-4 /dev/dm-5 /dev/mapper/3600d0230003414f30000203a7bc41a00 /dev/mapper/3600d0230003414f30000203a7bc41a00p1 /dev/mpath/3600d0230003414f30000203a7bc41a00 /dev/mpath/3600d0230003414f30000203a7bc41a00p1
25.22. Configuring Maximum Time for Error Recovery with eh_deadline
In most scenarios, you do not need to enable the eh_deadline parameter. Using the eh_deadline parameter can be useful in certain specific scenarios, for example if a link loss occurs between a Fibre Channel switch and a target port, and the Host Bus Adapter (HBA) does not receive Registered State Change Notifications (RSCNs). In such a case, I/O requests and error recovery commands all time out rather than encounter an error. Setting eh_deadline in this environment puts an upper limit on the recovery time, which enables the failed I/O to be retried on another available path by multipath.
However, if RSCNs are enabled, the HBA does not register the link becoming unavailable, or both, the eh_deadline functionality provides no additional benefit, as the I/O and error recovery commands fail immediately, which allows multipath to retry.
The SCSI host object eh_deadline parameter enables you to configure the maximum amount of time that the SCSI error handling mechanism attempts to perform error recovery before stopping and resetting the entire HBA.
The value of the eh_deadline is specified in seconds. The default setting is off, which disables the time limit and allows all of the error recovery to take place. In addition to using sysfs, a default value can be set for all SCSI HBAs by using the scsi_mod.eh_deadline kernel parameter.
Note that when eh_deadline expires, the HBA is reset, which affects all target paths on that HBA, not only the failing one. As a consequence, I/O errors can occur if some of the redundant paths are not available for other reasons. Enable eh_deadline only if you have a fully redundant multipath configuration on all targets.
Chapter 26. Device Mapper Multipathing (DM Multipath) and Storage for Virtual Machines
Red Hat Enterprise Linux 7 supports DM Multipath and storage for virtual machines.
26.1. Storage for Virtual Machines
Red Hat Enterprise Linux 7 supports the following file systems or online storage methods for virtual storage:
Fibre Channel
iSCSI
NFS
GFS2
Virtualization in Red Hat Enterprise Linux 7 uses libvirt to manage virtual instances. The libvirt utility uses the concept of storage pools to manage storage for virtualized guests. A storage pool is a storage that can be divided into smaller volumes or allocated directly to a guest. Volumes of a storage pool can be allocated to virtualized guests. Following are the categories of available storage pools:
Local storage pools
Local storage includes storage devices, files or directories directly attached to a host, local directories, directly attached disks, and LVM Volume Groups.
Networked (shared) storage poolsNetworked storage covers storage devices shared over a network using standard protocols. It includes shared storage devices using Fibre Channel, iSCSI, NFS, GFS2, and SCSI RDMA protocols, and is a requirement for migrating virtualized guests between hosts.
26.2. DM Multipath
DM Multipath is a feature that allows you to configure multiple I/O paths between server nodes and storage arrays into a single device. These I/O paths are physical SAN connections that can include separate cables, switches, and controllers. Multipathing aggregates the I/O paths, creating a new device that consists of the aggregated paths.
DM Multipath are used primarily for the following reasons:
Redundancy
DM Multipath can provide failover in an active/passive configuration. In an active/passive configuration, only half the paths are used at any time for I/O. If any element of an I/O path fails, DM Multipath switches to an alternate path.
Improved PerformanceDM Multipath can be configured in active/active mode, where I/O is spread over the paths in a round-robin fashion. In some configurations, DM Multipath can detect loading on the I/O paths and dynamically rebalance the load.
For more information, see the Red Hat DM Multipath guide.
Chapter 27. External Array Management (libStorageMgmt)
Red Hat Enterprise Linux 7 ships with a new external array management library called libStorageMgmt.
27.1. Introduction to libStorageMgmt
The libStorageMgmt library is a storage array independent Application Programming Interface (API). As a developer, you can use this API to manage different storage arrays and leverage the hardware accelerated features.
This library is used as a building block for other higher level management tools and applications. End system administrators can also use it as a tool to manually manage storage and automate storage management tasks with the use of scripts.
With the libStorageMgmt library, you can perform the following operations:
List storage pools, volumes, access groups, or file systems.
Create and delete volumes, access groups, file systems, or NFS exports.
Grant and remove access to volumes, access groups, or initiators.
Replicate volumes with snapshots, clones, and copies.
Create and delete access groups and edit members of a group.
Server resources such as CPU and interconnect bandwidth are not utilized because the operations are all done on the array.
The libstoragemgmt package provides:
A stable C and Python API for client application and plug-in developers.
A command-line interface that utilizes the library (lsmcli).
A daemon that executes the plug-in (lsmd).
A simulator plug-in that allows the testing of client applications (sim).
Plug-in architecture for interfacing with arrays.
This library and its associated tool have the ability to destroy any and all data located on the arrays it manages. It is highly recommended to develop and test applications and scripts against the storage simulator plug-in to remove any logic errors before working with production systems. Testing applications and scripts on actual non-production hardware before deploying to production is also strongly encouraged if possible.
The libStorageMgmt library in Red Hat Enterprise Linux 7 adds a default udev rule to handle the REPORTED LUNS DATA HAS CHANGED unit attention.
When a storage configuration change has taken place, one of several Unit Attention ASC/ASCQ codes reports the change. A uevent is then generated and is rescanned automatically with sysfs.
The file /lib/udev/rules.d/90-scsi-ua.rules contains example rules to enumerate other events that the kernel can generate.
The libStorageMgmt library uses a plug-in architecture to accommodate differences in storage arrays. For more information on libStorageMgmt plug-ins and how to write them, see the Red Hat Developer Guide.
27.2. libStorageMgmt Terminology
Different array vendors and storage standards use different terminology to refer to similar functionality. This library uses the following terminology.
Storage array
Any storage system that provides block access (FC, FCoE, iSCSI) or file access through Network Attached Storage (NAS).
VolumeStorage Area Network (SAN) Storage Arrays can expose a volume to the Host Bus Adapter (HBA) over different transports, such as FC, iSCSI, or FCoE. The host OS treats it as block devices. One volume can be exposed to many disks if multipath[2] is enabled).
This is also known as the Logical Unit Number (LUN), StorageVolume with SNIA terminology, or virtual disk.
PoolA group of storage spaces. File systems or volumes can be created from a pool. Pools can be created from disks, volumes, and other pools. A pool may also hold RAID settings or thin provisioning settings.
This is also known as a StoragePool with SNIA Terminology.
SnapshotA point in time, read only, space efficient copy of data.
This is also known as a read only snapshot.
CloneA point in time, read writeable, space efficient copy of data.
This is also known as a read writeable snapshot.
CopyA full bitwise copy of the data. It occupies the full space.
MirrorA continuously updated copy (synchronous and asynchronous).
Access groupCollections of iSCSI, FC, and FCoE initiators which are granted access to one or more storage volumes. This ensures that only storage volumes are accessible by the specified initiators.
This is also known as an initiator group.
Access GrantExposing a volume to a specified access group or initiator. The libStorageMgmt library currently does not support LUN mapping with the ability to choose a specific logical unit number. The libStorageMgmt library allows the storage array to select the next available LUN for assignment. If configuring a boot from SAN or masking more than 256 volumes be sure to read the OS, Storage Array, or HBA documents.
Access grant is also known as LUN Masking.
SystemRepresents a storage array or a direct attached storage RAID.
File systemA Network Attached Storage (NAS) storage array can expose a file system to host an OS through an IP network, using either NFS or CIFS protocol. The host OS treats it as a mount point or a folder containing files depending on the client operating system.
DiskThe physical disk holding the data. This is normally used when creating a pool with RAID settings.
This is also known as a DiskDrive using SNIA Terminology.
InitiatorIn Fibre Channel (FC) or Fibre Channel over Ethernet (FCoE), the initiator is the World Wide Port Name (WWPN) or World Wide Node Name (WWNN). In iSCSI, the initiator is the iSCSI Qualified Name (IQN). In NFS or CIFS, the initiator is the host name or the IP address of the host.
Child dependencySome arrays have an implicit relationship between the origin (parent volume or file system) and the child (such as a snapshot or a clone). For example, it is impossible to delete the parent if it has one or more depend children. The API provides methods to determine if any such relationship exists and a method to remove the dependency by replicating the required blocks.
27.3. Installing libStorageMgmt
To install libStorageMgmt for use of the command line, required run-time libraries and simulator plug-ins, use the following command:
# yum install libstoragemgmt libstoragemgmt-pythonTo develop C applications that utilize the library, install the libstoragemgmt-devel package with the following command:
# yum install libstoragemgmt-develTo install libStorageMgmt for use with hardware arrays, select one or more of the appropriate plug-in packages with the following command:
# yum install libstoragemgmt-name-pluginThe following plug-ins that are available include:
libstoragemgmt-smis-plugin
Generic SMI-S array support.
libstoragemgmt-netapp-pluginSpecific support for NetApp files.
libstoragemgmt-nstor-pluginSpecific support for NexentaStor.
libstoragemgmt-targetd-pluginSpecific support for targetd.
The daemon is then installed and configured to run at start after the next reboot. To use it immediately without rebooting, start the daemon manually.
Managing an array requires support through a plug-in. The base install package includes open source plug-ins for a number of different vendors. Additional plug-in packages will be available separately as array support improves. Currently supported hardware is constantly changing and improving.
The libStorageMgmt daemon (lsmd) behaves like any standard service for the system.
To check the status of the libStorageMgmt service:
# systemctl status libstoragemgmtTo stop the service:
# systemctl stop libstoragemgmtTo start the service:
# systemctl start libstoragemgmt27.4. Using libStorageMgmt
To use libStorageMgmt interactively, use the lsmcli tool.
The lsmcli tool requires two things to run:
A Uniform Resource Identifier (URI) which is used to identify the plug-in to connect to the array and any configurable options the array requires.
A valid user name and password for the array.
URI has the following form:
plugin+optional-transport://user-name@host:port/?query-string-parameters
Each plug-in has different requirements for what is needed.
Example 27.1. Examples of Different Plug-in Requirements
Simulator Plug-in That Requires No User Name or Password
sim://
NetApp Plug-in over SSL with User Name root
ontap+ssl://root@filer.company.com/
SMI-S Plug-in over SSL for EMC Array
smis+ssl://admin@provider.com:5989/?namespace=root/emc
There are three options to use the URI:
Pass the URI as part of the command.
$ lsmcli -u sim://Store the URI in an environmental variable.
$ export LSMCLI_URI=sim://Place the URI in the file ~/.lsmcli, which contains name-value pairs separated by "=". The only currently supported configuration is 'uri'.
Determining which URI to use needs to be done in this order. If all three are supplied, only the first one on the command line will be used.
Provide the password by specifying the -P option on the command line or by placing it in an environmental variable LSMCLI_PASSWORD.
Example 27.2. Example of lsmcli
An example for using the command line to create a new volume and making it visible to an initiator.
List arrays that are serviced by this connection:
$ lsmcli list --type SYSTEMS ID | Name | Status -------+-------------------------------+-------- sim-01 | LSM simulated storage plug-in | OKList storage pools:
$ lsmcli list --type POOLS -H ID | Name | Total space | Free space | System ID -----+---------------+----------------------+----------------------+----------- POO2 | Pool 2 | 18446744073709551616 | 18446744073709551616 | sim-01 POO3 | Pool 3 | 18446744073709551616 | 18446744073709551616 | sim-01 POO1 | Pool 1 | 18446744073709551616 | 18446744073709551616 | sim-01 POO4 | lsm_test_aggr | 18446744073709551616 | 18446744073709551616 | sim-01Create a volume:
$ lsmcli volume-create --name volume_name --size 20G --pool POO1 -H ID | Name | vpd83 | bs | #blocks | status | ... -----+-------------+----------------------------------+-----+----------+--------+---- Vol1 | volume_name | F7DDF7CA945C66238F593BC38137BD2F | 512 | 41943040 | OK | ...Create an access group with an iSCSI initiator in it:
$ lsmcli --create-access-group example_ag --id iqn.1994-05.com.domain:01.89bd01 --type ISCSI --system sim-01 ID | Name | Initiator ID |SystemID ---------------------------------+------------+----------------------------------+-------- 782d00c8ac63819d6cca7069282e03a0 | example_ag | iqn.1994-05.com.domain:01.89bd01 |sim-01Create an access group with an iSCSI intiator in it:
$ lsmcli access-group-create --name example_ag --init iqn.1994-05.com.domain:01.89bd01 --init-type ISCSI --sys sim-01 ID | Name | Initiator IDs | System ID ---------------------------------+------------+----------------------------------+----------- 782d00c8ac63819d6cca7069282e03a0 | example_ag | iqn.1994-05.com.domain:01.89bd01 | sim-01Allow the access group visibility to the newly created volume:
$ lsmcli access-group-grant --ag 782d00c8ac63819d6cca7069282e03a0 --vol Vol1 --access RWThe design of the library provides for a process separation between the client and the plug-in by means of inter-process communication (IPC). This prevents bugs in the plug-in from crashing the client application. It also provides a means for plug-in writers to write plug-ins with a license of their own choosing. When a client opens the library passing a URI, the client library looks at the URI to determine which plug-in should be used.
The plug-ins are technically stand alone applications but they are designed to have a file descriptor passed to them on the command line. The client library then opens the appropriate Unix domain socket which causes the daemon to fork and execute the plug-in. This gives the client library a point to point communication channel with the plug-in. The daemon can be restarted without affecting existing clients. While the client has the library open for that plug-in, the plug-in process is running. After one or more commands are sent and the plug-in is closed, the plug-in process cleans up and then exits.
The default behavior of lsmcli is to wait until the operation is complete. Depending on the requested operations, this could potentially could take many hours. To allow a return to normal usage, it is possible to use the -b option on the command line. If the exit code is 0 the command is completed. If the exit code is 7 the command is in progress and a job identifier is written to standard output. The user or script can then take the job ID and query the status of the command as needed by using lsmcli --jobstatus JobID. If the job is now completed, the exit value will be 0 and the results printed to standard output. If the command is still in progress, the return value will be 7 and the percentage complete will be printed to the standard output.
Example 27.3. An Asynchronous Example
Create a volume passing the -b option so that the command returns immediately.
$ lsmcli volume-create --name async_created --size 20G --pool POO1 -b JOB_3Check the exit value:
$ echo $? 77 indicates that the job is still in progress.
Check if the job is completed:
$ lsmcli job-status --job JOB_3 33Check the exit value. 7 indicates the job is still in progress so the standard output is the percentage done or 33% based on the given screen.
$ echo $? 7Wait for sometime and check the exit value again:
$ lsmcli job-status --job JOB_3 ID | Name | vpd83 | Block Size | ... -----+---------------+----------------------------------+-------------+----- Vol2 | async_created | 855C9BA51991B0CC122A3791996F6B15 | 512 | ...0 means success and standard out displays the new volume.
For scripting, pass the -t SeparatorCharacters option. This will make it easier to parse the output.
Example 27.4. Scripting Examples
$ lsmcli list --type volumes -t# Vol1#volume_name#049167B5D09EC0A173E92A63F6C3EA2A#512#41943040#21474836480#OK#sim-01#POO1 Vol2#async_created#3E771A2E807F68A32FA5E15C235B60CC#512#41943040#21474836480#OK#sim-01#POO1 $ lsmcli list --type volumes -t " | " Vol1 | volume_name | 049167B5D09EC0A173E92A63F6C3EA2A | 512 | 41943040 | 21474836480 | OK | 21474836480 | sim-01 | POO1 Vol2 | async_created | 3E771A2E807F68A32FA5E15C235B60CC | 512 | 41943040 | 21474836480 | OK | sim-01 | POO1 $ lsmcli list --type volumes -s --------------------------------------------- ID | Vol1 Name | volume_name VPD83 | 049167B5D09EC0A173E92A63F6C3EA2A Block Size | 512 #blocks | 41943040 Size | 21474836480 Status | OK System ID | sim-01 Pool ID | POO1 --------------------------------------------- ID | Vol2 Name | async_created VPD83 | 3E771A2E807F68A32FA5E15C235B60CC Block Size | 512 #blocks | 41943040 Size | 21474836480 Status | OK System ID | sim-01 Pool ID | POO1 ---------------------------------------------
It is recommended to use the Python library for non-trivial scripting.
For more information on lsmcli, see the lsmcli man page or lsmcli --help.
Chapter 28. Persistent Memory: NVDIMMs
Persistent memory (pmem), also called as storage class memory, is a combination of memory and storage. pmem combines the durability of storage with the low access latency and the high bandwidth of dynamic RAM (DRAM):
Persistent memory is byte-addressable, so it can be accessed by using CPU load and store instructions. In addition to read() or write() system calls that are required for accessing traditional block-based storage, pmem also supports direct load and store programming model.
The performance characteristics of persistent memory are similar to DRAM with very low access latency, typically in the tens to hundreds of nanoseconds.
Contents of persistent memory are preserved when the power is off, like with storage.
Using persistent memory is beneficial for use cases like:
Rapid start: data set is already in memory.Rapid start is also called the warm cache effect. A file server has none of the file contents in memory after starting. As clients connect and read and write data, that data is cached in the page cache. Eventually, the cache contains mostly hot data. After a reboot, the system must start the process again.
Persistent memory allows an application to keep the warm cache across reboots if the application is designed properly. In this instance, there would be no page cache involved: the application would cache data directly in the persistent memory.
Fast write-cacheFile servers often do not acknowledge a client's write request until the data is on durable media. Using persistent memory as a fast write cache enables a file server to acknowledge the write request quickly thanks to the low latency of pmem.
NVDIMMs Interleaving
Non-Volatile Dual In-line Memory Modules (NVDIMMs) can be grouped into interleave sets in the same way as regular DRAM. An interleave set is like a RAID 0 (stripe) across multiple DIMMs.
Following are the advantages of NVDIMMS interleaving:
Like DRAM, NVDIMMs benefit from increased performance when they are configured into interleave sets.
It can be used to combine multiple smaller NVDIMMs into one larger logical device.
Use the system BIOS or UEFI firmware to configure interleave sets.
In Linux, one region device is created per interleave set.
Following is the relation between region devices and labels:
If your NVDIMMs support labels, the region device can be further subdivided into namespaces.
If your NVDIMMs do not support labels, the region devices can only contain a single namespace. In this case, the kernel creates a default namespace which covers the entire region.
Persistent Memory Access Modes
You can use persistent memory devices in the sector, fsdax, devdax (device direct access) or raw mode:
sector mode
It presents the storage as a fast block device. Using sector mode is useful for legacy applications that have not been modified to use persistent memory, or for applications that make use of the full I/O stack, including the Device Mapper.
fsdax modeIt enables persistent memory devices to support direct access programming as described in the Storage Networking Industry Association (SNIA) Non-Volatile Memory (NVM) Programming Model specification. In this mode, I/O bypasses the storage stack of the kernel, and many Device Mapper drivers therefore cannot be used.
devdax modeThe devdax (device DAX) mode provides raw access to persistent memory by using a DAX character device node. Data on a devdax device can be made durable using CPU cache flushing and fencing instructions. Certain databases and virtual machine hypervisors might benefit from the devdax mode. File systems cannot be created on device devdax instances.
raw modeThe raw mode namespaces have several limitations and should not be used.
28.1. Configuring Persistent Memory with ndctl
Use the ndctl utility to configure persistent memory devices. To install ndctl utility, use the following command:
# yum install ndctlProcedure 28.1. Configuring Persistent Memory for a device that does not support labels
List the available pmem regions on your system. In the following example, the command lists an NVDIMM-N device that does not support labels:
# ndctl list --regions [ { "dev":"region1", "size":34359738368, "available_size":0, "type":"pmem" }, { "dev":"region0", "size":34359738368, "available_size":0, "type":"pmem" } ]Red Hat Enterprise Linux creates a default namespace for each region because the NVDIMM-N device here does not support labels. Hence, the available size is 0 bytes.
List all the inactive namespaces on your system:
# ndctl list --namespaces --idle [ { "dev":"namespace1.0", "mode":"raw", "size":34359738368, "state":"disabled", "numa_node":1 }, { "dev":"namespace0.0", "mode":"raw", "size":34359738368, "state":"disabled", "numa_node":0 } ]Reconfigure the inactive namespaces in order to make use of this space. For example, to use namespace0.0 for a file system that supports DAX, use the following command:
# ndctl create-namespace --force --reconfig=namespace0.0 --mode=fsdax --map=mem { "dev":"namespace0.0", "mode":"fsdax", "size":"32.00 GiB (34.36 GB)", "uuid":"ab91cc8f-4c3e-482e-a86f-78d177ac655d", "blockdev":"pmem0", "numa_node":0 }
Procedure 28.2. Configuring Persistent Memory for a device that supports labels
List the available pmem regions on your system. In the following example, the command lists an NVDIMM-N device that supports labels:
# ndctl list --regions [ { "dev":"region5", "size":270582939648, "available_size":270582939648, "type":"pmem", "iset_id":-7337419320239190016 }, { "dev":"region4", "size":270582939648, "available_size":270582939648, "type":"pmem", "iset_id":-137289417188962304 } ]If the NVDIMM device supports labels, default namespaces are not created, and you can allocate one or more namespaces from a region without using the --force or --reconfigure flags:
# ndctl create-namespace --region=region4 --mode=fsdax --map=dev --size=36G { "dev":"namespace4.0", "mode":"fsdax", "size":"35.44 GiB (38.05 GB)", "uuid":"9c5330b5-dc90-4f7a-bccd-5b558fa881fe", "blockdev":"pmem4", "numa_node":0 }Now, you can create another namespace from the same region:
# ndctl create-namespace --region=region4 --mode=fsdax --map=dev --size=36G { "dev":"namespace4.1", "mode":"fsdax", "size":"35.44 GiB (38.05 GB)", "uuid":"91868e21-830c-4b8f-a472-353bf482a26d", "blockdev":"pmem4.1", "numa_node":0 }You can also create namespaces of different types from the same region, using the following command:
# ndctl create-namespace --region=region4 --mode=devdax --align=2M --size=36G { "dev":"namespace4.2", "mode":"devdax", "size":"35.44 GiB (38.05 GB)", "uuid":"a188c847-4153-4477-81bb-7143e32ffc5c", "daxregion": { "id":4, "size":"35.44 GiB (38.05 GB)", "align":2097152, "devices":[ { "chardev":"dax4.2", "size":"35.44 GiB (38.05 GB)" }] }, "numa_node":0 }
For more information on ndctl utility, see man ndctl.
28.2. Configuring Persistent Memory for Use as a Block Device (Legacy Mode)
To use persistent memory as a fast block device, set the namespace to sector mode.
# ndctl create-namespace --force --reconfig=namespace1.0 --mode=sector { "dev":"namespace1.0", "mode":"sector", "size":17162027008, "uuid":"029caa76-7be3-4439-8890-9c2e374bcc76", "sector_size":4096, "blockdev":"pmem1s" }In the example, namespace1.0 is reconfigured to sector mode. Note that the block device name changed from pmem1 to pmem1s. This device can be used in the same way as any other block device on the system. For example, the device can be partitioned, you can create a file system on the device, the device can be configured as part of a software RAID set, and the device can be the cache device for dm-cache.
28.3. Configuring Persistent Memory for File System Direct Access
File system direct access requires the namespace to be configured to the fsdax mode. This mode allows for the direct access programming model. When a device is configured in the fsdax mode, a file system can be created on top of it, and then mounted with the -o fsdax mount option. Then, any application that performs an mmap() operation on a file on this file system gets direct access to its storage. See the following example:
# ndctl create-namespace --force --reconfig=namespace0.0 --mode=fsdax --map=mem { "dev":"namespace0.0", "mode":"fsdax", "size":17177772032, "uuid":"e6944638-46aa-4e06-a722-0b3f16a5acbf", "blockdev":"pmem0" }In the example, namespace0.0 is converted to namespace fsdax mode. With the --map=mem argument, ndctl puts operating system data structures used for Direct Memory Access (DMA) in system DRAM.
To perform DMA, the kernel requires a data structure for each page in the memory region. The overhead of this data structure is 64 bytes per 4-KiB page. For small devices, the amount of overhead is small enough to fit in DRAM with no problems. For example, the 16-GiB namespace only requires 256MiB for page structures. Because NVDIMM devices are usually small and expensive, storing the kernel’s page tracking data structures in DRAM is preferable, as indicated by the --map=mem parameter.
In the future, NVDIMM devices might be terabytes in size. For such devices, the amount of memory required to store the page tracking data structures might exceed the amount of DRAM in the system. One TiB of persistent memory requires 16 GiB just for page structures. As a result, specifying the --map=dev parameter to store the data structures in the persistent memory itself is preferable in such cases.
After configuring the namespace in the fsdax mode, the namespace is ready for a file system. Starting with Red Hat Enterprise Linux 7.3, both the Ext4 and XFS file system enable using persistent memory as a Technology Preview. File system creation requires no special arguments. To get the DAX functionality, mount the file system with the dax mount option. For example:
# mkfs -t xfs /dev/pmem0 # mount -o dax /dev/pmem0 /mnt/pmem/Then, applications can use persistent memory and create files in the /mnt/pmem/ directory, open the files, and use the mmap operation to map the files for direct access.
When creating partitions on a pmem device to be used for direct access, partitions must be aligned on page boundaries. On the Intel 64 and AMD64 architecture, at least 4KiB alignment for the start and end of the partition, but 2MiB is the preferred alignment. By default, the parted tool aligns partitions on 1MiB boundaries. For the first partition, specify 2MiB as the start of the partition. If the size of the partition is a multiple of 2MiB, all other partitions are also aligned.
28.4. Configuring Persistent Memory for use in Device DAX mode
Device DAX (devdax) provides a means for applications to directly access storage, without the involvement of a file system. The benefit of device DAX is that it provides a guaranteed fault granularity, which can be configured using the --align option with the ndctl utility:
# ndctl create-namespace --force --reconfig=namespace0.0 --mode=devdax --align=2MThe given command ensures that the operating system would fault in 2MiB pages at a time. For the Intel 64 and AMD64 architecture, the following fault granularities are supported:
4KiB
2MiB
1GiB
Device DAX nodes (/dev/daxN.M) only supports the following system call:
open()
close()
mmap()
fallocate()
read() and write() variants are not supported because the use case is tied to persistent memory programming.
28.5. Troubleshooting NVDIMM
28.5.1. Monitoring NVDIMM Health Using S.M.A.R.T.
Some NVDIMMs support Self-Monitoring, Analysis and Reporting Technology (S.M.A.R.T.) interfaces for retrieving health information.
Prerequisites
On some systems, the acpi_ipmi driver must be loaded to retrieve health information using the following command:
# modprobe acpi_ipmi
Procedure
To access the health information, use the following command:
# ndctl list --dimms --health ... { "dev":"nmem0", "id":"802c-01-1513-b3009166", "handle":1, "phys_id":22, "health": { "health_state":"ok", "temperature_celsius":25.000000, "spares_percentage":99, "alarm_temperature":false, "alarm_spares":false, "temperature_threshold":50.000000, "spares_threshold":20, "life_used_percentage":1, "shutdown_state":"clean" } } ...
28.5.2. Detecting and Replacing a Broken NVDIMM
If you find error messages related to NVDIMM reported in your system log or by S.M.A.R.T., it might mean an NVDIMM device is failing. In that case, it is necessary to:
Detect which NVDIMM device is failing,
Back up data stored on it, and
Physically replace the device.
Procedure 28.3. Detecting and Replacing a Broken NVDIMM
To detect the broken DIMM, use the following command:
# ndctl list --dimms --regions --health --media-errors --humanThe badblocks field shows which NVDIMM is broken. Note its name in the dev field. In the following example, the NVDIMM named nmem0 is broken:
Example 28.1. Health Status of NVDIMM Devices
# ndctl list --dimms --regions --health --media-errors --human ... "regions":[ { "dev":"region0", "size":"250.00 GiB (268.44 GB)", "available_size":0, "type":"pmem", "numa_node":0, "iset_id":"0xXXXXXXXXXXXXXXXX", "mappings":[ { "dimm":"nmem1", "offset":"0x10000000", "length":"0x1f40000000", "position":1 }, { "dimm":"nmem0", "offset":"0x10000000", "length":"0x1f40000000", "position":0 } ], "badblock_count":1, "badblocks":[ { "offset":65536, "length":1, "dimms":[ "nmem0" ] } ], "persistence_domain":"memory_controller" } ] }
Use the following command to find the phys_id attribute of the broken NVDIMM:
# ndctl list --dimms --humanFrom the previous example, you know that nmem0 is the broken NVDIMM. Therefore, find the phys_id attribute of nmem0. In the following example, the phys_id is 0x10:
Example 28.2. The phys_id Attributes of NVDIMMs
# ndctl list --dimms --human [ { "dev":"nmem1", "id":"XXXX-XX-XXXX-XXXXXXXX", "handle":"0x120", "phys_id":"0x1c" }, { "dev":"nmem0", "id":"XXXX-XX-XXXX-XXXXXXXX", "handle":"0x20", "phys_id":"0x10", "flag_failed_flush":true, "flag_smart_event":true } ]
Use the following command to find the memory slot of the broken NVDIMM:
# dmidecodeIn the output, find the entry where the Handle identifier matches the phys_id attribute of the broken NVDIMM. The Locator field lists the memory slot used by the broken NVDIMM. In the following example, the nmem0 device matches the 0x0010 identifier and uses the DIMM-XXX-YYYY memory slot:
Example 28.3. NVDIMM Memory Slot Listing
# dmidecode ... Handle 0x0010, DMI type 17, 40 bytes Memory Device Array Handle: 0x0004 Error Information Handle: Not Provided Total Width: 72 bits Data Width: 64 bits Size: 125 GB Form Factor: DIMM Set: 1 Locator: DIMM-XXX-YYYY Bank Locator: Bank0 Type: Other Type Detail: Non-Volatile Registered (Buffered) ...
Back up all data in the namespaces on the NVDIMM. If you do not back up the data before replacing the NVDIMM, the data will be lost when you remove the NVDIMM from your system.
In some cases, such as when the NVDIMM is completely broken, the backup might fail.
Use the following command to list the namespaces on the NVDIMM:
# ndctl list --namespaces --dimm=DIMM-ID-numberIn the following example, the nmem0 device contains the namespace0.0 and namespace0.2 namespaces, which you need to back up:
Example 28.4. NVDIMM Namespaces Listing
# ndctl list --namespaces --dimm=0 [ { "dev":"namespace0.2", "mode":"sector", "size":67042312192, "uuid":"XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", "raw_uuid":"XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", "sector_size":4096, "blockdev":"pmem0.2s", "numa_node":0 }, { "dev":"namespace0.0", "mode":"sector", "size":67042312192, "uuid":"XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", "raw_uuid":"XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", "sector_size":4096, "blockdev":"pmem0s", "numa_node":0 } ]
Replace the broken NVDIMM physically.
Chapter 29. Overview of NVMe over fabric devices
Non-volatile Memory Express (NVMe) is an interface that allows host software utility to communicate with solid state drives. Use the following types of fabric transport to configure NVMe over fabric devices:
When using FC and RDMA, the solid-state drive does not have to be local to your system; it can be configured remotely through a FC or RDMA controller.
29.1. NVMe over fabrics using RDMA
The following sections describe how to deploy an NVMe over RDMA (NVMe/RDMA) initiator configuration.
29.1.1. Configuring an NVMe over RDMA client
Use this procedure to configure an NVMe/RDMA client using the NVMe management command line interface (nvme-cli).
Install the nvme-cli package:
# yum install nvme-cliLoad the nvme-rdma module if it is not loaded:
# modprobe nvme-rdmaDiscover available subsystems on the NVMe target:
# nvme discover -t rdma -a 172.31.0.202 -s 4420 Discovery Log Number of Records 1, Generation counter 2 =====Discovery Log Entry 0====== trtype: rdma adrfam: ipv4 subtype: nvme subsystem treq: not specified, sq flow control disable supported portid: 1 trsvcid: 4420 subnqn: testnqn traddr: 172.31.0.202 rdma_prtype: not specified rdma_qptype: connected rdma_cms: rdma-cm rdma_pkey: 0x0000Connect to the discovered subsystems:
# nvme connect -t rdma -n testnqn -a 172.31.0.202 -s 4420 # lsblk NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT sda 8:0 0 465.8G 0 disk ├─sda1 8:1 0 1G 0 part /boot └─sda2 8:2 0 464.8G 0 part ├─rhel_rdma--virt--03-root 253:0 0 50G 0 lvm / ├─rhel_rdma--virt--03-swap 253:1 0 4G 0 lvm [SWAP] └─rhel_rdma--virt--03-home 253:2 0 410.8G 0 lvm /home nvme0n1 # cat /sys/class/nvme/nvme0/transport rdmaReplace testnqn with the NVMe subsystem name.
Replace 172.31.0.202 with the target IP address.
Replace 4420 with the port number.
List the NVMe devices that are currently connected:
# nvme listOptional: Disconnect from the target:
# nvme disconnect -n testnqn NQN:testnqn disconnected 1 controller(s) # lsblk NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT sda 8:0 0 465.8G 0 disk ├─sda1 8:1 0 1G 0 part /boot └─sda2 8:2 0 464.8G 0 part ├─rhel_rdma--virt--03-root 253:0 0 50G 0 lvm / ├─rhel_rdma--virt--03-swap 253:1 0 4G 0 lvm [SWAP] └─rhel_rdma--virt--03-home 253:2 0 410.8G 0 lvm /home
29.2. NVMe over fabrics using FC
The NVMe over Fibre Channel (FC-NVMe) is fully supported in initiator mode when used with certain Broadcom Emulex and Marvell Qlogic Fibre Channel adapters. As a system administrator, complete the tasks in the following sections to deploy the FC-NVMe:
29.2.2. Configuring the NVMe initiator for QLogic adapters
Use this procedure to configure NVMe initiator for Qlogic adapters client using the NVMe management command line interface (nvme-cli) tool.
Install the nvme-cli tool:
# yum install nvme-cliThis creates the hostnqn file in the /etc/nvme/ directory. The hostnqn file identifies the NVMe host. To generate a new hostnqn:
# nvme gen-hostnqnRemove and reload the qla2xxx module:
# rmmod qla2xxx # modprobe qla2xxxFind the WWNN and WWPN of the local and remote ports:
# dmesg |grep traddr [ 6.139862] qla2xxx [0000:04:00.0]-ffff:0: register_localport: host-traddr=nn-0x20000024ff19bb62:pn-0x21000024ff19bb62 on portID:10700 [ 6.241762] qla2xxx [0000:04:00.0]-2102:0: qla_nvme_register_remote: traddr=nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6 PortID:01050dUsing this host-traddr and traddr, find the subsystem NQN:
# nvme discover --transport fc --traddr nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6 --host-traddr nn-0x20000024ff19bb62:pn-0x21000024ff19bb62 Discovery Log Number of Records 2, Generation counter 49530 =====Discovery Log Entry 0====== trtype: fc adrfam: fibre-channel subtype: nvme subsystem treq: not specified portid: 0 trsvcid: none subnqn: nqn.1992-08.com.netapp:sn.c9ecc9187b1111e98c0800a098cbcac6:subsystem.vs_nvme_multipath_1_subsystem_468 traddr: nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6Replace nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6 with the traddr.
Replace nn-0x20000024ff19bb62:pn-0x21000024ff19bb62 with the host_traddr.
Connect to the NVMe target using the nvme-cli tool:
# nvme connect --transport fc --traddr nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6 --host_traddr nn-0x20000024ff19bb62:pn-0x21000024ff19bb62 -n nqn.1992-08.com.netapp:sn.c9ecc9187b1111e98c0800a098cbcac6:subsystem.vs_nvme_multipath_1_subsystem_468Replace nn-0x203b00a098cbcac6:pn-0x203d00a098cbcac6 with the traddr.
Replace nn-0x20000024ff19bb62:pn-0x21000024ff19bb62 with the host_traddr.
Replace nqn.1992-08.com.netapp:sn.c9ecc9187b1111e98c0800a098cbcac6:subsystem.vs_nvme_multipath_1_subsystem_468 with the subnqn.
Verify the NVMe devices are currently connected:
# nvme list Node SN Model Namespace Usage Format FW Rev ---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- -------- /dev/nvme0n1 80BgLFM7xMJbAAAAAAAC NetApp ONTAP Controller 1 107.37 GB / 107.37 GB 4 KiB + 0 B FFFFFFFF # lsblk |grep nvme nvme0n1 259:0 0 100G 0 disk
Part III. Data Deduplication and Compression with VDO
This part describes how to provide deduplicated block storage capabilities to existing storage management applications by enabling them to utilize Virtual Data Optimizer (VDO).
Chapter 30. VDO Integration
30.1. Theoretical Overview of VDO
Virtual Data Optimizer (VDO) is a block virtualization technology that allows you to easily create compressed and deduplicated pools of block storage.
Deduplication is a technique for reducing the consumption of storage resources by eliminating multiple copies of duplicate blocks.
Instead of writing the same data more than once, VDO detects each duplicate block and records it as a reference to the original block. VDO maintains a mapping from logical block addresses, which are used by the storage layer above VDO, to physical block addresses, which are used by the storage layer under VDO.
After deduplication, multiple logical block addresses may be mapped to the same physical block address; these are called shared blocks. Block sharing is invisible to users of the storage, who read and write blocks as they would if VDO were not present. When a shared block is overwritten, a new physical block is allocated for storing the new block data to ensure that other logical block addresses that are mapped to the shared physical block are not modified.
Compression is a data-reduction technique that works well with file formats that do not necessarily exhibit block-level redundancy, such as log files and databases. See Section 30.4.8, “Using Compression” for more detail.
The VDO solution consists of the following components:
kvdo
A kernel module that loads into the Linux Device Mapper layer to provide a deduplicated, compressed, and thinly provisioned block storage volume
udsA kernel module that communicates with the Universal Deduplication Service (UDS) index on the volume and analyzes data for duplicates.
Command line toolsFor configuring and managing optimized storage.
30.1.1. The UDS Kernel Module (uds)
The UDS index provides the foundation of the VDO product. For each new piece of data, it quickly determines if that piece is identical to any previously stored piece of data. If the index finds match, the storage system can then internally reference the existing item to avoid storing the same information more than once.
The UDS index runs inside the kernel as the uds kernel module.
30.1.2. The VDO Kernel Module (kvdo)
The kvdo Linux kernel module provides block-layer deduplication services within the Linux Device Mapper layer. In the Linux kernel, Device Mapper serves as a generic framework for managing pools of block storage, allowing the insertion of block-processing modules into the storage stack between the kernel's block interface and the actual storage device drivers.
The kvdo module is exposed as a block device that can be accessed directly for block storage or presented through one of the many available Linux file systems, such as XFS or ext4. When kvdo receives a request to read a (logical) block of data from a VDO volume, it maps the requested logical block to the underlying physical block and then reads and returns the requested data.
When kvdo receives a request to write a block of data to a VDO volume, it first checks whether it is a DISCARD or TRIM request or whether the data is uniformly zero. If either of these conditions holds, kvdo updates its block map and acknowledges the request. Otherwise, a physical block is allocated for use by the request.
Overview of VDO Write Policies
If the kvdo module is operating in synchronous mode:
It temporarily writes the data in the request to the allocated block and then acknowledges the request.
Once the acknowledgment is complete, an attempt is made to deduplicate the block by computing a MurmurHash-3 signature of the block data, which is sent to the VDO index.
If the VDO index contains an entry for a block with the same signature, kvdo reads the indicated block and does a byte-by-byte comparison of the two blocks to verify that they are identical.
If they are indeed identical, then kvdo updates its block map so that the logical block points to the corresponding physical block and releases the allocated physical block.
If the VDO index did not contain an entry for the signature of the block being written, or the indicated block does not actually contain the same data, kvdo updates its block map to make the temporary physical block permanent.
If kvdo is operating in asynchronous mode:
Instead of writing the data, it will immediately acknowledge the request.
It will then attempt to deduplicate the block in same manner as described above.
If the block turns out to be a duplicate, kvdo will update its block map and release the allocated block. Otherwise, it will write the data in the request to the allocated block and update the block map to make the physical block permanent.
30.1.3. VDO Volume
VDO uses a block device as a backing store, which can include an aggregation of physical storage consisting of one or more disks, partitions, or even flat files. When a VDO volume is created by a storage management tool, VDO reserves space from the volume for both a UDS index and the VDO volume, which interact together to provide deduplicated block storage to users and applications. Figure 30.1, “VDO Disk Organization” illustrates how these pieces fit together.
Figure 30.1. VDO Disk Organization
Slabs
The physical storage of the VDO volume is divided into a number of slabs, each of which is a contiguous region of the physical space. All of the slabs for a given volume will be of the same size, which may be any power of 2 multiple of 128 MB up to 32 GB.
The default slab size is 2 GB in order to facilitate evaluating VDO on smaller test systems. A single VDO volume may have up to 8192 slabs. Therefore, in the default configuration with 2 GB slabs, the maximum allowed physical storage is 16 TB. When using 32 GB slabs, the maximum allowed physical storage is 256 TB. At least one entire slab is reserved by VDO for metadata, and therefore cannot be used for storing user data.
Slab size has no effect on the performance of the VDO volume.
Table 30.1. Recommended VDO Slab Sizes by Physical Volume Size
10–99 GB | 1 GB |
100 GB – 1 TB | 2 GB |
2–256 TB | 32 GB |
The size of a slab can be controlled by providing the --vdoSlabSize=megabytes option to the vdo create command.
Physical Size and Available Physical Size
Both physical size and available physical size describe the amount of disk space on the block device that VDO can utilize:
Physical size is the same size as the underlying block device. VDO uses this storage for:
User data, which might be deduplicated and compressed
VDO metadata, such as the UDS index
Available physical size is the portion of the physical size that VDO is able to use for user data.
It is equivalent to the physical size minus the size of the metadata, minus the remainder after dividing the volume into slabs by the given slab size.
Logical Size
If the --vdoLogicalSize option is not specified, the logical volume size defaults to the available physical volume size. Note that, in Figure 30.1, “VDO Disk Organization”, the VDO deduplicated storage target sits completely on top of the block device, meaning the physical size of the VDO volume is the same size as the underlying block device.
VDO currently supports any logical size up to 254 times the size of the physical volume with an absolute maximum logical size of 4PB.
30.1.4. Command Line Tools
VDO includes the following command line tools for configuration and management:
vdo
Creates, configures, and controls VDO volumes
vdostatsProvides utilization and performance statistics
30.2. System Requirements
Processor Architectures
One or more processors implementing the Intel 64 instruction set are required: that is, a processor of the AMD64 or Intel 64 architecture.
RAM
Each VDO volume has two distinct memory requirements:
The VDO module requires 370 MB plus an additional 268 MB per each 1 TB of physical storage managed.
The Universal Deduplication Service (UDS) index requires a minimum of 250 MB of DRAM, which is also the default amount that deduplication uses. For details on the memory usage of UDS, see Section 30.2.1, “UDS Index Memory Requirements”.
Storage
A VDO volume is a thinly provisioned block device. To prevent running out of physical space, place the volume on top of storage that you can expand at a later time. Examples of such expandable storage are LVM volumes or MD RAID arrays.
A single VDO volume can be configured to use up to 256 TB of physical storage. See Section 30.2.2, “VDO Storage Space Requirements” for the calculations to determine the usable size of a VDO-managed volume from the physical size of the storage pool the VDO is given.
Additional System Software
VDO depends on the following software:
LVM
Python 2.7
The yum package manager will install all necessary software dependencies automatically.
Placement of VDO in the Storage Stack
As a general rule, you should place certain storage layers under VDO and others on top of VDO:
-
Under VDO: DM-Multipath, DM-Crypt, and software RAID (LVM or mdraid).
On top of VDO: LVM cache, LVM snapshots, and LVM Thin Provisioning.
The following configurations are not supported:
VDO on top of VDO volumes: storage → VDO → LVM → VDO
VDO on top of LVM Snapshots
VDO on top of LVM Cache
VDO on top of the loopback device
VDO on top of LVM Thin Provisioning
Encrypted volumes on top of VDO: storage → VDO → DM-Crypt
Partitions on a VDO volume: fdisk, parted, and similar partitions
RAID (LVM, MD, or any other type) on top of a VDO volume
VDO supports two write modes: sync and async. When VDO is in sync mode, writes to the VDO device are acknowledged when the underlying storage has written the data permanently. When VDO is in async mode, writes are acknowledged before being written to persistent storage.
It is critical to set the VDO write policy to match the behavior of the underlying storage. By default, VDO write policy is set to the auto option, which selects the appropriate policy automatically.
30.2.1. UDS Index Memory Requirements
The UDS index consists of two parts:
A compact representation is used in memory that contains at most one entry per unique block.
An on-disk component which records the associated block names presented to the index as they occur, in order.
UDS uses an average of 4 bytes per entry in memory (including cache).
The on-disk component maintains a bounded history of data passed to UDS. UDS provides deduplication advice for data that falls within this deduplication window, containing the names of the most recently seen blocks. The deduplication window allows UDS to index data as efficiently as possible while limiting the amount of memory required to index large data repositories. Despite the bounded nature of the deduplication window, most datasets which have high levels of deduplication also exhibit a high degree of temporal locality — in other words, most deduplication occurs among sets of blocks that were written at about the same time. Furthermore, in general, data being written is more likely to duplicate data that was recently written than data that was written a long time ago. Therefore, for a given workload over a given time interval, deduplication rates will often be the same whether UDS indexes only the most recent data or all the data.
Because duplicate data tends to exhibit temporal locality, it is rarely necessary to index every block in the storage system. Were this not so, the cost of index memory would outstrip the savings of reduced storage costs from deduplication. Index size requirements are more closely related to the rate of data ingestion. For example, consider a storage system with 100 TB of total capacity but with an ingestion rate of 1 TB per week. With a deduplication window of 4 TB, UDS can detect most redundancy among the data written within the last month.
UDS's Sparse Indexing feature (the recommended mode for VDO) further exploits temporal locality by attempting to retain only the most relevant index entries in memory. UDS can maintain a deduplication window that is ten times larger while using the same amount of memory. While the sparse index provides the greatest coverage, the dense index provides more advice. For most workloads, given the same amount of memory, the difference in deduplication rates between dense and sparse indexes is negligible.
The memory required for the index is determined by the desired size of the deduplication window:
For a dense index, UDS will provide a deduplication window of 1 TB per 1 GB of RAM. A 1 GB index is generally sufficient for storage systems of up to 4 TB.
For a sparse index, UDS will provide a deduplication window of 10 TB per 1 GB of RAM. A 1 GB sparse index is generally sufficient for up to 40 TB of physical storage.
30.2.2. VDO Storage Space Requirements
VDO requires storage space both for VDO metadata and for the actual UDS deduplication index:
VDO writes two types of metadata to its underlying physical storage:
The first type scales with the physical size of the VDO volume and uses approximately 1 MB for each 4 GB of physical storage plus an additional 1 MB per slab.
The second type scales with the logical size of the VDO volume and consumes approximately 1.25 MB for each 1 GB of logical storage, rounded up to the nearest slab.
The UDS index is stored within the VDO volume group and is managed by the associated VDO instance. The amount of storage required depends on the type of index and the amount of RAM allocated to the index. For each 1 GB of RAM, a dense UDS index will use 17 GB of storage, and a sparse UDS index will use 170 GB of storage.
30.2.3. Examples of VDO System Requirements by Physical Volume Size
The following tables provide approximate system requirements of VDO based on the size of the underlying physical volume. Each table lists requirements appropriate to the intended deployment, such as primary storage or backup storage.
The exact numbers depend on your configuration of the VDO volume.
Primary Storage Deployment
In the primary storage case, the UDS index is between 0.01% to 25% the size of the physical volume.
Table 30.2. VDO Storage and Memory Requirements for Primary Storage
RAM Usage | 250 MB | Dense: 1 GB Sparse: 250 MB | 2 GB | 3 GB | 12 GB |
Disk Usage | 2.5 GB | Dense: 10 GB Sparse: 22 GB | 170 GB | 255 GB | 1020 GB |
Index Type | Dense | Dense or Sparse | Sparse | Sparse | Sparse |
Backup Storage Deployment
In the backup storage case, the UDS index covers the size of the backup set but is not bigger than the physical volume. If you expect the backup set or the physical size to grow in the future, factor this into the index size.
Table 30.3. VDO Storage and Memory Requirements for Backup Storage
RAM Usage | 250 MB | 2 GB | 10 GB | 20 GB | 26 GB |
Disk Usage | 2.5 GB | 170 GB | 850 GB | 1700 GB | 3400 GB |
Index Type | Dense | Sparse | Sparse | Sparse | Sparse |
30.3. Getting Started with VDO
30.3.1. Introduction
Virtual Data Optimizer (VDO) provides inline data reduction for Linux in the form of deduplication, compression, and thin provisioning. When you set up a VDO volume, you specify a block device on which to construct your VDO volume and the amount of logical storage you plan to present.
When hosting active VMs or containers, Red Hat recommends provisioning storage at a 10:1 logical to physical ratio: that is, if you are utilizing 1 TB of physical storage, you would present it as 10 TB of logical storage.
For object storage, such as the type provided by Ceph, Red Hat recommends using a 3:1 logical to physical ratio: that is, 1 TB of physical storage would present as 3 TB logical storage.
In either case, you can simply put a file system on top of the logical device presented by VDO and then use it directly or as part of a distributed cloud storage architecture.
This chapter describes the following use cases of VDO deployment:
the direct-attached use case for virtualization servers, such as those built using Red Hat Virtualization, and
the cloud storage use case for object-based distributed storage clusters, such as those built using Ceph Storage.
VDO deployment with Ceph is currently not supported.
This chapter provides examples for configuring VDO for use with a standard Linux file system that can be easily deployed for either use case; see the diagrams in Section 30.3.5, “Deployment Examples”.
30.3.2. Installing VDO
VDO is deployed using the following RPM packages:
vdo
kmod-kvdo
To install VDO, use the yum package manager to install the RPM packages:
# yum install vdo kmod-kvdo30.3.3. Creating a VDO Volume
Create a VDO volume for your block device. Note that multiple VDO volumes can be created for separate devices on the same machine. If you choose this approach, you must supply a different name and device for each instance of VDO on the system.
In all the following steps, replace vdo_name with the identifier you want to use for your VDO volume; for example, vdo1.
Create the VDO volume using the VDO Manager:
# vdo create \ --name=vdo_name \ --device=block_device \ --vdoLogicalSize=logical_size \ [--vdoSlabSize=slab_size]Replace block_device with the persistent name of the block device where you want to create the VDO volume. For example, /dev/disk/by-id/scsi-3600508b1001c264ad2af21e903ad031f.
Use a persistent device name. If you use a non-persistent device name, then VDO might fail to start properly in the future if the device name changes.
Replace logical_size with the amount of logical storage that the VDO volume should present:
For active VMs or container storage, use logical size that is ten times the physical size of your block device. For example, if your block device is 1 TB in size, use 10T here.
For object storage, use logical size that is three times the physical size of your block device. For example, if your block device is 1 TB in size, use 3T here.
If the block device is larger than 16 TiB, add the --vdoSlabSize=32G to increase the slab size on the volume to 32 GiB.
Using the default slab size of 2 GiB on block devices larger than 16 TiB results in the vdo create command failing with the following error:
vdo: ERROR - vdoformat: formatVDO failed on '/dev/device': VDO Status: Exceeds maximum number of slabs supported
Example 30.1. Creating VDO for Container Storage
For example, to create a VDO volume for container storage on a 1 TB block device, you might use:
# vdo create \ --name=vdo1 \ --device=/dev/disk/by-id/scsi-3600508b1001c264ad2af21e903ad031f \ --vdoLogicalSize=10TWhen a VDO volume is created, VDO adds an entry to the /etc/vdoconf.yml configuration file. The vdo.service systemd unit then uses the entry to start the volume by default.
Create a file system:
For the XFS file system:
# mkfs.xfs -K /dev/mapper/vdo_nameFor the ext4 file system:
# mkfs.ext4 -E nodiscard /dev/mapper/vdo_name
Mount the file system:
# mkdir -m 1777 /mnt/vdo_name # mount /dev/mapper/vdo_name /mnt/vdo_nameTo configure the file system to mount automatically, use either the /etc/fstab file or a systemd mount unit:
If you decide to use the /etc/fstab configuration file, add one of the following lines to the file:
For the XFS file system:
/dev/mapper/vdo_name /mnt/vdo_name xfs defaults,_netdev,x-systemd.device-timeout=0,x-systemd.requires=vdo.service 0 0For the ext4 file system:
/dev/mapper/vdo_name /mnt/vdo_name ext4 defaults,_netdev,x-systemd.device-timeout=0,x-systemd.requires=vdo.service 0 0
Alternatively, if you decide to use a systemd unit, create a systemd mount unit file with the appropriate filename. For the mount point of your VDO volume, create the /etc/systemd/system/mnt-vdo_name.mount file with the following content:
[Unit] Description = VDO unit file to mount file system name = vdo_name.mount Requires = vdo.service After = multi-user.target Conflicts = umount.target [Mount] What = /dev/mapper/vdo_name Where = /mnt/vdo_name Type = xfs [Install] WantedBy = multi-user.targetAn example systemd unit file is also installed at /usr/share/doc/vdo/examples/systemd/VDO.mount.example.
Enable the discard feature for the file system on your VDO device. Both batch and online operations work with VDO.
30.3.4. Monitoring VDO
Because VDO is thin provisioned, the file system and applications will only see the logical space in use and will not be aware of the actual physical space available.
VDO space usage and efficiency can be monitored using the vdostats utility:
# vdostats --human-readable Device 1K-blocks Used Available Use% Space saving% /dev/mapper/node1osd1 926.5G 21.0G 905.5G 2% 73% /dev/mapper/node1osd2 926.5G 28.2G 898.3G 3% 64%When the physical storage capacity of a VDO volume is almost full, VDO reports a warning in the system log, similar to the following:
Oct 2 17:13:39 system lvm[13863]: Monitoring VDO pool vdo_name. Oct 2 17:27:39 system lvm[13863]: WARNING: VDO pool vdo_name is now 80.69% full. Oct 2 17:28:19 system lvm[13863]: WARNING: VDO pool vdo_name is now 85.25% full. Oct 2 17:29:39 system lvm[13863]: WARNING: VDO pool vdo_name is now 90.64% full. Oct 2 17:30:29 system lvm[13863]: WARNING: VDO pool vdo_name is now 96.07% full.Monitor physical space on your VDO volumes to prevent out-of-space situations. Running out of physical blocks might result in losing recently written, unacknowledged data on the VDO volume.
30.3.5. Deployment Examples
The following examples illustrate how VDO might be used in KVM and other deployments.
VDO Deployment with KVM
Figure 30.2. VDO Deployment with KVM
More Deployment Scenarios
30.4. Administering VDO
30.4.1. Starting or Stopping VDO
To start a given VDO volume, or all VDO volumes, and the associated UDS index(es), storage management utilities should invoke one of these commands:
# vdo start --name=my_vdo # vdo start --allTo stop a given VDO volume, or all VDO volumes, and the associated UDS index(es), use one of these commands:
# vdo stop --name=my_vdo # vdo stop --allStopping a VDO volume takes time based on the speed of your storage device and the amount of data that the volume needs to write:
The volume always writes around 1GiB for every 1GiB of the UDS index.
With a sparse UDS index, the volume additionally writes the amount of data equal to the block map cache size plus up to 8MiB per slab.
If restarted after an unclean shutdown, VDO will perform a rebuild to verify the consistency of its metadata and will repair it if necessary. Rebuilds are automatic and do not require user intervention. See Section 30.4.5, “Recovering a VDO Volume After an Unclean Shutdown” for more information on the rebuild process.
VDO might rebuild different writes depending on the write mode:
In synchronous mode, all writes that were acknowledged by VDO prior to the shutdown will be rebuilt.
In asynchronous mode, all writes that were acknowledged prior to the last acknowledged flush request will be rebuilt.
In either mode, some writes that were either unacknowledged or not followed by a flush may also be rebuilt.
30.4.2. Selecting VDO Write Modes
VDO supports three write modes, sync, async, and auto:
When VDO is in sync mode, the layers above it assume that a write command writes data to persistent storage. As a result, it is not necessary for the file system or application, for example, to issue FLUSH or Force Unit Access (FUA) requests to cause the data to become persistent at critical points.
VDO must be set to sync mode only when the underlying storage guarantees that data is written to persistent storage when the write command completes. That is, the storage must either have no volatile write cache, or have a write through cache.
When VDO is in async mode, the data is not guaranteed to be written to persistent storage when a write command is acknowledged. The file system or application must issue FLUSH or FUA requests to ensure data persistence at critical points in each transaction.
VDO must be set to async mode if the underlying storage does not guarantee that data is written to persistent storage when the write command completes; that is, when the storage has a volatile write back cache.
When VDO is running in async mode, it is not compliant with Atomicity, Consistency, Isolation, Durability (ACID). When there is an application or a file system that assumes ACID compliance on top of the VDO volume, async mode might cause unexpected data loss.
The auto mode automatically selects sync or async based on the characteristics of each device. This is the default option.
To set a write policy, use the --writePolicy option. This can be specified either when creating a VDO volume as in Section 30.3.3, “Creating a VDO Volume” or when modifying an existing VDO volume with the changeWritePolicy subcommand:
# vdo changeWritePolicy --writePolicy=sync|async|auto --name=vdo_nameUsing the incorrect write policy might result in data loss on power failure.
Checking for a Volatile Cache
To see whether a device has a writeback cache, read the /sys/block/block_device/device/scsi_disk/identifier/cache_type sysfs file. For example:
Device sda indicates that it has a writeback cache:
$ cat '/sys/block/sda/device/scsi_disk/7:0:0:0/cache_type' write backDevice sdb indicates that it does not have a writeback cache:
$ cat '/sys/block/sdb/device/scsi_disk/1:2:0:0/cache_type' None
Additionally, in the kernel boot log, you can find whether the above mentioned devices have a write cache or not:
sd 7:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA sd 1:2:0:0: [sdb] Write cache: disabled, read cache: disabled, supports DPO and FUASee the Viewing and Managing Log Files chapter in the System Administrator's Guide for more information on reading the system log.
In these examples, use the following write policies for VDO:
async mode for the sda device
sync mode for the sdb device
You should configure VDO to use the sync write policy if the cache_type value is none or write through.
30.4.3. Removing VDO Volumes
A VDO volume can be removed from the system by running:
# vdo remove --name=my_vdoPrior to removing a VDO volume, unmount file systems and stop applications that are using the storage. The vdo remove command removes the VDO volume and its associated UDS index, as well as logical volumes where they reside.
30.4.3.1. Removing an Unsuccessfully Created Volume
If a failure occurs when the vdo utility is creating a VDO volume, the volume is left in an intermediate state. This might happen when, for example, the system crashes, power fails, or the administrator interrupts a running vdo create command.
To clean up from this situation, remove the unsuccessfully created volume with the --force option:
# vdo remove --force --name=my_vdoThe --force option is required because the administrator might have caused a conflict by changing the system configuration since the volume was unsuccessfully created. Without the --force option, the vdo remove command fails with the following message:
[...] A previous operation failed. Recovery from the failure either failed or was interrupted. Add '--force' to 'remove' to perform the following cleanup. Steps to clean up VDO my_vdo: umount -f /dev/mapper/my_vdo udevadm settle dmsetup remove my_vdo vdo: ERROR - VDO volume my_vdo previous operation (create) is incomplete30.4.4. Configuring the UDS Index
VDO uses a high-performance deduplication index called UDS to detect duplicate blocks of data as they are being stored. The deduplication window is the number of previously written blocks which the index remembers. The size of the deduplication window is configurable. For a given window size, the index will requires a specific amount of RAM and a specific amount of disk space. The size of the window is usually determined by specifying the size of the index memory using the --indexMem=size option. The amount of disk space to use will then be determined automatically.
In general, Red Hat recommends using a sparse UDS index for all production use cases. This is an extremely efficient indexing data structure, requiring approximately one-tenth of a byte of DRAM per block in its deduplication window. On disk, it requires approximately 72 bytes of disk space per block. The minimum configuration of this index uses 256 MB of DRAM and approximately 25 GB of space on disk. To use this configuration, specify the --sparseIndex=enabled --indexMem=0.25 options to the vdo create command. This configuration results in a deduplication window of 2.5 TB (meaning it will remember a history of 2.5 TB). For most use cases, a deduplication window of 2.5 TB is appropriate for deduplicating storage pools that are up to 10 TB in size.
The default configuration of the index, however, is to use a dense index. This index is considerably less efficient (by a factor of 10) in DRAM, but it has much lower (also by a factor of 10) minimum required disk space, making it more convenient for evaluation in constrained environments.
In general, a deduplication window which is one quarter of the physical size of a VDO volume is a recommended configuration. However, this is not an actual requirement. Even small deduplication windows (compared to the amount of physical storage) can find significant amounts of duplicate data in many use cases. Larger windows may also be used, but it in most cases, there will be little additional benefit to doing so.
Speak with your Red Hat Technical Account Manager representative for additional guidelines on tuning this important system parameter.
30.4.5. Recovering a VDO Volume After an Unclean Shutdown
If a volume is restarted without having been shut down cleanly, VDO will need to rebuild a portion of its metadata to continue operating, which occurs automatically when the volume is started. (Also see Section 30.4.5.2, “Forcing a Rebuild” to invoke this process on a volume that was cleanly shut down.)
Data recovery depends on the write policy of the device:
If VDO was running on synchronous storage and write policy was set to sync, then all data written to the volume will be fully recovered.
If the write policy was async, then some writes may not be recovered if they were not made durable by sending VDO a FLUSH command, or a write I/O tagged with the FUA flag (force unit access). This is accomplished from user mode by invoking a data integrity operation like fsync, fdatasync, sync, or umount.
30.4.5.1. Online Recovery
In the majority of cases, most of the work of rebuilding an unclean VDO volume can be done after the VDO volume has come back online and while it is servicing read and write requests. Initially, the amount of space available for write requests may be limited. As more of the volume's metadata is recovered, more free space may become available. Furthermore, data written while the VDO is recovering may fail to deduplicate against data written before the crash if that data is in a portion of the volume which has not yet been recovered. Data may be compressed while the volume is being recovered. Previously compressed blocks may still be read or overwritten.
During an online recovery, a number of statistics will be unavailable: for example, blocks in use and blocks free. These statistics will become available once the rebuild is complete.
30.4.5.2. Forcing a Rebuild
VDO can recover from most hardware and software errors. If a VDO volume cannot be recovered successfully, it is placed in a read-only mode that persists across volume restarts. Once a volume is in read-only mode, there is no guarantee that data has not been lost or corrupted. In such cases, Red Hat recommends copying the data out of the read-only volume and possibly restoring the volume from backup. (The operating mode attribute of vdostats indicates whether a VDO volume is in read-only mode.)
If the risk of data corruption is acceptable, it is possible to force an offline rebuild of the VDO volume metadata so the volume can be brought back online and made available. Again, the integrity of the rebuilt data cannot be guaranteed.
To force a rebuild of a read-only VDO volume, first stop the volume if it is running:
# vdo stop --name=my_vdoThen restart the volume using the --forceRebuild option:
# vdo start --name=my_vdo --forceRebuild30.4.6. Automatically Starting VDO Volumes at System Boot
During system boot, the vdo systemd unit automatically starts all VDO devices that are configured as activated.
To prevent certain existing volumes from being started automatically, deactivate those volumes by running either of these commands:
To deactivate a specific volume:
# vdo deactivate --name=my_vdoTo deactivate all volumes:
# vdo deactivate --all
Conversely, to activate volumes, use one of these commands:
To activate a specific volume:
# vdo activate --name=my_vdoTo activate all volumes:
# vdo activate --all
You can also create a VDO volume that does not start automatically by adding the --activate=disabled option to the vdo create command.
For systems that place LVM volumes on top of VDO volumes as well as beneath them (for example, Figure 30.5, “Deduplicated Unified Storage”), it is vital to start services in the right order:
The lower layer of LVM must be started first (in most systems, starting this layer is configured automatically when the LVM2 package is installed).
The vdo systemd unit must then be started.
Finally, additional scripts must be run in order to start LVM volumes or other services on top of the now running VDO volumes.
30.4.7. Disabling and Re-enabling Deduplication
In some instances, it may be desirable to temporarily disable deduplication of data being written to a VDO volume while still retaining the ability to read to and write from the volume. While disabling deduplication will prevent subsequent writes from being deduplicated, data which was already deduplicated will remain so.
To stop deduplication on a VDO volume, use the following command:
# vdo disableDeduplication --name=my_vdoThis stops the associated UDS index and informs the VDO volume that deduplication is no longer active.
To restart deduplication on a VDO volume, use the following command:
# vdo enableDeduplication --name=my_vdoThis restarts the associated UDS index and informs the VDO volume that deduplication is active again.
You can also disable deduplication when creating a new VDO volume by adding the --deduplication=disabled option to the vdo create command.
30.4.8. Using Compression
30.4.8.1. Introduction
In addition to block-level deduplication, VDO also provides inline block-level compression using the HIOPS Compression™ technology. While deduplication is the optimal solution for virtual machine environments and backup applications, compression works very well with structured and unstructured file formats that do not typically exhibit block-level redundancy, such as log files and databases.
Compression operates on blocks that have not been identified as duplicates. When unique data is seen for the first time, it is compressed. Subsequent copies of data that have already been stored are deduplicated without requiring an additional compression step. The compression feature is based on a parallelized packaging algorithm that enables it to handle many compression operations at once. After first storing the block and responding to the requestor, a best-fit packing algorithm finds multiple blocks that, when compressed, can fit into a single physical block. After it is determined that a particular physical block is unlikely to hold additional compressed blocks, it is written to storage and the uncompressed blocks are freed and reused. By performing the compression and packaging operations after having already responded to the requestor, using compression imposes a minimal latency penalty.
30.4.8.2. Enabling and Disabling Compression
VDO volume compression is on by default.
When creating a volume, you can disable compression by adding the --compression=disabled option to the vdo create command.
Compression can be stopped on an existing VDO volume if necessary to maximize performance or to speed processing of data that is unlikely to compress.
To stop compression on a VDO volume, use the following command:
# vdo disableCompression --name=my_vdoTo start it again, use the following command:
# vdo enableCompression --name=my_vdo
30.4.9. Managing Free Space
Because VDO is a thinly provisioned block storage target, the amount of physical space VDO uses may differ from the size of the volume presented to users of the storage. Integrators and systems administrators can exploit this disparity to save on storage costs but must take care to avoid unexpectedly running out of storage space if the data written does not achieve the expected rate of deduplication.
Whenever the number of logical blocks (virtual storage) exceeds the number of physical blocks (actual storage), it becomes possible for file systems and applications to unexpectedly run out of space. For that reason, storage systems using VDO must provide storage administrators with a way of monitoring the size of the VDO's free pool. The size of this free pool may be determined by using the vdostats utility; see Section 30.7.2, “vdostats” for details. The default output of this utility lists information for all running VDO volumes in a format similar to the Linux df utility. For example:
Device 1K-blocks Used Available Use% /dev/mapper/my_vdo 211812352 105906176 105906176 50%When the physical storage capacity of a VDO volume is almost full, VDO reports a warning in the system log, similar to the following:
Oct 2 17:13:39 system lvm[13863]: Monitoring VDO pool my_vdo. Oct 2 17:27:39 system lvm[13863]: WARNING: VDO pool my_vdo is now 80.69% full. Oct 2 17:28:19 system lvm[13863]: WARNING: VDO pool my_vdo is now 85.25% full. Oct 2 17:29:39 system lvm[13863]: WARNING: VDO pool my_vdo is now 90.64% full. Oct 2 17:30:29 system lvm[13863]: WARNING: VDO pool my_vdo is now 96.07% full.If the size of VDO's free pool drops below a certain level, the storage administrator can take action by deleting data (which will reclaim space whenever the deleted data is not duplicated), adding physical storage, or even deleting LUNs.
Monitor physical space on your VDO volumes to prevent out-of-space situations. Running out of physical blocks might result in losing recently written, unacknowledged data on the VDO volume.
Reclaiming Space on File Systems
VDO cannot reclaim space unless file systems communicate that blocks are free using DISCARD, TRIM, or UNMAP commands. For file systems that do not use DISCARD, TRIM, or UNMAP, free space may be manually reclaimed by storing a file consisting of binary zeros and then deleting that file.
File systems may generally be configured to issue DISCARD requests in one of two ways:
Realtime discard (also online discard or inline discard)
When realtime discard is enabled, file systems send REQ_DISCARD requests to the block layer whenever a user deletes a file and frees space. VDO recieves these requests and returns space to its free pool, assuming the block was not shared.
For file systems that support online discard, you can enable it by setting the discard option at mount time.
Batch discardBatch discard is a user-initiated operation that causes the file system to notify the block layer (VDO) of any unused blocks. This is accomplished by sending the file system an ioctl request called FITRIM.
You can use the fstrim utility (for example from cron) to send this ioctl to the file system.
Reclaiming Space Without a File System
It is also possible to manage free space when the storage is being used as a block storage target without a file system. For example, a single VDO volume can be carved up into multiple subvolumes by installing the Logical Volume Manager (LVM) on top of it. Before deprovisioning a volume, the blkdiscard command can be used in order to free the space previously used by that logical volume. LVM supports the REQ_DISCARD command and will forward the requests to VDO at the appropriate logical block addresses in order to free the space. If other volume managers are being used, they would also need to support REQ_DISCARD, or equivalently, UNMAP for SCSI devices or TRIM for ATA devices.
Reclaiming Space on Fibre Channel or Ethernet Network
VDO volumes (or portions of volumes) can also be provisioned to hosts on a Fibre Channel storage fabric or an Ethernet network using SCSI target frameworks such as LIO or SCST. SCSI initiators can use the UNMAP command to free space on thinly provisioned storage targets, but the SCSI target framework will need to be configured to advertise support for this command. This is typically done by enabling thin provisioning on these volumes. Support for UNMAP can be verified on Linux-based SCSI initiators by running the following command:
# sg_vpd --page=0xb0 /dev/deviceIn the output, verify that the "Maximum unmap LBA count" value is greater than zero.
30.4.10. Increasing Logical Volume Size
Management applications can increase the logical size of a VDO volume using the vdo growLogical subcommand. Once the volume has been grown, the management should inform any devices or file systems on top of the VDO volume of its new size. The volume may be grown as follows:
# vdo growLogical --name=my_vdo --vdoLogicalSize=new_logical_sizeThe use of this command allows storage administrators to initially create VDO volumes which have a logical size small enough to be safe from running out of space. After some period of time, the actual rate of data reduction can be evaluated, and if sufficient, the logical size of the VDO volume can be grown to take advantage of the space savings.
30.4.11. Increasing Physical Volume Size
To increase the amount of physical storage available to a VDO volume:
Increase the size of the underlying device.
Use the growPhysical option to add the new physical storage space to the VDO volume:
# vdo growPhysical --name=my_vdo
It is not possible to shrink a VDO volume with this command.
30.4.12. Automating VDO with Ansible
You can use the Ansible tool to automate VDO deployment and administration. For details, see:
30.5. Deployment Scenarios
VDO can be deployed in a variety of ways to provide deduplicated storage for both block and file access and for both local and remote storage. Because VDO exposes its deduplicated storage as a standard Linux block device, it can be used with standard file systems, iSCSI and FC target drivers, or as unified storage.
30.5.1. iSCSI Target
As a simple example, the entirety of the VDO storage target can be exported as an iSCSI Target to remote iSCSI initiators.
Figure 30.3. Deduplicated Block Storage Target
30.5.2. File Systems
If file access is desired instead, file systems can be created on top of VDO and exposed to NFS or CIFS users via either the Linux NFS server or Samba.
Figure 30.4. Deduplicated NAS
30.5.3. LVM
More feature-rich systems may make further use of LVM to provide multiple LUNs that are all backed by the same deduplicated storage pool. In Figure 30.5, “Deduplicated Unified Storage”, the VDO target is registered as a physical volume so that it can be managed by LVM. Multiple logical volumes (LV1 to LV4) are created out of the deduplicated storage pool. In this way, VDO can support multiprotocol unified block/file access to the underlying deduplicated storage pool.
Figure 30.5. Deduplicated Unified Storage
Deduplicated unified storage design allows for multiple file systems to collectively use the same deduplication domain through the LVM tools. Also, file systems can take advantage of LVM snapshot, copy-on-write, and shrink or grow features, all on top of VDO.
30.5.4. Encryption
Data security is critical today. More and more companies have internal policies regarding data encryption. Linux Device Mapper mechanisms such as DM-Crypt are compatible with VDO. Encrypting VDO volumes will help ensure data security, and any file systems above VDO still gain the deduplication feature for disk optimization. Note that applying encryption above VDO results in little if any data deduplication; encryption renders duplicate blocks different before VDO can deduplicate them.
Figure 30.6. Using VDO with Encryption
30.6. Tuning VDO
30.6.1. Introduction to VDO Tuning
As with tuning databases or other complex software, tuning VDO involves making trade-offs between numerous system constraints, and some experimentation is required. The primary controls available for tuning VDO are the number of threads assigned to different types of work, the CPU affinity settings for those threads, and cache settings.
30.6.2. Background on VDO Architecture
The VDO kernel driver is multi-threaded to improve performance by amortizing processing costs across multiple concurrent I/O requests. Rather than have one thread process an I/O request from start to finish, it delegates different stages of work to one or more threads or groups of threads, with messages passed between them as the I/O request makes its way through the pipeline. This way, one thread can serialize all access to a global data structure without having to lock and unlock it each time an I/O operation is processed. If the VDO driver is well-tuned, each time a thread completes a requested processing stage there will usually be another request queued up for that same processing. Keeping these threads busy reduces the overhead of context switching and scheduling, improving performance. Separate threads are also used for parts of the operating system that can block, such as enqueueing I/O operations to the underlying storage system or messages to UDS.
The various worker thread types used by VDO are:
Logical zone threads
The logical threads, with process names including the string kvdo:logQ, maintain the mapping between the logical block numbers (LBNs) presented to the user of the VDO device and the physical block numbers (PBNs) in the underlying storage system. They also implement locking such that two I/O operations attempting to write to the same block will not be processed concurrently. Logical zone threads are active during both read and write operations.
LBNs are divided into chunks (a block map page contains a bit over 3 MB of LBNs) and these chunks are grouped into zones that are divided up among the threads.
Processing should be distributed fairly evenly across the threads, though some unlucky access patterns may occasionally concentrate work in one thread or another. For example, frequent access to LBNs within a given block map page will cause one of the logical threads to process all of those operations.
Physical zone threadsPhysical, or kvdo:physQ, threads manage data block allocation and maintain reference counts. They are active during write operations.
Like LBNs, PBNs are divided into chunks called slabs, which are further divided into zones and assigned to worker threads that distribute the processing load.
I/O submission threadskvdo:bioQ threads submit block I/O (bio) operations from VDO to the storage system. They take I/O requests enqueued by other VDO threads and pass them to the underlying device driver. These threads may communicate with and update data structures associated with the device, or set up requests for the device driver's kernel threads to process. Submitting I/O requests can block if the underlying device's request queue is full, so this work is done by dedicated threads to avoid processing delays.
If these threads are frequently shown in D state by ps or top utilities, then VDO is frequently keeping the storage system busy with I/O requests. This is generally good if the storage system can service multiple requests in parallel, as some SSDs can, or if the request processing is pipelined. If thread CPU utilization is very low during these periods, it may be possible to reduce the number of I/O submission threads.
CPU usage and memory contention are dependent on the device driver(s) beneath VDO. If CPU utilization per I/O request increases as more threads are added then check for CPU, memory, or lock contention in those device drivers.
The number of I/O submission threads can be controlled using the --vdoBioThreads=thread count option of the vdo command.
CPU-processing threadskvdo:cpuQ threads exist to perform any CPU-intensive work such as computing hash values or compressing data blocks that do not block or require exclusive access to data structures associated with other thread types.
The number of CPU-processing threads can be controlled using the --vdoCpuThreads=thread count option of the vdo command.
I/O acknowledgement threadsThe kvdo:ackQ threads issue the callbacks to whatever sits atop VDO (for example, the kernel page cache, or application program threads doing direct I/O) to report completion of an I/O request. CPU time requirements and memory contention will be dependent on this other kernel-level code.
The number of acknowledgement threads can be controlled using the --vdoAckThreads=thread count option of the vdo command.
Non-scalable VDO kernel threads:Deduplication threadThe kvdo:dedupeQ thread takes queued I/O requests and contacts UDS. Since the socket buffer can fill up if the server cannot process requests quickly enough or if kernel memory is constrained by other system activity, this work is done by a separate thread so if a thread should block, other VDO processing can continue. There is also a timeout mechanism in place to skip an I/O request after a long delay (several seconds).
Journal threadThe kvdo:journalQ thread updates the recovery journal and schedules journal blocks for writing. A VDO device uses only one journal, so this work cannot be split across threads.
Packer threadThe kvdo:packerQ thread, active in the write path when compression is enabled, collects data blocks compressed by the kvdo:cpuQ threads to minimize wasted space. There is one packer data structure, and thus one packer thread, per VDO device.
30.6.3. Values to tune
30.6.3.1. CPU/memory
30.6.3.1.1. Logical, physical, cpu, ack thread countsThe logical, physical, cpu, and I/O acknowledgement work can be spread across multiple threads, the number of which can be specified during initial configuration or later if the VDO device is restarted.
One core, or one thread, can do a finite amount of work during a given time. Having one thread compute all data-block hash values, for example, would impose a hard limit on the number of data blocks that could be processed per second. Dividing the work across multiple threads (and cores) relieves that bottleneck.
As a thread or core approaches 100% usage, more work items will tend to queue up for processing. While this may result in CPU having fewer idle cycles, queueing delays and latency for individual I/O requests will typically increase. According to some queueing theory models, utilization levels above 70% or 80% can lead to excessive delays that can be several times longer than the normal processing time. Thus it may be helpful to distribute work further for a thread or core with 50% or higher utilization, even if those threads or cores are not always busy.
In the opposite case, where a thread or CPU is very lightly loaded (and thus very often asleep), supplying work for it to do is more likely to incur some additional cost. (A thread attempting to wake another thread must acquire a global lock on the scheduler's data structures, and may potentially send an inter-processor interrupt to transfer work to another core). As more cores are configured to run VDO threads, it becomes less likely that a given piece of data will be cached as work is moved between threads or as threads are moved between cores — so too much work distribution can also degrade performance.
The work performed by the logical, physical, and CPU threads per I/O request will vary based on the type of workload, so systems should be tested with the different types of workloads they are expected to service.
Write operations in sync mode involving successful deduplication will entail extra I/O operations (reading the previously stored data block), some CPU cycles (comparing the new data block to confirm that they match), and journal updates (remapping the LBN to the previously-stored data block's PBN) compared to writes of new data. When duplication is detected in async mode, data write operations are avoided at the cost of the read and compare operations described above; only one journal update can happen per write, whether or not duplication is detected.
If compression is enabled, reads and writes of compressible data will require more processing by the CPU threads.
Blocks containing all zero bytes (a zero block) are treated specially, as they commonly occur. A special entry is used to represent such data in the block map, and the zero block is not written to or read from the storage device. Thus, tests that write or read all-zero blocks may produce misleading results. The same is true, to a lesser degree, of tests that write over zero blocks or uninitialized blocks (those that were never written since the VDO device was created) because reference count updates done by the physical threads are not required for zero or uninitialized blocks.
Acknowledging I/O operations is the only task that is not significantly affected by the type of work being done or the data being operated upon, as one callback is issued per I/O operation.
30.6.3.1.2. CPU Affinity and NUMAAccessing memory across NUMA node boundaries takes longer than accessing memory on the local node. With Intel processors sharing the last-level cache between cores on a node, cache contention between nodes is a much greater problem than cache contention within a node.
Tools such as top can not distinguish between CPU cycles that do work and cycles that are stalled. These tools interpret cache contention and slow memory accesses as actual work. As a result, moving a thread between nodes may appear to reduce the thread's apparent CPU utilization while increasing the number of operations it performs per second.
While many of VDO's kernel threads maintain data structures that are accessed by only one thread, they do frequently exchange messages about the I/O requests themselves. Contention may be high if VDO threads are run on multiple nodes, or if threads are reassigned from one node to another by the scheduler. If it is possible to run other VDO-related work (such as I/O submissions to VDO, or interrupt processing for the storage device) on the same node as the VDO threads, contention may be further reduced. If one node does not have sufficient cycles to run all VDO-related work, memory contention should be considered when selecting threads to move onto other nodes.
If practical, collect VDO threads on one node using the taskset utility. If other VDO-related work can also be run on the same node, that may further reduce contention. In that case, if one node lacks the CPU power to keep up with processing demands then memory contention must be considered when choosing threads to move onto other nodes. For example, if a storage device's driver has a significant number of data structures to maintain, it may help to move both the device's interrupt handling and VDO's I/O submissions (the bio threads that call the device's driver code) to another node. Keeping I/O acknowledgment (ack threads) and higher-level I/O submission threads (user-mode threads doing direct I/O, or the kernel's page cache flush thread) paired is also good practice.
30.6.3.1.3. Frequency throttlingIf power consumption is not an issue, writing the string performance to the /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor files if they exist might produce better results. If these sysfs nodes do not exist, Linux or the system's BIOS may provide other options for configuring CPU frequency management.
Performance measurements are further complicated by CPUs that dynamically vary their frequencies based on workload, because the time needed to accomplish a specific piece of work may vary due to other work the CPU has been doing, even without task switching or cache contention.
30.6.3.2. Caching
30.6.3.2.1. Block Map CacheVDO caches a number of block map pages for efficiency. The cache size defaults to 128 MB, but it can be increased with the --blockMapCacheSize=megabytes option of the vdo command. Using a larger cache may produce significant benefits for random-access workloads.
30.6.3.2.2. Read CacheA second cache may be used for caching data blocks read from the storage system to verify VDO's deduplication advice. If similar data blocks are seen within a short time span, the number of I/O operations needed may be reduced.
The read cache also holds storage blocks containing compressed user data. If multiple compressible blocks were written within a short period of time, their compressed versions may be located within the same storage system block. Likewise, if they are read within a short time, caching may avoid the need for additional reads from the storage system.
The vdo command's --readCache={enabled | disabled} option controls whether a read cache is used. If enabled, the cache has a minimum size of 8 MB, but it can be increased with the --readCacheSize=megabytes option. Managing the read cache incurs a slight overhead, so it may not increase performance if the storage system is fast enough. The read cache is disabled by default.
30.6.3.3. Storage System I/O
30.6.3.3.1. Bio ThreadsFor generic hard drives in a RAID configuration, one or two bio threads may be sufficient for submitting I/O operations. If the storage device driver requires its I/O submission threads to do significantly more work (updating driver data structures or communicating with the device) such that one or two threads are very busy and storage devices are often idle, the bio thread count can be increased to compensate. However, depending on the driver implementation, raising the thread count too high may lead to cache or spin lock contention. If device access timing is not uniform across all NUMA nodes, it may be helpful to run bio threads on the node "closest" to the storage device controllers.
30.6.3.3.2. IRQ HandlingIf a device driver does significant work in its interrupt handler and does not use a threaded IRQ handler, it may prevent the scheduler from providing the best performance. CPU time spent servicing hardware interrupts may look like normal VDO (or other) kernel thread execution in some ways. For example, if hardware IRQ handling required 30% of a core's cycles, a busy kernel thread on the same core could only use the remaining 70%. However, if the work queued up for that thread demanded 80% of the core's cycles, the thread would never catch up, and the scheduler might simply leave that thread to run impeded on that core instead of switching that thread to a less busy core.
Using such a device driver under a heavy VDO workload may require a large number of cycles to service hardware interrupts (the %hi indicator in the header of the top display). In that case it may help to assign IRQ handling to certain cores and adjust the CPU affinity of VDO kernel threads not to run on those cores.
30.6.3.4. Maximum Discard Sectors
The maximum allowed size of DISCARD (TRIM) operations to a VDO device can be tuned via /sys/kvdo/max_discard_sectors, based on system usage. The default is 8 sectors (that is, one 4 KB block). Larger sizes may be specified, though VDO will still process them in a loop, one block at a time, ensuring that metadata updates for one discarded block are written to the journal and flushed to disk before starting on the next block.
When using a VDO volume as a local file system, Red Hat testing found that a small discard size works best, as the generic block-device code in the Linux kernel will break large discard requests into multiple smaller ones and submit them in parallel. If there is low I/O activity on the device, VDO can process many smaller requests concurrently and much more quickly than one large request.
If the VDO device is to be used as a SCSI target, the initiator and target software introduce additional factors to consider. If the target SCSI software is SCST, it reads the maximum discard size and relays it to the initiator. (Red Hat has not attempted to tune VDO configurations in conjunction with LIO SCSI target code.)
Because the Linux SCSI initiator code allows only one discard operation at a time, discard requests that exceed the maximum size would be broken into multiple smaller discards and sent, one at a time, to the target system (and to VDO). So, in addition to VDO processing a number of small discard operations in serial, the round-trip communication time between the two systems adds additional latency.
Setting a larger maximum discard size can reduce this communication overhead, though that larger request is passed in its entirety to VDO and processed one 4 KB block at a time. While there is no per-block communication delay, additional processing time for the larger block may cause the SCSI initiator software to time out.
For SCSI target usage, Red Hat recommends configuring the maximum discard size to be moderately large while still keeping the typical discard time well within the initiator's timeout setting. An extra round-trip cost every few seconds, for example, should not significantly affect performance and SCSI initiators with timeouts of 30 or 60 seconds should not time out.
30.6.4. Identifying Bottlenecks
There are several key factors that affect VDO performance, and many tools available to identify those having the most impact.
Thread or CPU utilization above 70%, as seen in utilities such as top or ps, generally implies that too much work is being concentrated in one thread or on one CPU. However, in some cases it could mean that a VDO thread was scheduled to run on the CPU but no work actually happened; this scenario could occur with excessive hardware interrupt handler processing, memory contention between cores or NUMA nodes, or contention for a spin lock.
When using the top utility to examine system performance, Red Hat suggests running top -H to show all process threads separately and then entering the 1 f j keys, followed by the Enter/Return key; the top command then displays the load on individual CPU cores and identifies the CPU on which each process or thread last ran. This information can provide the following insights:
If a core has low %id (idle) and %wa (waiting-for-I/O) values, it is being kept busy with work of some kind.
If the %hi value for a core is very low, that core is doing normal processing work, which is being load-balanced by the kernel scheduler. Adding more cores to that set may reduce the load as long as it does not introduce NUMA contention.
If the %hi for a core is more than a few percent and only one thread is assigned to that core, and %id and %wa are zero, the core is over-committed and the scheduler is not addressing the situation. In this case the kernel thread or the device interrupt handling should be reassigned to keep them on separate cores.
The perf utility can examine the performance counters of many CPUs. Red Hat suggests using the perf top subcommand as a starting point to examine the work a thread or processor is doing. If, for example, the bioQ threads are spending many cycles trying to acquire spin locks, there may be too much contention in the device driver below VDO, and reducing the number of bioQ threads might alleviate the situation. High CPU use (in acquiring spin locks or elsewhere) could also indicate contention between NUMA nodes if, for example, the bioQ threads and the device interrupt handler are running on different nodes. If the processor supports them, counters such as stalled-cycles-backend, cache-misses, and node-load-misses may be of interest.
The sar utility can provide periodic reports on multiple system statistics. The sar -d 1 command reports block device utilization levels (percentage of the time they have at least one I/O operation in progress) and queue lengths (number of I/O requests waiting) once per second. However, not all block device drivers can report such information, so the sar usefulness might depend on the device drivers in use.
30.7. VDO Commands
This section describes the following VDO utilities:
vdo
The vdo utility manages both the kvdo and UDS components of VDO.
It is also used to enable or disable compression.
vdostatsThe vdostats utility displays statistics for each configured (or specified) device in a format similar to the Linux df utility.
30.7.1. vdo
The vdo utility manages both the kvdo and UDS components of VDO.
Synopsis
vdo { activate | changeWritePolicy | create | deactivate | disableCompression | disableDeduplication | enableCompression | enableDeduplication | growLogical | growPhysical | list | modify | printConfigFile | remove | start | status | stop } [ options... ]Sub-Commands
Options
Table 30.5. VDO Options
--indexMem=gigabytes | Specifies the amount of UDS server memory in gigabytes; the default size is 1 GB. The special decimal values 0.25, 0.5, 0.75 can be used, as can any positive integer. |
--sparseIndex={enabled | disabled} | Enables or disables sparse indexing. The default is disabled. |
--all | Indicates that the command should be applied to all configured VDO volumes. May not be used with --name. |
--blockMapCacheSize=megabytes | Specifies the amount of memory allocated for caching block map pages; the value must be a multiple of 4096. Using a value with a B(ytes), K(ilobytes), M(egabytes), G(igabytes), T(erabytes), P(etabytes) or E(xabytes) suffix is optional. If no suffix is supplied, the value will be interpreted as megabytes. The default is 128M; the value must be at least 128M and less than 16T. Note that there is a memory overhead of 15%. |
--blockMapPeriod=period | A value between 1 and 16380 which determines the number of block map updates which may accumulate before cached pages are flushed to disk. Higher values decrease recovery time after a crash at the expense of decreased performance during normal operation. The default value is 16380. Speak with your Red Hat representative before tuning this parameter. |
--compression={enabled | disabled} | Enables or disables compression within the VDO device. The default is enabled. Compression may be disabled if necessary to maximize performance or to speed processing of data that is unlikely to compress. |
--confFile=file | Specifies an alternate configuration file. The default is /etc/vdoconf.yml. |
--deduplication={enabled | disabled} | Enables or disables deduplication within the VDO device. The default is enabled. Deduplication may be disabled in instances where data is not expected to have good deduplication rates but compression is still desired. |
--emulate512={enabled | disabled} | Enables 512-byte block device emulation mode. The default is disabled. |
--force | Unmounts mounted file systems before stopping a VDO volume. |
--forceRebuild | Forces an offline rebuild before starting a read-only VDO volume so that it may be brought back online and made available. This option may result in data loss or corruption. |
--help | Displays documentation for the vdo utility. |
--logfile=pathname | Specify the file to which this script's log messages are directed. Warning and error messages are always logged to syslog as well. |
--name=volume | Operates on the specified VDO volume. May not be used with --all. |
--device=device | Specifies the absolute path of the device to use for VDO storage. |
--activate={enabled | disabled} | The argument disabled indicates that the VDO volume should only be created. The volume will not be started or enabled. The default is enabled. |
--vdoAckThreads=thread count | Specifies the number of threads to use for acknowledging completion of requested VDO I/O operations. The default is 1; the value must be at least 0 and less than or equal to 100. |
--vdoBioRotationInterval=I/O count | Specifies the number of I/O operations to enqueue for each bio-submission thread before directing work to the next. The default is 64; the value must be at least 1 and less than or equal to 1024. |
--vdoBioThreads=thread count | Specifies the number of threads to use for submitting I/O operations to the storage device. Minimum is 1; maximum is 100. The default is 4; the value must be at least 1 and less than or equal to 100. |
--vdoCpuThreads=thread count | Specifies the number of threads to use for CPU- intensive work such as hashing or compression. The default is 2; the value must be at least 1 and less than or equal to 100. |
--vdoHashZoneThreads=thread count | Specifies the number of threads across which to subdivide parts of the VDO processing based on the hash value computed from the block data. The default is 1; the value must be at least 0 and less than or equal to 100. vdoHashZoneThreads, vdoLogicalThreads and vdoPhysicalThreads must be either all zero or all non-zero. |
--vdoLogicalThreads=thread count | Specifies the number of threads across which to subdivide parts of the VDO processing based on the hash value computed from the block data. The value must be at least 0 and less than or equal to 100. A logical thread count of 9 or more will require explicitly specifying a sufficiently large block map cache size, as well. vdoHashZoneThreads, vdoLogicalThreads, and vdoPhysicalThreads must be either all zero or all non‐zero. The default is 1. |
--vdoLogLevel=level | Specifies the VDO driver log level: critical, error, warning, notice, info, or debug. Levels are case sensitive; the default is info. |
--vdoLogicalSize=megabytes | Specifies the logical VDO volume size in megabytes. Using a value with a S(ectors), B(ytes), K(ilobytes), M(egabytes), G(igabytes), T(erabytes), P(etabytes) or E(xabytes) suffix is optional. Used for over- provisioning volumes. This defaults to the size of the storage device. |
--vdoPhysicalThreads=thread count | Specifies the number of threads across which to subdivide parts of the VDO processing based on physical block addresses. The value must be at least 0 and less than or equal to 16. Each additional thread after the first will use an additional 10 MB of RAM. vdoPhysicalThreads, vdoHashZoneThreads, and vdoLogicalThreads must be either all zero or all non‐zero. The default is 1. |
--readCache={enabled | disabled} | Enables or disables the read cache within the VDO device. The default is disabled. The cache should be enabled if write workloads are expected to have high levels of deduplication, or for read intensive workloads of highly compressible data. |
--readCacheSize=megabytes | Specifies the extra VDO device read cache size in megabytes. This space is in addition to a system- defined minimum. Using a value with a B(ytes), K(ilobytes), M(egabytes), G(igabytes), T(erabytes), P(etabytes) or E(xabytes) suffix is optional. The default is 0M. 1.12 MB of memory will be used per MB of read cache specified, per bio thread. |
--vdoSlabSize=megabytes | Specifies the size of the increment by which a VDO is grown. Using a smaller size constrains the total maximum physical size that can be accommodated. Must be a power of two between 128M and 32G; the default is 2G. Using a value with a S(ectors), B(ytes), K(ilobytes), M(egabytes), G(igabytes), T(erabytes), P(etabytes) or E(xabytes) suffix is optional. If no suffix is used, the value will be interpreted as megabytes. |
--verbose | Prints commands before executing them. |
--writePolicy={ auto | sync | async } | Specifies the write policy:
|
The status subcommand returns the following information in YAML format, divided into keys as follows:
Table 30.6. VDO Status Output
VDO Status | Information in this key covers the name of the host and date and time at which the status inquiry is being made. Parameters reported in this area include: |
Node | The host name of the system on which VDO is running. |
Date | The date and time at which the vdo status command is run. |
Kernel Module | Information in this key covers the configured kernel. |
Loaded | Whether or not the kernel module is loaded (True or False). |
Version Information | Information on the version of kvdo that is configured. |
Configuration | Information in this key covers the location and status of the VDO configuration file. |
File | Location of the VDO configuration file. |
Last modified | The last-modified date of the VDO configuration file. |
VDOs | Provides configuration information for all VDO volumes. Parameters reported for each VDO volume include: |
Block size | The block size of the VDO volume, in bytes. |
512 byte emulation | Indicates whether the volume is running in 512-byte emulation mode. |
Enable deduplication | Whether deduplication is enabled for the volume. |
Logical size | The logical size of the VDO volume. |
Physical size | The size of a VDO volume's underlying physical storage. |
Write policy | The configured value of the write policy (sync or async). |
VDO Statistics | Output of the vdostats utility. |
30.7.2. vdostats
The vdostats utility displays statistics for each configured (or specified) device in a format similar to the Linux df utility.
The output of the vdostats utility may be incomplete if it is not run with root privileges.
Synopsis
vdostats [ --verbose | --human-readable | --si | --all ] [ --version ] [ device ...]Options
Table 30.7. vdostats Options
--verbose | Displays the utilization and block I/O (bios) statistics for one (or more) VDO devices. See Table 30.9, “vdostats --verbose Output” for details. |
--human-readable | Displays block values in readable form (Base 2: 1 KB = 210 bytes = 1024 bytes). |
--si | The --si option modifies the output of the --human-readable option to use SI units (Base 10: 1 KB = 103 bytes = 1000 bytes). If the --human-readable option is not supplied, the --si option has no effect. |
--all | This option is only for backwards compatibility. It is now equivalent to the --verbose option. |
--version | Displays the vdostats version. |
device ... | Specifies one or more specific volumes to report on. If this argument is omitted, vdostats will report on all devices. |
Output
Device 1K-blocks Used Available Use% Space Saving% /dev/mapper/my_vdo 1932562432 427698104 1504864328 22% 21%Table 30.8. Default vdostats Output
Device | The path to the VDO volume. |
1K-blocks | The total number of 1K blocks allocated for a VDO volume (= physical volume size * block size / 1024) |
Used | The total number of 1K blocks used on a VDO volume (= physical blocks used * block size / 1024) |
Available | The total number of 1K blocks available on a VDO volume (= physical blocks free * block size / 1024) |
Use% | The percentage of physical blocks used on a VDO volume (= used blocks / allocated blocks * 100) |
Space Saving% | The percentage of physical blocks saved on a VDO volume (= [logical blocks used - physical blocks used] / logical blocks used) |
The --human-readable option converts block counts into conventional units (1 KB = 1024 bytes):
Device Size Used Available Use% Space Saving% /dev/mapper/my_vdo 1.8T 407.9G 1.4T 22% 21%The --human-readable and --si options convert block counts into SI units (1 KB = 1000 bytes):
Device Size Used Available Use% Space Saving% /dev/mapper/my_vdo 2.0T 438G 1.5T 22% 21%Statistics printed in bold in Table 30.9, “vdostats --verbose Output” will continue to be reported in future releases. The remaining fields are primarily intended for software support and are subject to change in future releases; management tools should not rely upon them. Management tools should also not rely upon the order in which any of the statistics are reported.
Table 30.9. vdostats --verbose Output
Version | The version of these statistics. |
Release version | The release version of the VDO. |
Data blocks used | The number of physical blocks currently in use by a VDO volume to store data. |
Overhead blocks used | The number of physical blocks currently in use by a VDO volume to store VDO metadata. |
Logical blocks used | The number of logical blocks currently mapped. |
Physical blocks | The total number of physical blocks allocated for a VDO volume. |
Logical blocks | The maximum number of logical blocks that can be mapped by a VDO volume. |
1K-blocks | The total number of 1K blocks allocated for a VDO volume (= physical volume size * block size / 1024) |
1K-blocks used | The total number of 1K blocks used on a VDO volume (= physical blocks used * block size / 1024) |
1K-blocks available | The total number of 1K blocks available on a VDO volume (= physical blocks free * block size / 1024) |
Used percent | The percentage of physical blocks used on a VDO volume (= used blocks / allocated blocks * 100) |
Saving percent | The percentage of physical blocks saved on a VDO volume (= [logical blocks used - physical blocks used] / logical blocks used) |
Block map cache size | The size of the block map cache, in bytes. |
Write policy | The active write policy (sync or async). This is configured via vdo changeWritePolicy --writePolicy=auto|sync|async. |
Block size | The block size of a VDO volume, in bytes. |
Completed recovery count | The number of times a VDO volume has recovered from an unclean shutdown. |
Read-only recovery count | The number of times a VDO volume has been recovered from read-only mode (via vdo start --forceRebuild). |
Operating mode | Indicates whether a VDO volume is operating normally, is in recovery mode, or is in read-only mode. |
Recovery progress (%) | Indicates online recovery progress, or N/A if the volume is not in recovery mode. |
Compressed fragments written | The number of compressed fragments that have been written since the VDO volume was last restarted. |
Compressed blocks written | The number of physical blocks of compressed data that have been written since the VDO volume was last restarted. |
Compressed fragments in packer | The number of compressed fragments being processed that have not yet been written. |
Slab count | The total number of slabs. |
Slabs opened | The total number of slabs from which blocks have ever been allocated. |
Slabs reopened | The number of times slabs have been re-opened since the VDO was started. |
Journal disk full count | The number of times a request could not make a recovery journal entry because the recovery journal was full. |
Journal commits requested count | The number of times the recovery journal requested slab journal commits. |
Journal entries batching | The number of journal entry writes started minus the number of journal entries written. |
Journal entries started | The number of journal entries which have been made in memory. |
Journal entries writing | The number of journal entries in submitted writes minus the number of journal entries committed to storage. |
Journal entries written | The total number of journal entries for which a write has been issued. |
Journal entries committed | The number of journal entries written to storage. |
Journal blocks batching | The number of journal block writes started minus the number of journal blocks written. |
Journal blocks started | The number of journal blocks which have been touched in memory. |
Journal blocks writing | The number of journal blocks written (with metadatata in active memory) minus the number of journal blocks committed. |
Journal entries written | The total number of journal blocks for which a write has been issued. |
Journal blocks committed | The number of journal blocks written to storage. |
Slab journal disk full count | The number of times an on-disk slab journal was full. |
Slab journal flush count | The number of times an entry was added to a slab journal that was over the flush threshold. |
Slab journal blocked count | The number of times an entry was added to a slab journal that was over the blocking threshold. |
Slab journal blocks written | The number of slab journal block writes issued. |
Slab journal tail busy count | The number of times write requests blocked waiting for a slab journal write. |
Slab summary blocks written | The number of slab summary block writes issued. |
Reference blocks written | The number of reference block writes issued. |
Block map dirty pages | The number of dirty pages in the block map cache. |
Block map clean pages | The number of clean pages in the block map cache. |
Block map free pages | The number of free pages in the block map cache. |
Block map failed pages | The number of block map cache pages that have write errors. |
Block map incoming pages | The number of block map cache pages that are being read into the cache. |
Block map outgoing pages | The number of block map cache pages that are being written. |
Block map cache pressure | The number of times a free page was not available when needed. |
Block map read count | The total number of block map page reads. |
Block map write count | The total number of block map page writes. |
Block map failed reads | The total number of block map read errors. |
Block map failed writes | The total number of block map write errors. |
Block map reclaimed | The total number of block map pages that were reclaimed. |
Block map read outgoing | The total number of block map reads for pages that were being written. |
Block map found in cache | The total number of block map cache hits. |
Block map discard required | The total number of block map requests that required a page to be discarded. |
Block map wait for page | The total number of requests that had to wait for a page. |
Block map fetch required | The total number of requests that required a page fetch. |
Block map pages loaded | The total number of page fetches. |
Block map pages saved | The total number of page saves. |
Block map flush count | The total number of flushes issued by the block map. |
Invalid advice PBN count | The number of times the index returned invalid advice |
No space error count. | The number of write requests which failed due to the VDO volume being out of space. |
Read only error count | The number of write requests which failed due to the VDO volume being in read-only mode. |
Instance | The VDO instance. |
512 byte emulation | Indicates whether 512 byte emulation is on or off for the volume. |
Current VDO IO requests in progress. | The number of I/O requests the VDO is current processing. |
Maximum VDO IO requests in progress | The maximum number of simultaneous I/O requests the VDO has processed. |
Current dedupe queries | The number of deduplication queries currently in flight. |
Maximum dedupe queries | The maximum number of in-flight deduplication queries. |
Dedupe advice valid | The number of times deduplication advice was correct. |
Dedupe advice stale | The number of times deduplication advice was incorrect. |
Dedupe advice timeouts | The number of times deduplication queries timed out. |
Flush out | The number of flush requests submitted by VDO to the underlying storage. |
Bios in... Bios in partial... Bios out... Bios meta... Bios journal... Bios page cache... Bios out completed... Bio meta completed... Bios journal completed... Bios page cache completed... Bios acknowledged... Bios acknowledged partial... Bios in progress... | These statistics count the number of bios in each category with a given flag. The categories are:
There are three types of flags:
|
Read cache accesses | The number of times VDO searched the read cache. |
Read cache hits | The number of read cache hits. |
30.8. Statistics Files in /sys
Statistics for a running VDO volume may be read from files in the /sys/kvdo/volume_name/statistics directory, where volume_name is the name of thhe VDO volume. This provides an alternate interface to the data produced by the vdostats utility suitable for access by shell scripts and management software.
There are files in the statistics directory in addition to the ones listed in the table below. These additional statistics files are not guaranteed to be supported in future releases.
Table 30.10. Statistics files
dataBlocksUsed | The number of physical blocks currently in use by a VDO volume to store data. |
logicalBlocksUsed | The number of logical blocks currently mapped. |
physicalBlocks | The total number of physical blocks allocated for a VDO volume. |
logicalBlocks | The maximum number of logical blocks that can be mapped by a VDO volume. |
mode | Indicates whether a VDO volume is operating normally, is in recovery mode, or is in read-only mode. |
Chapter 31. VDO Evaluation
31.1. Introduction
VDO is software that provides inline block-level deduplication, compression, and thin provisioning capabilities for primary storage. VDO installs within the Linux device mapper framework, where it takes ownership of existing physical block devices and remaps these to new, higher-level block devices with data-efficiency properties. Specifically, VDO can multiply the effective capacity of these devices by ten or more. These benefits require additional system resources, so it is therefore necessary to measure VDO's impact on system performance.
Storage vendors undoubtedly have existing in-house test plans and expertise that they use to evaluate new storage products. Since the VDO layer helps to identify deduplication and compression, different tests may be required. An effective test plan requires studying the VDO architecture and exploring these items:
VDO-specific configurable properties (performance tuning end-user applications)
Impact of being a native 4 KB block device
Response to access patterns and distributions of deduplication and compression
Performance in high-load environments (very important)
Analyze cost vs. capacity vs. performance, based on application
Failure to consider such factors up front has created situations that have invalidated certain tests and required customers to repeat testing and data collection efforts.
31.1.1. Expectations and Deliverables
This Evaluation Guide is meant to augment, not replace, a vendor's internal evaluation effort. With a modest investment of time, it will help evaluators produce an accurate assessment of VDO's integration into existing storage devices. This guide is designed to:
Help engineers identify configuration settings that elicit optimal responses from the test device
Provide an understanding of basic tuning parameters to help avoid product misconfigurations
Create a performance results portfolio as a reference to compare against "real" application results
Identify how different workloads affect performance and data efficiency
Expedite time-to-market with VDO implementations
The test results will help Red Hat engineers assist in understanding VDO's behavior when integrated into specific storage environments. OEMs will understand how to design their deduplication and compression capable devices, and also how their customers can tune their applications to best use those devices.
Be aware that the procedures in this document are designed to provide conditions under which VDO can be most realistically evaluated. Altering test procedures or parameters may invalidate results. Red Hat Sales Engineers are available to offer guidance when modifying test plans.
31.2. Test Environment Preparations
Before evaluating VDO, it is important to consider the host system configuration, VDO configuration, and the workloads that will be used during testing. These choices will affect benchmarking both in terms of data optimization (space efficiency) and performance (bandwidth and latency). Items that should be considered when developing test plans are listed in the following sections.
31.2.1. System Configuration
Number and type of CPU cores available. This can be controlled by using the taskset utility.
Available memory and total installed memory.
Configuration of storage devices.
Linux kernel version. Note that Red Hat Enterprise Linux 7 provides only one Linux kernel version.
Packages installed.
31.2.2. VDO Configuration
Partitioning scheme
File system(s) used on VDO volumes
Size of the physical storage assigned to a VDO volume
Size of the logical VDO volume created
Sparse or dense indexing
UDS Index in memory size
VDO's thread configuration
31.2.3. Workloads
Types of tools used to generate test data
Number of concurrent clients
The quantity of duplicate 4 KB blocks in the written data
Read and write patterns
The working set size
VDO volumes may need to be re-created in between certain tests to ensure that each test is performed on the same disk environment. Read more about this in the testing section.
31.2.4. Supported System Configurations
Red Hat has tested VDO with Red Hat Enterprise Linux 7 on the Intel 64 architecture.
The following utilities are recommended when evaluating VDO:
Flexible I/O Tester version 2.08 or higher; available from the fio package
sysstat version 8.1.2-2 or higher; available from the sysstat package
31.2.5. Pre-Test System Preparations
This section describes how to configure system settings to achieve optimal performance during the evaluation. Testing beyond the implicit bounds established in any particular test may result in loss of testing time due to abnormal results. For example, this guide describes a test that conducts random reads over a 100 GB address range. To test a working set of 500 GB, the amount of DRAM allocated for the VDO block map cache should be increased accordingly.
System Configuration
Ensure that your CPU is running at its highest performance setting.
Disable frequency scaling if possible using the BIOS configuration or the Linux cpupower utility.
Enable Turbo mode if possible to achieve maximum throughput. Turbo mode introduces some variability in test results, but performance will meet or exceed that of testing without Turbo.
Linux Configuration
For disk-based solutions, Linux offers several I/O scheduler algorithms to handle multiple read/write requests as they are queued. By default, Red Hat Enterprise Linux uses the CFQ (completely fair queuing) scheduler, which arranges requests in a way that improves rotational disk (hard disk) access in many situations. We instead suggest using the Deadline scheduler for rotational disks, having found that it provides better throughput and latency in Red Hat lab testing. Change the device settings as follows:
# echo "deadline" > /sys/block/device/queue/schedulerFor flash-based solutions, the noop scheduler demonstrates superior random access throughput and latency in Red Hat lab testing. Change the device settings as follows:
# echo "noop" > /sys/block/device/queue/scheduler
Storage device configuration
File systems (ext4, XFS, etc.) may have unique impacts on performance; they often skew performance measurements, making it harder to isolate VDO's impact on the results. If reasonable, we recommend measuring performance on the raw block device. If this is not possible, format the device using the file system that would be used in the target implementation.
31.2.6. VDO Internal Structures
We believe that a general understanding of VDO mechanisms is essential for a complete and successful evaluation. This understanding becomes especially important when testers wish to deviate from the test plan or devise new stimuli to emulate a particular application or use case. For more information, see Chapter 30, VDO Integration.
The Red Hat test plan was written to operate with a default VDO configuration. When developing new tests, some of the VDO parameters listed in the next section must be adjusted.
31.2.7. VDO Optimizations
High Load
Perhaps the most important strategy for producing optimal performance is determining the best I/O queue depth, a characteristic that represents the load on the storage system. Most modern storage systems perform optimally with high I/O depth. VDO's performance is best demonstrated with many concurrent requests.
Synchronous vs. Asynchronous Write Policy
VDO might operate with either of two write policies, synchronous or asynchronous. By default, VDO automatically chooses the appropriate write policy for your underlying storage device.
When testing performance, you need to know which write policy VDO selected. The following command shows the write policy of your VDO volume:
# vdo status --name=my_vdoMetadata Caching
VDO maintains a table of mappings from logical block addresses to physical block addresses, and VDO must look up the relevant mapping when accessing any particular block. By default, VDO allocates 128 MB of metadata cache in DRAM to support efficient access to 100 GB of logical space at a time. The test plan generates workloads appropriate to this configuration option.
Working sets larger than the configured cache size will require additional I/Os to service requests, in which case performance degradation will occur. If additional memory is available, the block map cache should be made larger. If the working set is larger than what the block map cache can hold in memory, additional I/O hover head can occur to lookup associated block map pages.
VDO Multithreading Configuration
VDO's thread configuration must be tuned to achieve optimal performance. Review the VDO Integration Guide for information on how to modify these settings when creating a VDO volume. Contact your Red Hat Sales Engineer to discuss how to design a test to find the optimal setting.
Data Content
Because VDO performs deduplication and compression, test data sets must be chosen to effectively exercise these capabilities.
31.2.8. Special Considerations for Testing Read Performance
When testing read performance, these factors must be considered:
If a 4 KB block has never been written, VDO will not perform I/O to the storage and will immediately respond with a zero block.
If a 4 KB block has been written but contains all zeros, VDO will not perform I/O to the storage and will immediately respond with a zero block.
This behavior results in very fast read performance when there is no data to read. This makes it imperative that read tests prefill with actual data.
31.2.9. Cross Talk
To prevent one test from affecting the results of another, it is suggested that a new VDO volume be created for each iteration of each test.
31.3. Data Efficiency Testing Procedures
Successful validation of VDO is dependent upon following a well-structured test procedure. This section provides a series of steps to follow, along with the expected results, as examples of tests to consider when participating in an evaluation.
Test Environment
The test cases in the next section make the following assumptions about the test environment:
One or more Linux physical block devices are available.
The target block device (for example, /dev/sdb) is larger than 512 GB.
Flexible I/O Tester (fio) version 2.1.1 or later is installed.
VDO is installed.
The following information should be recorded at the start of each test in order to ensure that the test environment is fully understood:
The Linux build used, including the kernel build number.
A complete list of installed packages, as obtained from the rpm -qa command.
Complete system specifications:
CPU type and quantity (available in /proc/cpuinfo).
Installed memory and the amount available after the base OS is running (available in /proc/meminfo).
Type(s) of drive controller(s) used.
Type(s) and quantity of disk(s) used.
A complete list of running processes (from ps aux or a similar listing).
Name of the Physical Volume and the Volume Group created for use with VDO (pvs and vgs listings).
File system used when formatting the VDO volume (if any).
Permissions on the mounted directory.
Contents of /etc/vdoconf.yaml.
Location of the VDO files.
You can capture much of the required information by running sosreport.
Workloads
Effectively testing VDO requires the use of data sets that simulate real world workloads. The data sets should provide a balance between data that can be deduplicated and/or compressed and data that cannot in order to demonstrate performance under different conditions.
There are several tools that can synthetically generate data with repeatable characteristics. Two utilities in particular, VDbench and fio, are recommended for use during testing.
This guide uses fio. Understanding the arguments is critical to a successful evaluation:
Table 31.1. fio Options
--size | The quantity of data fio will send to the target per job (see numjobs below). | 100 GB |
--bs | The block size of each read/write request produced by fio. Red Hat recommends a 4 KB block size to match VDO's 4 KB default | 4k |
--numjobs | The number of jobs that fio will create to run the benchmark. Each job sends the amount of data specified by the --size parameter. The first job sends data to the device at the offset specified by the --offset parameter. Subsequent jobs write the same region of the disk (overwriting) unless the --offset_increment parameter is provided, which will offset each job from where the previous job began by that value. To achieve peak performance on flash at least two jobs are recommended. One job is typically enough to saturate rotational disk (HDD) throughput. | 1 (HDD) 2 (SSD) |
--thread | Instructs fio jobs to be run in threads rather than being forked, which may provide better performance by limiting context switching. | <N/A> |
--ioengine | There are several I/O engines available in Linux that are able to be tested using fio. Red Hat testing uses the asynchronous unbuffered engine (libaio). If you are interested in another engine, discuss that with your Red Hat Sales Engineer. The Linux libaio engine is used to evaluate workloads in which one or more processes are making random requests simultaneously. libaio allows multiple requests to be made asynchronously from a single thread before any data has been retrieved, which limits the number of context switches that would be required if the requests were provided by manythreads via a synchronous engine. | libaio |
--direct | When set, direct allows requests to be submitted to the device bypassing the Linux Kernel's page cache. Libaio Engine: libaio must be used with direct enabled (=1) or the kernel may resort to the sync API for all I / O requests. | 1 (libaio) |
--iodepth | The number of I / O buffers in flight at any time. A high iodepth will usually increase performance, particularly for random reads or writes. High depths ensure that the controller always has requests to batch. However, setting iodepth too high (greater than 1K, typically) may cause undesirable latency. While Red Hat recommends an iodepth between 128 and 512, the final value is a trade-off and depends on how your application tolerates latency. | 128 (minimum) |
--iodepth_batch_submit | The number of I/Os to create when the iodepth buffer pool begins to empty. This parameter limits task switching from I / O to buffer creation during the test. | 16 |
--iodepth_batch_complete | The number of I/Os to complete before submitting a batch (iodepth_batch_complete). This parameter limits task switching from I / O to buffer creation during the test. | 16 |
--gtod_reduce | Disables time-of-day calls to calculate latency. This setting will lower throughput if enabled (=0), so it should be enabled (=1) unless latency measurement is necessary. | 1 |
31.3.1. Configuring a VDO Test Volume
1. Create a VDO Volume with a Logical Size of 1 TB on a 512 GB Physical Volume
Create a VDO volume.
To test the VDO async mode on top of synchronous storage, create an asynchronous volume using the --writePolicy=async option:
# vdo create --name=vdo0 --device=/dev/sdb \ --vdoLogicalSize=1T --writePolicy=async --verboseTo test the VDO sync mode on top of synchronous storage, create a synchronous volume using the --writePolicy=sync option:
# vdo create --name=vdo0 --device=/dev/sdb \ --vdoLogicalSize=1T --writePolicy=sync --verbose
Format the new device with an XFS or ext4 file system.
For XFS:
# mkfs.xfs -K /dev/mapper/vdo0For ext4:
# mkfs.ext4 -E nodiscard /dev/mapper/vdo0
Mount the formatted device:
# mkdir /mnt/VDOVolume # mount /dev/mapper/vdo0 /mnt/VDOVolume && \ chmod a+rwx /mnt/VDOVolume
31.3.2. Testing VDO Efficiency
2. Test Reading and Writing to the VDO Volume
Write 32 GB of random data to the VDO volume:
$ dd if=/dev/urandom of=/mnt/VDOVolume/testfile bs=4096 count=8388608Read the data from the VDO volume and write it to another location not on the VDO volume:
$ dd if=/mnt/VDOVolume/testfile of=/home/user/testfile bs=4096Compare the two files using diff, which should report that the files are the same:
$ diff -s /mnt/VDOVolume/testfile /home/user/testfileCopy the file to a second location on the VDO volume:
$ dd if=/home/user/testfile of=/mnt/VDOVolume/testfile2 bs=4096Compare the third file to the second file. This should report that the files are the same:
$ diff -s /mnt/VDOVolume/testfile2 /home/user/testfile
3. Remove the VDO Volume
Unmount the file system created on the VDO volume:
# umount /mnt/VDOVolumeRun the command to remove the VDO volume vdo0 from the system:
# vdo remove --name=vdo0Verify that the volume has been removed. There should be no listing in vdo list for the VDO partition:
# vdo list --all | grep vdo
4. Measure Deduplication
Create 10 directories on the VDO volume named vdo1 through vdo10 to hold 10 copies of the test data set:
$ mkdir /mnt/VDOVolume/vdo{01..10}Examine the amount of disk space used according to the file system:
$ df -h /mnt/VDOVolume Filesystem Size Used Avail Use% Mounted on /dev/mapper/vdo0 1.5T 198M 1.4T 1% /mnt/VDOVolumeConsider tabulating the results in a table:
StatisticBare File SystemAfter SeedAfter 10 Copies File System Used Size 198 MB VDO Data Used VDO Logical Used Run the following command and record the values. "Data blocks used" is the number of blocks used by user data on the physical device running under VDO. "Logical blocks used" is the number of blocks used before optimization. It will be used as the starting point for measurements
# vdostats --verbose | grep "blocks used" data blocks used : 1090 overhead blocks used : 538846 logical blocks used : 6059434Create a data source file in the top level of the VDO volume
$ dd if=/dev/urandom of=/mnt/VDOVolume/sourcefile bs=4096 count=1048576 4294967296 bytes (4.3 GB) copied, 540.538 s, 7.9 MB/sRe-examine the amount of used physical disk space in use. This should show an increase in the number of blocks used corresponding to the file just written:
$ df -h /mnt/VDOVolume Filesystem Size Used Avail Use% Mounted on /dev/mapper/vdo0 1.5T 4.2G 1.4T 1% /mnt/VDOVolume # vdostats --verbose | grep "blocks used" data blocks used : 1050093 (increased by 4GB) overhead blocks used : 538846 (Did not change) logical blocks used : 7108036 (increased by 4GB)Copy the file to each of the 10 subdirectories:
$ for i in {01..10}; do cp /mnt/VDOVolume/sourcefile /mnt/VDOVolume/vdo$i doneOnce again, check the amount of physical disk space used (data blocks used). This number should be similar to the result of step 6 above, with only a slight increase due to file system journaling and metadata:
$ df -h /mnt/VDOVolume Filesystem Size Used Avail Use% Mounted on /dev/mapper/vdo0 1.5T 45G 1.3T 4% /mnt/VDOVolume # vdostats --verbose | grep "blocks used" data blocks used : 1050836 (increased by 3M) overhead blocks used : 538846 logical blocks used : 17594127 (increased by 41G)Subtract this new value of the space used by the file system from the value found before writing the test data. This is the amount of space consumed by this test from the file system's perspective.
Observe the space savings in your recorded statistics:
Note:In the following table, values have been converted to MB/GB. vdostats "blocks" are 4,096 B.
StatisticBare File SystemAfter SeedAfter 10 Copies File System Used Size 198 MB 4.2 GB 45 GB VDO Data Used 4 MB 4.1 GB 4.1 GB VDO Logical Used 23.6 GB* 27.8 GB 68.7 GB * File system overhead for 1.6 TB formatted drive
5. Measure Compression
Create a VDO volume of at least 10 GB of physical and logical size. Add options to disable deduplication and enable compression:
# vdo create --name=vdo0 --device=/dev/sdb \ --vdoLogicalSize=10G --verbose \ --deduplication=disabled --compression=enabledInspect VDO statistics before transfer; make note of data blocks used and logical blocks used (both should be zero):
# vdostats --verbose | grep "blocks used"-
Format the new device with an XFS or ext4 file system.
For XFS:
# mkfs.xfs -K /dev/mapper/vdo0For ext4:
# mkfs.ext4 -E nodiscard /dev/mapper/vdo0
Mount the formatted device:
# mkdir /mnt/VDOVolume # mount /dev/mapper/vdo0 /mnt/VDOVolume && \ chmod a+rwx /mnt/VDOVolumeSynchronize the VDO volume to complete any unfinished compression:
# sync && dmsetup message vdo0 0 sync-dedupeInspect VDO statistics again. Logical blocks used — data blocks used is the number of 4 KB blocks saved by compression for the file system alone. VDO optimizes file system overhead as well as actual user data:
# vdostats --verbose | grep "blocks used"Copy the contents of /lib to the VDO volume. Record the total size:
# cp -vR /lib /mnt/VDOVolume ... sent 152508960 bytes received 60448 bytes 61027763.20 bytes/sec total size is 152293104 speedup is 1.00Synchronize Linux caches and the VDO volume:
# sync && dmsetup message vdo0 0 sync-dedupeInspect VDO statistics once again. Observe the logical and data blocks used:
# vdostats --verbose | grep "blocks used"Logical blocks used - data blocks used represents the amount of space used (in units of 4 KB blocks) for the copy of your /lib files.
Remove the VDO volume:
# umount /mnt/VDOVolume && vdo remove --name=vdo0
6. Test VDO Compression Efficiency
Experiment with your own datasets.
7. Understanding TRIM and DISCARD
Thin provisioning allows a logical or virtual storage space to be larger than the underlying physical storage. Applications such as file systems benefit from running on the larger virtual layer of storage, and data-efficiency techniques such as data deduplication reduce the number of physical data blocks needed to store all of the data. To benefit from these storage savings, the physical storage layer needs to know when application data has been deleted.
Traditional file systems did not have to inform the underlying storage when data was deleted. File systems that work with thin provisioned storage send TRIM or DISCARD commands to inform the storage system when a logical block is no longer required. These commands can be sent whenever a block is deleted using the discard mount option, or these commands can be sent in a controlled manner by running utilities such as fstrim that tell the file system to detect which logical blocks are unused and send the information to the storage system in the form of a TRIM or DISCARD command.
To see how this works:
Trim the file system to remove any unneeded blocks (this may take a long time):
# fstrim /mnt/VDOVolumeRecord the initial state in following table below by entering:
$ df -m /mnt/VDOVolumeto see how much capacity is used in the file system, and run vdostats to see how many physical and logical data blocks are being used.
Create a 1 GB file with non-duplicate data in the file system running on top of VDO:
$ dd if=/dev/urandom of=/mnt/VDOVolume/file bs=1M count=1Kand then collect the same data. The file system should have used an additional 1 GB, and the data blocks used and logical blocks used have increased similarly.
Run fstrim /mnt/VDOVolume and confirm that this has no impact after creating a new file.
Delete the 1 GB file:
$ rm /mnt/VDOVolume/fileCheck and record the parameters. The file system is aware that a file has been deleted, but there has been no change to the number of physical or logical blocks because the file deletion has not been communicated to the underlying storage.
Run fstrim /mnt/VDOVolume and record the same parameters. fstrim looks for free blocks in the file system and sends a TRIM command to the VDO volume for unused addresses, which releases the associated logical blocks, and VDO processes the TRIM to release the underlying physical blocks.
StepFile Space Used (MB)Data Blocks UsedLogical Blocks Used Initial Add 1 GB File Run fstrim Delete 1 GB File Run fstrim
From this exercise, the TRIM process is needed so the underlying storage can have an accurate knowledge of capacity utilization. fstrim is a command line tool that analyzes many blocks at once for greater efficiency. An alternative method is to use the file system discard option when mounting. The discard option will update the underlying storage after each file system block is deleted, which can slow throughput but provides for great utilization awareness. It is also important to understand that the need to TRIM or DISCARD unused blocks is not unique to VDO; any thin-provisioned storage system has the same challenge
31.4. Performance Testing Procedures
The goal of this section is to construct a performance profile of the device with VDO installed. Each test should be run with and without VDO installed, so that VDO's performance can be evaluated relative to the performance of the base system.
31.4.1. Phase 1: Effects of I/O Depth, Fixed 4 KB Blocks
The goal of this test is to determine the I/O depth that produces the optimal throughput and the lowest latency for your appliance. VDO uses a 4 KB sector size rather than the traditional 512 B used on legacy storage devices. The larger sector size allows it to support higher-capacity storage, improve performance, and match the cache buffer size used by most operating systems.
Perform four-corner testing at 4 KB I/O, and I/O depth of 1, 8, 16, 32, 64, 128, 256, 512, 1024:
Sequential 100% reads, at fixed 4 KB *
Sequential 100% write, at fixed 4 KB
Random 100% reads, at fixed 4 KB *
Random 100% write, at fixed 4 KB **
* Prefill any areas that may be read during the read test by performing a write fio job first
** Re-create the VDO volume after 4 KB random write I/O runs
Example shell test input stimulus (write):
# for depth in 1 2 4 8 16 32 64 128 256 512 1024 2048; do fio --rw=write --bs=4096 --name=vdo --filename=/dev/mapper/vdo0 \ --ioengine=libaio --numjobs=1 --thread --norandommap --runtime=300\ --direct=1 --iodepth=$depth --scramble_buffers=1 --offset=0 \ --size=100g doneRecord throughput and latency at each data point, and then graph.
Repeat test to complete four-corner testing: --rw=randwrite, --rw=read, and --rw=randread.
The result is a graph as shown below. Points of interest are the behavior across the range and the points of inflection where increased I / O depth proves to provide diminishing throughput gains. Likely, sequential access and random access will peak at different values, but it may be different for all types of storage configurations. In Figure 31.1, “I/O Depth Analysis” notice the "knee" in each performance curve. Marker 1 identifies the peak sequential throughput at point X, and marker 2 identifies peak random 4 KB throughput at point Z.
This particular appliance does not benefit from sequential 4 KB I / O depth > X. Beyond that depth, there are diminishing bandwidth bandwidth gains, and average request latency will increase 1:1 for each additional I / O request.
This particular appliance does not benefit from random 4 KB I / O depth > Z. Beyond that depth, there are diminishing bandwidth gains, and average request latency will increase 1:1 for each additional I / O request.
Figure 31.1. I/O Depth Analysis
Figure 31.2, “Latency Response of Increasing I/O for Random Writes” shows an example of the random write latency after the "knee" of the curve in Figure 31.1, “I/O Depth Analysis”. Benchmarking practice should test at these points for maximum throughput that incurs the least response time penalty. As we move forward in the test plan for this example appliance, we will collect additional data with I / O depth = Z
Figure 31.2. Latency Response of Increasing I/O for Random Writes
31.4.2. Phase 2: Effects of I/O Request Size
The goal of this test is to understand the block size that produces the best performance of the system under test at the optimal I/O depth determined in the previous step.
Perform four-corner testing at fixed I/O depth, with varied block size (powers of 2) over the range 8 KB to 1 MB. Remember to prefill any areas to be read and to recreate volumes between tests.
Example test input stimulus (write):
# z=[see previous step] # for iosize in 4 8 16 32 64 128 256 512 1024; do fio --rw=write --bs=$iosize\k --name=vdo --filename=/dev/mapper/vdo0 --ioengine=libaio --numjobs=1 --thread --norandommap --runtime=300 --direct=1 --iodepth=$z --scramble_buffers=1 --offset=0 --size=100g doneRecord throughput and latency at each data point, and then graph.
Repeat test to complete four-corner testing: --rw=randwrite, --rw=read, and --rw=randread.
There are several points of interest that you may find in the results. In this example:
Sequential writes reach a peak throughput at request size Y. This curve demonstrates how applications that are configurable or naturally dominated by certain request sizes may perceive performance. Larger request sizes often provide more throughput because 4 KB I/Os may benefit from merging.
Sequential reads reach a similar peak throughput at point Z. Remember that after these peaks, overall latency before the I/O completes will increase with no additional throughput. It would be wise to tune the device to not accept I/Os larger than this size.
Random reads achieve peak throughput at point X. Some devices may achieve near-sequential throughput rates at large request size random accesses, while others suffer more penalty when varying from purely sequential access.
Random writes achieve peak throughput at point Y. Random writes involve the most interaction of a deduplication device, and VDO achieves high performance especially when request sizes and/or I/O depths are large.
The results from this test Figure 31.3, “Request Size vs. Throughput Analysis and Key Inflection Points” help in understanding the characteristics of the storage device and the user experience for specific applications. Consult with a Red Hat Sales Engineer to determine if there may be further tuning needed to increase performance at different request sizes.
Figure 31.3. Request Size vs. Throughput Analysis and Key Inflection Points
31.4.3. Phase 3: Effects of Mixing Read & Write I/Os
The goal of this test is to understand how your appliance with VDO behaves when presented with mixed I/O loads (read/write), analyzing the effects of read/write mix at the optimal random queue depth and request sizes from 4 KB to 1 MB. You should use whatever is appropriate in your case.
Perform four-corner testing at fixed I/O depth, varied block size (powers of 2) over the 8 KB to 256 KB range, and set read percentage at 10% increments, beginning with 0%. Remember to prefill any areas to be read and to recreate volumes between tests.
Example test input stimulus (read/write mix):
# z=[see previous step] # for readmix in 0 10 20 30 40 50 60 70 80 90 100; do for iosize in 4 8 16 32 64 128 256 512 1024; do fio --rw=rw --rwmixread=$readmix --bs=$iosize\k --name=vdo \ --filename=/dev/mapper/vdo0 --ioengine=libaio --numjobs=1 --thread \ --norandommap --runtime=300 --direct=0 --iodepth=$z --scramble_buffers=1 \ --offset=0 --size=100g done doneRecord throughput and latency at each data point, and then graph.
Figure 31.4. Performance Is Consistent across Varying Read/Write Mixes
Performance (aggregate) and latency (aggregate) are relatively consistent across the range of mixing reads and writes, trending from the lower max write throughput to the higher max read throughput.
This behavior may vary with different storage, but the important observation is that the performance is consistent under varying loads and/or that you can understand performance expectation for applications that demonstrate specific read/write mixes. If you discover any unexpected results, Red Hat Sales Engineers will be able to help you understand if it is VDO or the storage device itself that needs modification.
Note: Systems that do not exhibit a similar response consistency often signify a sub-optimal configuration. Contact your Red Hat Sales Engineer if this occurs.
31.4.4. Phase 4: Application Environments
The goal of these final tests is to understand how the system with VDO behaves when deployed in a real application environment. If possible, use real applications and use the knowledge learned so far; consider limiting the permissible queue depth on your appliance, and if possible tune the application to issue requests with those block sizes most beneficial to VDO performance.
Request sizes, I/O loads, read/write patterns, etc., are generally hard to predict, as they will vary by application use case (i.e., filers vs. virtual desktops vs. database), and applications often vary in the types of I/O based on the specific operation or due to multi-tenant access.
The final test shows general VDO performance in a mixed environment. If more specific details are known about your expected environment, test those settings as well.
Example test input stimulus (read/write mix):
# for readmix in 20 50 80; do for iosize in 4 8 16 32 64 128 256 512 1024; do fio --rw=rw --rwmixread=$readmix --bsrange=4k-256k --name=vdo \ --filename=/dev/mapper/vdo0 --ioengine=libaio --numjobs=1 --thread \ --norandommap --runtime=300 --direct=0 --iodepth=$iosize \ --scramble_buffers=1 --offset=0 --size=100g done doneFigure 31.5. Mixed Environment Performance
31.5. Issue Reporting
In the event that an issue is encountered while working with VDO, it is important to gather as much information as possible to assist Red Hat Sales Engineers in attempting to reproduce the issue.
Issue reports should include the following:
The VDO configuration
The use case that generated the issue
The actions that were being performed at the time of the error
The text of any error messages on the console or terminal
-
The kernel log files
Kernel crash dumps, if available
The result of sosreport, which will capture data describing the entire Linux environment
31.6. Conclusion
Going through this or any other well-structured evaluation plan is an important step in integrating VDO into any storage system. The evaluation process is important to understanding performance and catching any potential compatibility issues. The collection of results from this evaluation not only demonstrates deduplication and compression, but also provides a performance profile of your system implementing VDO. The results help determine whether the results achieved in real applications are as expected and plausible or whether they fall short of expectations. Finally, we can also use these results to help predict the kinds of applications that will operate favorably with VDO.