5 Managing LSM Objects

This chapter describes how to manage LSM objects using LSM commands. For more information on an LSM command, see the reference page corresponding to its name. For example, for more information on the volassist command, enter:

# man volassist

5.1 Managing LSM Disks

The following sections describe how to use LSM commands to manage LSM disks.

5.1.1 Displaying LSM Disk Information

To display detailed information for an LSM disk:

# voldisk list disk

The following example contains information for an LSM disk named dsk12:

Device:    dsk12
devicetag: dsk12
type:      sliced
hostid:    hostname.com
disk:      name=dsk12 id=1012859934.2400.potamus.zk3.dec.com
group:     name=dg2 id=1012859945.2405.potamus.zk3.dec.com
flags:     online ready autoimport imported
pubpaths:  block=/dev/disk/dsk12g char=/dev/rdisk/dsk12g
privpaths: block=/dev/disk/dsk12h char=/dev/rdisk/dsk12h
version:   2.1
iosize:    min=512 (bytes) max=2048 (blocks)
public:    slice=6 offset=16 len=2046748
private:   slice=7 offset=0 len=4096
update:    time=1012859947 seqno=0.1
headers:   0 248
configs:   count=1 len=2993
logs:      count=1 len=453
Defined regions:
 config   priv     17-   247[   231]: copy=01 offset=000000 enabled
 config   priv    249-  3010[  2762]: copy=01 offset=000231 enabled
 log      priv   3011-  3463[   453]: copy=01 offset=000000 enabled

5.1.2 Renaming LSM Disks

When you initialize an LSM disk, you can assign it a disk media name or use the default disk media name, which is the same as the disk access name.

Caution

Each disk in a disk group must have a unique name. To avoid confusion, you might want to ensure that no two disk groups contain disks with the same name. For example, both the rootdg disk group and another disk group might contain disks with a disk media name of disk03. Because most LSM commands operate on the rootdg disk group unless you specify otherwise, you can inadvertently perform operations on the wrong disk if multiple disk groups contain identically named disks.
The voldisk list command displays a list of all the LSM disks in all disk groups on the system.

To rename an LSM disk:

# voledit rename old_disk_media_name new_disk_media_name

For example, to rename an LSM disk from disk03 to disk01:

# voledit rename disk03 disk01

5.1.3 Placing LSM Disks Off Line

You can place an LSM disk off line to:

Prevent LSM from accessing it

Enable you to move the disk to a different physical location and have the disk retain its LSM identity

Placing a disk off line closes its device file. If a disk is in use, you cannot place it off line.

To place an LSM disk off line:

Remove the LSM disk from its disk group:
```
# voldg [-g disk_group] rmdisk disk
```

Place the LSM disk off line:
```
# voldisk offline disk
```

5.1.4 Placing LSM Disks On Line

To restore access to an LSM disk that you placed off line, place it on line. The LSM disk is placed in the free disk pool and is accessible to LSM again. After placing an LSM disk on line, you must add it to a disk group before an LSM volume can use it. If the disk belonged to a disk group previously, you can add it to the same disk group.

To place an LSM disk on line:

# voldisk online disk

For information on adding an LSM disk to a disk group, see Section 5.2.2.

5.1.5 Moving Data Off an LSM Disk

You can move (evacuate) LSM volume data to other LSM disks in the same disk group if there is sufficient free space. If you do not specify a target LSM disk, LSM uses any available LSM disk in the disk group that has sufficient free space.

You might want to move data off an LSM disk in the following circumstances:

To move data off a nopriv disk (created when you encapsulate a disk, disk partition, or AdvFS domain) to sliced or simple disks in the disk group.
This is recommended, because nopriv disks are more complex to manage and are intended to be used only temporarily, as a way to place existing data under LSM control. The exception: nopriv disks created by encapsulating the boot disk partitions on a standalone system.
For information on verifying there is sufficient free space in the disk group to support the move, see Section 5.2.1.

To move LSM objects relocated by a hot-sparing operation. See Section 5.1.6 for specific information.

To redistribute one or more volumes that use space on the same disk, if this is causing contention and slow performance. You can determine this by using the volstat command (Section 6.1.2).

To move data off a disk, use one of the following commands:

The volassist move command, which moves a specific volume off a disk, but leaves objects from other volumes on the disk.

The volevac command, which moves all LSM objects off a particular disk (for example, multiple LSM volumes or logs).

Note

Do not move the contents of an LSM disk to another LSM disk that contains data from the same volume. The resulting layout might not preserve redundancy for volumes that use mirror plexes or a RAID5 plex.

To move all data (for example, multiple LSM volumes) off an LSM disk:

# volevac [-g disk_group] source_disk target_disk [target_disk...]

For example:

To move all the data from LSM disk dsk8 to dsk9 (in the rootdg disk group):
```
# volevac dsk8 dsk9
```

To move data from a nopriv disk to any available sliced or simple disk:
```
# volevac dsk24c [disk...]
```
If you specify the target disks, make sure they have sufficient free space.

To move a volume off an LSM disk:

# volassist [-g disk_group] move volume [\]!source_disk \
[target_disk...]

For example, suppose three volumes in rootdg use space on dsk1:

# volprint -s | grep dsk1

sd dsk1-04      rootvol-04   ENABLED  65       FPA      -        -       -
sd dsk1-01      vol_1-03     ENABLED  65       LOG      -        -       -
sd dsk1-02      vol_2-03     ENABLED  65       LOG      -        -       -
 
.
.
.

The following commands move volumes vol_1 and vol_2 off dsk1 (using the quoting convention for the C shell to correctly interpret the ! symbol):

# volassist move vol_1 \!dsk1
# volassist move vol_2 \!dsk1

The following command confirms that the only volume using dsk1 after the operation is rootvol:

# volprint | grep dsk1-

sd dsk1-04   rootvol-04  ENABLED  65     FPA     -       -      -

The following command displays the disks now used in volumes vol_1 and vol_2:

# volprint vol_1 vol_2

Disk group: rootdg
 
TY NAME         ASSOC        KSTATE   LENGTH   PLOFFS   STATE    TUTIL0  PUTIL0
v  vol_1        fsgen        ENABLED  262144   -        ACTIVE   -       -
pl vol_1-01     vol_1        ENABLED  262144   -        ACTIVE   -       -
sd dsk0-01      vol_1-01     ENABLED  262144   0        -        -       -
pl vol_1-03     vol_1        ENABLED  LOGONLY  -        ACTIVE   -       -
sd dsk4-01      vol_1-03     DETACHED 65       LOG      RECOVER  -       -
 
v  vol_2        fsgen        ENABLED  262144   -        ACTIVE   -       -
pl vol_2-01     vol_2        ENABLED  262144   -        ACTIVE   -       -
sd dsk3-01      vol_2-01     ENABLED  262144   0        -        -       -
pl vol_2-03     vol_2        ENABLED  LOGONLY  -        ACTIVE   -       -
sd dsk6-01      vol_2-03     ENABLED  65       LOG      -        -       -

5.1.6 Moving LSM Objects Relocated by Hot-Sparing

When LSM objects are moved from a failing disk to a hot-spare disk by the hot-sparing feature, their new locations might not provide the same performance or have the same data layout that existed before. After hot-sparing occurs, you might want to move the relocated LSM objects to a different disk to improve performance, to keep the hot-spare disk space free for future hot-sparing needs, or to restore the LSM configuration to its previous state.

Note

The following procedure assumes you have initialized a new disk to replace the hot-spare disk. For more information on adding disks for LSM use, see Section 4.1. For more information on replacing a failed disk, see Section 6.4.5.

In the following procedure, the hot-spare disk to which LSM moved the data is called the relocation disk. The disk you choose to move the data to is called the target disk. Use the disk media name of both the relocation disk and the target disk.

To move LSM objects that were relocated by hot-sparing:

Note the characteristics of the LSM objects before they were relocated.
This information is available from the mail notification about the failure that the volwatch daemon sent to the root account. For example, look for a message similar to the following:
```
To: root
Subject: Logical Storage Manager failures on host teal
 
Attempting to relocate subdisk disk02-03 from plex home-02.
Dev_offset 0 length 1164 dm_name disk02 da_name dsk2.
The available plex home-01 will be used to recover the data.
```

Note the new location of the relocated LSM object (the relocation disk).
This information is available from the mail notification about the failure that the volwatch daemon sent to the root account. For example, look for a message similar to the following:
```
To: root
Subject: Attempting LSM relocation on host teal
 
Volume home Subdisk disk02-03 relocated to disk05-01,
but not yet recovered.
```

Find a suitable target disk in the same disk group, and ensure that the target disk is not already in use by the same volume.
For example, as shown in the sample mail message in the previous step, do not attempt to relocate the data to a disk in use by the volume home. For more information on finding unused space in a disk group, see Section 5.2.1.

Move the objects from the relocation disk to the target disk:

# volevac [-g disk_group] dm_relocation_disk dm_target_disk

5.1.7 Removing Disks from LSM Control

You can remove a disk from LSM control if you no longer need it.

For information on removing an LSM disk from a disk group, see Section 5.2.3. For information on deporting a disk group, see Section 5.2.4.

To remove an LSM disk from LSM control:

Display the disk media name and disk access name of the disk:

# voldisk list

Information similar to the following is displayed (edited for brevity):

DEVICE       TYPE      DISK         GROUP        STATUS
 
.
.
.
dsk25        sliced    -            -            unknown
dsk26        sliced    newdisk      rootdg       online
dsk27        sliced    dsk27        rootdg       online

Remove the disk media name from the disk group:
```
# voldg [-g disk_group] rmdisk disk
```
For example, to remove dsk26, which has a disk media name of newdisk, from the rootdg disk group:
```
# voldg rmdisk newdisk
```
When you remove an LSM disk from a disk group, it no longer has a disk media name. The disk keeps its disk access name until you remove the disk from LSM control. The voldisk list command shows that dsk26 does not belong to any disk group, and it has a status of online, indicating it is under LSM control:
```
DEVICE       TYPE      DISK         GROUP        STATUS
 
.
.
.
dsk25        sliced    -            -            unknown
dsk26        sliced    -            -            online
dsk27        sliced    dsk27        rootdg       online
```

Remove the disk access name from LSM control:
```
# voldisk rm disk
```
For example, to remove disk dsk26:
```
# voldisk rm dsk26
```

All the disk partition labels are changed to unused. To use the disk for other purposes, reinitialize it using the disklabel command. For more information, see disklabel(8).

5.2 Managing Disk Groups

The following sections describe how to use LSM commands to manage disk groups.

5.2.1 Displaying Disk Group Information

To display a list of all LSM disks and the disk group to which each belongs, and disks that LSM recognizes but which are not under LSM control:

# voldisk list

DEVICE       TYPE      DISK         GROUP        STATUS  
dsk0         sliced    -            -            unknown 
dsk1         sliced    -            -            unknown 
dsk2         sliced    dsk2         rootdg       online  
dsk3         sliced    dsk3         rootdg       online  
dsk4         sliced    dsk4         rootdg       online  
dsk5         sliced    dsk5         rootdg       online  
dsk6         sliced    dsk6         dg1          online  
dsk7         sliced    dsk7         dg1          online  
dsk8         sliced    dsk8         dg1          online  
dsk9         sliced    dsk9         dg2          online  
dsk10        sliced    dsk10        dg2          online  
dsk11        sliced    dsk11        dg2          online  
dsk12        sliced    -            -            unknown 
dsk13        sliced    -            -            unknown

To display the free space in one or all disk groups:

# voldg [-g disk_group] free

GROUP        DISK         DEVICE       TAG          OFFSET    LENGTH    FLAGS  
rootdg       dsk2         dsk2         dsk2         2097217   2009151   -  
rootdg       dsk3         dsk3         dsk3         2097152   2009216   -  
rootdg       dsk4         dsk4         dsk4         0         4106368   -  
rootdg       dsk5         dsk5         dsk5         0         4106368   -  
dg1          dsk6         dsk6         dsk6         0         2046748   -  
dg1          dsk8         dsk8         dsk8         0         2046748   -

The value in the LENGTH column indicates the amount of free disk space in 512-byte blocks. (2048 blocks equal 1 MB.)

To display the the largest volume you can create in the disk group, use the volassist maxsize command.
The value returned will vary depending on the properties of the volume you want to create, such as striped or mirrored.
For example, to display the largest volume you can create that is striped over four disks, with a stripe width of 512K bytes:
```
# volassist [-g disk_group] maxsize stwidth=512k ncolumn=4
Maximum volume size: 160051200 (78150Mb)
```
You can then create a volume with those properties, of any size up through that value; for example:
```
# volassist -g dg1 make megavol 78150m stwidth=512k ncolumn=4
```

5.2.2 Adding LSM Disks to Disk Groups

You can add unassigned LSM disks to any disk group. To display a list of unassigned disks, enter the voldisk list command. Unassigned LSM disks are those with a status of online and dashes (-) in the DISK and GROUP columns of the output.

To add one or more LSM disks to an existing disk group:

# voldg -g disk_group adddisk disk [disk...]

For example, to add the disk dsk10 to the disk group dg1:

# voldg -g dg1 adddisk dsk10

To initialize a disk for LSM use and either add it to a disk group or use it to initialize a new disk group in one step, use the voldiskadd script (Section 4.2.1).

5.2.3 Removing LSM Disks from Disk Groups

You can remove an LSM disk from a disk group; however, you cannot remove:

The last disk in a disk group unless the disk group is deported. For information on deporting a disk group, see Section 5.2.4.

Any disk that is in use (for example, disks that contain active LSM volume data). If you attempt to remove a disk that is in use, LSM displays an error message and does not remove the disk.
For information on moving data off an LSM disk, see Section 5.1.5. For information on removing LSM volumes, see Section 5.4.6.

To remove an LSM disk from a disk group:

Verify that the LSM disk is not in use by listing all subdisks:

# volprint -st

Disk group: rootdg
 
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
sd dsk1-01      klavol-01    dsk1     0        1408     0/0       dsk1     ENA
sd dsk2-02      klavol-03    dsk2     0        65       LOG       dsk2     ENA
sd dsk2-01      klavol-01    dsk2     65       1408     1/0       dsk2     ENA
sd dsk3-01      klavol-01    dsk3     0        1408     2/0       dsk3     ENA
sd dsk4-01      klavol-02    dsk4     0        1408     0/0       dsk4     ENA
sd dsk5-01      klavol-02    dsk5     0        1408     1/0       dsk5     ENA
sd dsk6-01      klavol-02    dsk6     0        1408     2/0       dsk6     ENA

The disks in the DISK column are currently in use by an LSM volume, and therefore you cannot remove those disks from a disk group.

Remove the LSM disk from the disk group:
```
# voldg -g disk_group rmdisk disk
```
For example, to remove the LSM disk dsk8 from the rootdg disk group:
```
# voldg rmdisk dsk8
```

The disk remains under LSM control. You can:

Add the LSM disk to a different disk group (Section 5.2.2).

Use the disk to create a new disk group (Section 4.2).

Remove the disk from LSM control (Section 5.1.7).

5.2.4 Deporting Disk Groups

Deporting a disk group makes its volumes inaccessible. You can deport a disk group to:

Rename the disk group.

Reuse the disks for other purposes.

Move the disk group to another system (Section 7.3.2).

You cannot deport the rootdg disk group.

Caution

The voldisk list command displays the disks in a deported disk group as available (with a status of online). However, removing or reusing the disks in a deported disk group can result in data loss.

To deport a disk group:

If applicable, stop the volumes:
```
# volume -g disk_group stopall
```

Deport the disk group:
- To deport the disk group with no changes:
```
# voldg deport disk_group
```
- To deport the disk group and assign it a new name:
```
# voldg [-n newname] deport disk_group
```
  For more information on assigning a new name to a disk group, see voldg(8).
You must import a disk group (Section 5.2.5) before you can use it.
If you no longer need the disk group, you can:
- Add the disks to different disk groups (Section 5.2.2).
- Use the disks to create new disk groups (Section 4.2).
- Remove the disks from LSM control (Section 5.1.7).

5.2.5 Importing Disk Groups

Importing a disk group makes the disk group and its volumes accessible. You cannot import a disk group if you used any of its associated disks while it was deported.

To import a disk group and restart its volumes:

Import the disk group:
```
# voldg import disk_group
```

Start all volumes within the disk group:
```
# volume -g disk_group startall
```

5.2.6 Renaming Disk Groups

Renaming a disk group involves deporting and then importing the disk group. You cannot rename a disk group while it is in use. All activity on all volumes in the disk group must stop, and the volumes in the disk group are inaccessible while the disk group is deported.

Because renaming a disk group involves an interruption of service to the volumes, you should perform this task during a planned shutdown or maintenance period. Choose the new disk group name carefully, and ensure that the new name is easy to remember and use. Renaming a disk group updates the /etc/fstab file.

Note

You cannot rename the rootdg disk group.

To rename a disk group:

Deport the disk group, assigning it a new name (Section 5.2.4).

Import the disk group using its new name (Section 5.2.5).

5.2.7 Creating a Clone Disk Group

The volclonedg command lets you create a copy of a disk group using disks that are hardware clones. This command is available on both standalone systems and clusters.

Note

You must create hardware clones of all the disks in a disk group before cloning the disk group. The disk group cannot contain any nopriv disks.

The volclonedg command creates a new disk group containing the same configuration of LSM objects as the parent disk group, using the the volsave and volrestore commands to save the configuration of the parent disk group and create the same configuration in the clone disk group. LSM starts all possible volumes in the clone disk group, performing recovery of mirrored volumes if necessary. For more information, see volclonedg(8).

To clone a disk group:

Display the disks in the disk group to be cloned:

# voldisk -g disk_group list

DEVICE       TYPE      DISK         GROUP        STATUS
dsk10        sliced    dsk10        dg1          online
dsk11        sliced    dsk11        dg1          online
dsk12        sliced    dsk12        dg1          online
dsk13        sliced    dsk13        dg1          online

Note

If the disk group contains any nopriv disks, add new sliced or simple disks to the disk group, move the data from the nopriv disks to the new disks (using the volevac command, for example), and remove the nopriv disks from the disk group before you clone the disk group.

Create hardware clones of the disks. See your hardware documentation for more information.

Run the hwmgr command to update the system or cluster with the new disk information. For more information, see hwmgr(8).

Verify that LSM can access and display the cloned disks:

# voldisk list

Cloned disks show a status of online aliased.

In the following output, the original disks are dsk10, dsk11, dsk12, and dsk13. The hardware disk clones are dsk14, dsk15, dsk16, and dsk17.

DEVICE       TYPE      DISK         GROUP        STATUS
dsk0         sliced    dsk0         rootdg       online
dsk1         sliced    dsk1         rootdg       online
dsk2         sliced    dsk2         rootdg       online spare
dsk3         sliced    dsk3         rootdg       online spare
dsk4         sliced    dsk4         rootdg       online
dsk5         sliced    dsk5         rootdg       online spare
dsk6         sliced    dsk6         rootdg       online
dsk7         sliced    dsk7         rootdg       online
dsk8         sliced    dsk8         rootdg       online
dsk9         sliced    dsk9         rootdg       online
dsk10        sliced    dsk10        dg1          online
dsk11        sliced    dsk11        dg1          online
dsk12        sliced    dsk12        dg1          online
dsk13        sliced    dsk13        dg1          online
dsk14        sliced    -            -            online aliased
dsk15        sliced    -            -            online aliased
dsk16        sliced    -            -            online aliased
dsk17        sliced    -            -            online aliased

Use the names of the disk clones to clone the disk group dg1, optionally assigning a name other than the default (in this case, dg1_clone):
```
# volclonedg -g dg1 [-N name] dsk14 dsk15 dsk16 dsk17
```
LSM creates the clone disk group and starts its volumes.

For more information, including an example of creating a clone disk group on a different system from the parent disk group, see volclonedg(8).

5.3 Managing the LSM Configuration Database

This section describes how to manage the LSM configuration database, including:

Backing up the configuration database (Section 5.3.1)

Restoring the configuration database from backup (Section 5.3.2)

Changing the size and number of configuration database copies (Section 5.3.3)

5.3.1 Backing Up the LSM Configuration Database

Use the volsave utility to periodically create a copy of the LSM configuration. You can then use the volrestore command to recreate the LSM configuration if you lose a disk group configuration.

The saved configuration database (also called a description set) is a record of the objects in the LSM configuration (the LSM disks, subdisks, plexes, and volumes) and the disk group to which each object belongs.

Whenever you make a change to the LSM configuration, the backup copy becomes obsolete. Like any backup, the content is useful only as long as it accurately represents the current information. Any time the number, nature, or name of LSM objects change, consider making a backup of the LSM configuration database.

The following list describes some of the changes that will invalidate a configuration database backup:

Creating disk groups

Adding or removing disks from disk groups or from LSM control

Creating or removing volumes

Changing the properties of volumes, such as the plex layout or number of logs

Note

Backing up the configuration database does not save the data in the volumes. For information on backing up volume data, see Section 5.4.2.
The volsave command does not save information relating to volumes used for the root, /usr, or /var file systems or for swap space.
Depending on the nature of a boot disk failure, you might need to restore the system partitions from backups or installation media to return to a state where the system partitions are not under LSM control. From there, you can redo the procedures to encapsulate the boot disk partitions to LSM volumes and add mirror plexes to those volumes.
For more information about recovering from a boot disk failure under LSM control, see Section 6.4.6.

By default, LSM saves the entire configuration database to a time-stamped directory called /usr/var/lsm/db/LSM.date.hostname. You can specify a different directory for the backup, but the directory must not already exist.

The backup directory contains the following files and directories:

A copy of the volboot file.

A file called header, which contains host ID and checksum information, and a list of the other files in this directory.

A file called voldisk.list, which contains a list of all LSM disks, their type (sliced, simple, nopriv), the size of their private and public regions, their disk group, and other information.

A subdirectory called rootdg.d, which contains the allvol.DF file. The allvol.DF file contains detailed descriptions of every LSM subdisk, plex, and volume, describing all their properties and attributes.

To back up the LSM configuration database:

Enter the following command, optionally specifying a directory location other than the default to store the LSM configuration database:
```
# volsave [-d directory]
```

Save the backup to tape or other removable media.

The volsave command saves multiple versions of the configuration database; each new backup is saved in the /usr/var/lsm/db directory with its own date and time stamp, as shown in the following example:

dr-xr-x---   3 root     system      8192 May  5 09:36 LSM.20000505093612.hostname
dr-xr-x---   3 root     system      8192 May 10 10:53 LSM.20000510105256.hostname

5.3.2 Restoring the LSM Configuration Database from Backup

The volrestore command restores an LSM configuration database, provided you saved it with the volsave command (Section 5.3.1).

You can restore the configuration database of a specific disk group or volume or the entire configuration (all disk groups and volumes except those associated with the boot disk). If you have multiple backups of the configuration database (a new one is created each time you run the volsave command), you can choose a specific one to restore. Otherwise, LSM restores the most recent version.

Note

Restoring the configuration database does not restore data in the LSM volumes. For information on restoring volumes, see Section 5.4.3.
The volrestore command does not restore volumes associated with the root (/), /usr, and /var file systems and the primary swap area on a standalone system. If volumes for these partitions are corrupted or destroyed, you must reencapsulate the system partitions to use LSM volumes.

To restore a backed-up LSM configuration database:

Optionally, display a list of all available database backups:
```
# ls /usr/var/lsm/db
```
If you saved the configuration database to a different directory, specify that directory.

Restore the chosen configuration database:
- To restore the entire configuration database:
```
# volrestore [-d directory]
```
- To restore a specific disk group configuration database:
```
# volrestore [-d directory] -g disk_group
```
- To restore a specific volume configuration:
```
# volrestore [-d directory] -v volume
```
- To restore a configuration database interactively, enabling you to select or skip specific objects:
```
# volrestore [-d directory] -i
```

Start the restored LSM volumes:
```
# volume -g disk_group startall
```
If the volumes will not start, you might need to manually edit the plex state. See Section 6.5.2.2.

If necessary, restore the volume data from backup. For more information, see Section 5.4.3.

5.3.3 Changing the Size and Number of Configuration Database Copies

LSM maintains copies of the configuration database on separate physical disks within each disk group. When the disk group runs out of space in the configuration database, LSM displays the following message:

volmake: No more space in disk group configuration

This can happen if:

One or more disks in the disk group contain two copies of the configuration database. Whenever a configuration change occurs, all active copies are updated. If one disk's copies cannot be updated because they have grown too large, then none of the copies for the whole disk group can be updated.

The disk group came from a system running Tru64 UNIX Version 4.0, which uses a smaller default private region size.

If the configuration database runs out of space and you find that some disks have two copies of the configuration database, you can remove one copy from each disk that has two (Section 5.3.3.1). However, make sure that the disk group still has sufficient copies of the configuration database available for redundancy. For example, if the disk group has a total of four copies and two are on the same disk, remove one copy from that disk and enable a copy on another disk that does not have one.

If all copies of the configuration database are the same size and no disk has more than one copy, this might indicate that the private regions of the disks are too small. For example, the disks were initialized on a system running an earlier version of LSM, with a smaller default private region. To resolve this problem, add new disks to LSM, which will have the larger default private region size, add the new disks to the disk group, and delete the copies of the configuration database on the other disks (Section 5.3.3.2).

5.3.3.1 Reducing the Number of Configuration Database Copies on an LSM Disk

To reduce the number of configuration database copies on an LSM disk:

Display information about the disk group's configuration database:

# voldg list disk_group

Group:   rootdg
dgid:   783105689.1025.lsm
import-id: 0.1
flags:
config:  seqno=0.1112 permlen=173 free=166 templen=6 loglen=26
config disk dsk13 copy 1 len=173 state=clean online
config disk dsk13 copy 2 len=173 state=clean online
config disk dsk11g copy 1 len=347 state=clean online
config disk dsk10g copy 1 len=347 state=clean online
log disk dsk11g copy 1 len=52
log disk dsk13 copy 1 len=26
log disk dsk13 copy 2 len=26
log disk dsk10g copy 1 len=52

Identify a disk that has multiple copies of the configuration database.
For example, disk dsk13 has two copies of the configuration database. This halves the total configuration space available in memory for the disk group and is therefore the limiting factor.

Reduce the number of copies on a disk that has two copies:
```
# voldisk moddb disk nconfig=n
```
For example, to reduce the number of configuration copies on dsk13 from two to one:
```
# voldisk moddb dsk13 nconfig=1
```

Display the new configuration:
```
# voldg list disk_group
```

If necessary, add a copy to another disk to maintain the appropriate number of copies for the disk group:
1. Display a list of all disks in the disk group:
```
# voldisk -g disk_group list
```
2. Compare the disks listed in the output of the voldisk list command to those listed in the output of the voldg list command to identify a disk in the disk group that does not have a copy of the configuration database.
3. Enable a configuration database copy on a disk that does not have one, using the disk access name:
```
# voldisk moddb disk_access_name nconfig=1
```

5.3.3.2 Removing Configuration Database Copies on LSM Disks with Small Private Regions

To remove configuration database copies on an LSM disk with a small private region:

Display information about the disk group's configuration database:

# voldg list disk_group

Group:     rootdg
dgid:      921610896.1026.hostname
import-id: 0.1  flags:
copies:    nconfig=default nlog=default
config:    seqno=0.1081 permlen=347 free=341 templen=3 loglen=52
config disk dsk7 copy 1 len=347 state=clean online
config disk dsk8 copy 1 len=2993 state=clean online
config disk dsk9 copy 1 len=2993 state=clean online
config disk dsk10 copy 1 len=2993 state=clean online
log disk dsk7 copy 1 len=52
log disk dsk8 copy 1 len=453
log disk dsk9 copy 1 len=453
log disk dsk10 copy 1 len=453

Disk dsk7 has a smaller private region than the other disks, as shown by the len= information in the lines beginning with config disk and log disk, and therefore has less space to store copies of the configuration database and log. This restricts the disk group's ability to store additional records, because the smallest private region sets the limit for the group.

Remove all configuration database copies from the disk with the smallest private region:
```
# voldisk moddb disk nconfig=0
```
For example, to remove the copies on the disk with the smallest private region, dsk7:
```
# voldisk moddb dsk7 nconfig=0
```

Display the new configuration:
```
# voldg list disk_group
```

If necessary, add a copy to another disk to maintain the appropriate number of copies for the disk group:
1. Display a list of all disks in the disk group:
```
# voldisk -g disk_group list
```
2. Compare the disks listed in the output of the voldisk list command to those listed in the output of the voldg list command to identify a disk in the disk group that does not have a copy of the configuration database.
3. Enable a copy on a disk that does not have one, using the disk access name:
```
# voldisk moddb disk_access_name nconfig=1
```

5.4 Managing LSM Volumes

The following sections describe how to use LSM commands to manage LSM volumes. For information on creating LSM volumes, see Chapter 4.

5.4.1 Displaying LSM Volume Information

The volprint command displays information about LSM objects, including LSM disks, subdisks, plexes, and volumes.

To display the complete hierarchy of objects in an LSM volume:

# volprint [-g disk_group] -ht volume

Disk group: rootdg   [1]
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  data01       fsgen        ENABLED  ACTIVE   512000   SELECT    -   [2]
pl data01-01    data01       ENABLED  ACTIVE   512256   STRIPE    3/128    RW   [3]
sd dsk2-01      data01-01    dsk2     0        170752   0/0       dsk2     ENA  [4]
sd dsk5-01      data01-01    dsk5     0        170752   1/0       dsk5     ENA
sd dsk6-01      data01-01    dsk6     0        170752   2/0       dsk6     ENA
pl data01-02    data01       ENABLED  ACTIVE   512256   STRIPE    3/128    RW
sd dsk7-01      data01-02    dsk7     0        170752   0/0       dsk7     ENA
sd dsk8-01      data01-02    dsk8     65       170752   1/0       dsk8     ENA
sd dsk9-01      data01-02    dsk9     0        170752   2/0       dsk9     ENA
pl data01-03    data01       ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk8-02      data01-03    dsk8     0        65       LOG       dsk8     ENA

The preceding example shows output for a volume with mirrored, three-column, striped plexes:

Disk group name. [Return to example]

Volume information: Name (data01), usage type (fsgen), state (ENABLED ACTIVE), and size (51200 blocks, or 250 MB). [Return to example]

Plex information: Two data plexes (data01-01 and data01-02) and a DRL plex (data01-03). [Return to example]

Subdisk information for each plex. [Return to example]

To display a listing of LSM volumes:

# volprint [-g disk_group] -vt

Disk group: rootdg
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
 
v  rootvol      root         ENABLED  ACTIVE   524288   ROUND     -
v  swapvol      swap         ENABLED  ACTIVE   520192   ROUND     -
v  vol-dsk24c   fsgen        ENABLED  ACTIVE   17773524 SELECT    -
v  vol-dsk25g   fsgen        ENABLED  ACTIVE   2296428  SELECT    -
v  vol-dsk25h   fsgen        ENABLED  ACTIVE   765476   SELECT    -
v  vol-01       fsgen        ENABLED  ACTIVE   768000   ROUND     -
v  vstripe      fsgen        ENABLED  ACTIVE   256000   SELECT    vstripe-01

5.4.2 Backing Up LSM Volumes

One of the more common tasks of a system administrator is helping users recover lost or corrupted files. To perform that task effectively, establish procedures for backing up LSM volumes and the LSM configuration database at frequent and regular intervals. You will need the saved configuration database as well as the backed-up data if you need to restore a volume after a major failure. (For example, if multiple disks in the same volume failed, or all the disks containing the active configuration records for a disk group failed.)

For information on backing up the LSM configuration database, see Section 5.3.1.

For a thorough discussion of the backup and restore options available on a Tru64 UNIX system, see the System Administration manual. If you are using AdvFS, also see the AdvFS Administration manual.

LSM commands do not actually create the backup of the volume data, but they provide several ways for you to make volume data available to back up. In many cases, you do not have to unmount file systems or bring the system to single-user mode, which prevents possible corruption of the backup data if users write to a file system before the backup is complete.

To create a backup of LSM volume data, you can use the following commands:

volassist snapfast and volassist snapback
These commands use the Fast Plex Attach feature to create a secondary volume from an existing eligible plex in a mirrored volume. They create a log that keeps track of changes to the original (primary) volume from the time a plex is detached with snapfast to the time the plex is reattached to the original volume with snapback.
Advantages — These commands save time by using an existing eligible plex and reduce the time required for the returning plex to resynchronize to the original volume.
Disadvantages — These commands remove one plex to create the temporary volume; they might leave the primary volume unmirrored until the snapback operation occurs.
For a detailed description of the Fast Plex Attach feature and information on how to use it, see Section 5.4.2.1 and Section 5.4.2.2.

volassist snapstart and volassist snapshot
These commands attach a new plex to a volume for the express purpose of creating a temporary volume for backups.
The snapstart keyword uses available space in the disk group to create and attach a new plex to the volume and performs a full synchronization of the new plex to the volume's contents.
The snapshot keyword uses the new plex to create a temporary volume with the name you specify. The temporary volume is the one you use to perform your backups; this leaves the original volume running and available for use.
After performing the backup, you can optionally detach the plex from the temporary volume and reattach it to the original volume. This performs a full resynchronization.
Advantages — These commands do not reduce the number of mirrors in a volume; they use a new plex created specifically for the backup. You can use this method to mirror a nonmirrored volume (except RAID5), if sufficient disk space is available.
Disadvantages — These commands use additional disk space (equal to the size of the volume), and take time for the new plex to fully synchronize to the original volume and (optionally) for the reattached plex to resynchronize to the original volume.
For more information, see Section 5.4.2.3.

volplex det, volmake, volume start, voledit, and volplex att (low-level commands)
These commands use an existing eligible plex from a mirrored volume to create the backup volume.
Advantages — These commands give you control over which plex to use and save time by using an existing eligible plex.
Disadvantages — These commands are considered low-level commands, introducing more chance of error. They remove one plex to create the temporary volume, which might leave the original volume unmirrored. If you reattach the plex to the original volume, it takes time for the plex to resynchronize.
For more information, see Section 5.4.2.4.

To back up a RAID 5 volume or a volume that use a single concatenated or striped plex that you cannot or choose not to mirror, see Section 5.4.2.5.

5.4.2.1 Overview of the Fast Plex Attach Feature

For mirrored LSM volumes, you can use the Fast Plex Attach feature to make a temporary copy of the volume data available for backup. You use the temporary volume to perform your backups, leaving the original volume running and available for use.

You can use the Fast Plex Attach feature on any mirrored volume on a standalone system or a cluster, including the rootvol, cluster_rootvol, and other volumes for encapsulated standalone system partitions and clusterwide file system domains. The Fast Plex Attach feature cannot be used on mirrored volumes used as swap space.

The volassist command provides two keywords (snapfast and snapback) to create and remove a backup volume using the Fast Plex Attach feature.

Figure 5-1 shows a three-way mirrored volume before a Fast Plex Attach operation.

Figure 5-1: Volume Before Fast Plex Attach

A complete Fast Plex Attach operation goes through the following phases:

When you run the volassist snapfast command, LSM examines the volume (known as the primary volume) to ensure it has at least two complete, read-write plexes and arbitrarily selects one plex as a candidate for Fast Plex Attach (FPA) support. This plex becomes the migrant plex. If any plex has a state of SNAPDONE (the result of a prior snapstart operation), LSM uses it as the migrant plex.
If the volume has only one plex, the command fails. If the volume has only two plexes, the command exits and displays a message stating that you must use the force (-f) option, because the operation will leave the volume with only one plex and therefore unmirrored.
LSM creates an FPA subdisk and attaches it to a separate FPA plex (plex-05) for the primary volume (Figure 5-2).
Figure 5-2: Process of the volassist snapfast Command: Phase 1

LSM creates an FPA subdisk for the migrant plex and attaches this to the migrant plex as its FPA log. LSM then detaches the migrant plex from the primary volume and creates a secondary volume (Figure 5-3).
Figure 5-3: Process of the volassist snapfast Command: Phase 2

LSM uses available disk space in the disk group for both FPA logs, using disks marked as hot-spares only if no other suitable space is available. LSM will not use space on disks marked as reserved or volatile. (For a description of these attributes, see voledit(8) and voldisk(8).)
As writes occur to both volumes, their respective FPA logs keep track of the regions that changed (Figure 5-4). The FPA log subdisk attached to the migrant plex keeps track of changes to the secondary volume (such as the I/O that occurs when you mount the secondary volume so you can back it up).
Figure 5-4: Writes Occurring to Primary and Secondary LSM Volumes

Back up the secondary volume.

When the backup is complete, reattach the migrant plex to the primary volume with the volassist snapback command.

LSM removes the FPA log subdisk from the secondary volume and merges it with the FPA log subdisk on the primary volume (Figure 5-5).
Figure 5-5: Process of volassist snapback Command: Phase 1

LSM reattaches the migrant plex to the primary volume (and removes the secondary volume if it has no other data plexes) and starts the process of resynchronizing the migrant plex to the primary volume according to the regions marked in the merged FPA log (Figure 5-6).
Figure 5-6: Process of volassist snapback Command: Phase 2

Any writes that occurred to the secondary volume are ignored; the corresponding regions of the primary volume are written to the returning migrant plex. Also, only the regions of the primary volume that changed in the interim are written to the returning migrant plex, instead of resynchronizing the entire volume. This can greatly reduce the time it takes for a volume to resynchronize and thereby reduce the performance impact.
Depending on how long the migrant plex was away and how active the primary volume was during its absence, resynchronizing only the regions marked in each volume's logs is likely to be faster than resynchronizing the entire volume.

5.4.2.2 Creating a Backup Volume Using the Fast Plex Attach Feature (volassist snapfast and snapback)

You can use the Fast Plex Attach feature to create a temporary backup volume from one plex of a mirrored volume. This is sometimes called splitting a mirror, but is more accurately described as detaching a mirror. The volume must have at least two complete, enabled plexes before you begin. If the volume has only two plexes, the volume will not be mirrored during the time one plex is detached. You must use the force (-f) option to detach a plex in this case.

Note

When you use the Fast Plex Attach feature to create a backup volume for rootvol, which has the special usage type of root, LSM creates the backup volume with the usage type of gen. Only the rootvol volume can have a usage type of root. In output from the volprint command, you might notice the different usage type for the backup volume for rootvol. The usage type does not affect backup operations.

The Fast Plex Attach feature does not let you specify a particular plex to use as the migrant plex. To control which plex is used, you can:

Run the volassist snapstart command. When complete, the volume will have a new plex with a state of SNAPDONE. You can then run the volassist snapfast command and LSM will use the plex marked SNAPDONE as the migrant plex.
There must be available space in the disk group equal to the size of the volume to use the volassist snapstart command.

Use the following commands to detach a specific plex from the primary volume, enable Fast Plex Attach logging on both the primary volume and the migrant plex, and create the secondary volume from that plex:
```
# volplex det -o fpa plex
# volplex att plex secondary_volume
```
Back up the secondary volume as described in the following procedure. When finished, you can use the volassist snapback command as shown in step 4.

To back up a volume using the Fast Plex Attach feature:

Verify that the volume you want to back up has more than two complete, enabled plexes available; for example:

# volprint -vht 3wayvol

Disk group: rootdg
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  3wayvol      fsgen        ENABLED  ACTIVE   1024     SELECT    -
pl 3wayvol-01   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk0-01      3wayvol-01   dsk0     0        1024     0         dsk0     ENA
pl 3wayvol-02   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk2-01      3wayvol-02   dsk2     0        1024     0         dsk2     ENA
pl 3wayvol-03   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk7-02      3wayvol-03   dsk7     128      1024     0         dsk7     ENA
pl 3wayvol-04   3wayvol      ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk1-01      3wayvol-04   dsk1     0        65       LOG       dsk1     ENA

Enter the following command to enable Fast Plex Attach logging on the primary volume, detach one plex using the force (-f) option if necessary, and create the secondary volume with the name of your choice:

# volassist [-f] snapfast primary_vol secondary_vol

For example:

# volassist -f snapfast 3wayvol 3wayvol_bk

LSM creates a secondary volume from one plex of the primary volume. The volumes look like the following:

v  3wayvol      fsgen        ENABLED  ACTIVE   1024     SELECT    -
pl 3wayvol-01   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk0-01      3wayvol-01   dsk0     0        1024     0         dsk0     ENA
pl 3wayvol-02   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk2-01      3wayvol-02   dsk2     0        1024     0         dsk2     ENA
pl 3wayvol-04   3wayvol      ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk1-01      3wayvol-04   dsk1     0        65       LOG       dsk1     ENA
pl 3wayvol-05   3wayvol      ENABLED  ACTIVE   FPAONLY  CONCAT    -        RW
sd dsk4-05      3wayvol-05   dsk4     524546   65       FPA       dsk4     ENA
 
v  3wayvol_bk   fsgen        ENABLED  ACTIVE   1024     ROUND     -
pl 3wayvol-03   3wayvol_bk   ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk0-03      3wayvol-03   dsk0     1152     65       FPA       dsk0     ENA
sd dsk7-02      3wayvol-03   dsk7     128      1024     0         dsk7     ENA

Back up the secondary volume using your preferred backup method. For more information, see Section 5.4.2.6.

When the backup is complete, reattach the migrant plex to the primary volume; for example:

# volassist snapback 3wayvol_bk 3wayvol

If the secondary volume has no other plexes, it is removed completely.

The FPA log plex remains attached to the primary volume to support future snapfast operations, as shown:

# volprint

v  3wayvol      fsgen        ENABLED  ACTIVE   1024     SELECT    -
pl 3wayvol-01   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk0-01      3wayvol-01   dsk0     0        1024     0         dsk0     ENA
pl 3wayvol-02   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk2-01      3wayvol-02   dsk2     0        1024     0         dsk2     ENA
pl 3wayvol-03   3wayvol      ENABLED  ACTIVE   1024     CONCAT    -        RW
sd dsk7-02      3wayvol-03   dsk7     128      1024     0         dsk7     ENA
pl 3wayvol-04   3wayvol      ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk1-01      3wayvol-04   dsk1     0        65       LOG       dsk1     ENA
pl 3wayvol-05   3wayvol      ENABLED  ACTIVE   FPAONLY  CONCAT    -        RW
sd dsk4-05      3wayvol-05   dsk4     524546   65       FPA       dsk4     ENA

5.4.2.3 Creating a Backup Volume by Attaching a New Plex (volassist snapstart and snapshot)

The following procedure describes how to add a plex to a volume (mirror the volume) and then detach the new plex to create a separate volume for backups. Detaching a plex is sometimes called splitting a mirror; in this case, you create the new mirror to detach.

This backup method uses available disk space in the disk group for the new plex, equal to the size of the volume, and requires extra time for the new plex to be fully synchronized to the volume before you can detach it to make the backup volume. However, this backup method eliminates the need to detach a plex from a volume and therefore preserves redundancy for a mirrored volume.

If you do not have sufficient disk space or do not want to mirror the volume, see Section 5.4.2.5.

To back up an LSM volume by adding and then detaching a new plex:

Display the size of the LSM volume and the disks it uses:
```
# volprint -v [-g disk_group] volume
```

Ensure there is enough free space in the disk group to add a plex to the LSM volume:
```
# voldg [-g disk_group] free
```
The amount of free space must be at least equal to the size of the volume and must be on disks that are not used in the volume you want to back up.

Add a new plex to the volume, optionally specifying the disks to use:
```
# volassist snapstart volume [disk...]
```
This step initiates a full synchronization, which might take several minutes or longer, depending on the size of the volume.

Create a temporary volume from the new plex. (The snapshot keyword uses the plex created in step 4 to create the new volume.)
```
# volassist snapshot volume temp_volume
```
The following example creates a temporary volume named vol3_backup for a volume named vol3:
```
# volassist snapshot vol3 vol3_backup
```

Start the temporary volume:
```
# volume start temp_volume
```

Back up the temporary volume using your preferred backup method. For more information, see Section 5.4.2.6.

When the backup is complete, do one of the following:
- If you no longer need the backup volume, stop and remove the volume:
```
# volume stop temp_volume
# voledit -r rm temp_volume
```
- If you want to reattach the plex to the original volume:
  1. Stop the backup volume:
```
# volume stop temp_volume
```
  2. Dissociate the plex from the backup volume (requires the force [-f] option):
```
# voledit -f dis plex
```
  3. Remove the now empty backup volume:
```
# voledit -o rm temp_volume
```
  4. Attach the backup plex to the original volume:
```
# volplex att plex original_volume
```
    This initiates a full resynchronization of the volume's contents to the plex.

5.4.2.4 Creating a Backup Volume by Detaching an Existing Plex

This procedure is recommended only for adminstrators who have experience with using the low-level LSM commands and who have particular reasons for using this method instead of those described in the previous sections.

If used on a volume with only two enabled plexes, this procedure leaves the original volume unmirrored while one of the plexes is detached.

To back up an LSM volume from one of its existing plexes using the low-level commands:

Dissociate one of the volume's complete, enabled plexes, which leaves the plex with an image of the LSM volume at the time of dissociation:
```
# volplex dis plex
```
For example:
```
# volplex dis data-02
```

Create a temporary LSM volume using the dissociated plex:

# volmake -U fsgen vol temp_volume plex=plex

For example:

# volmake -U fsgen vol data_temp plex=data-02

Start the temporary volume:

# volume start temp_volume

For example:

# volume start data_temp

Back up the temporary volume using your preferred backup method. For more information, see Section 5.4.2.6.

When the backup is complete, stop and remove the temporary volume:
```
# volume stop temp_volume
# voledit -r rm temp_volume
```

Reattach the dissociated plex to the original volume.
```
# volplex att volume plex
```
LSM automatically resynchronizes the plexes when you reattach the dissociated plex. This operation might take a long time, depending on the size of the volume. Running this process in the background returns control of the system to you immediately instead of waiting until the resynchronization is complete.

5.4.2.5 Backing Up a Nonredundant or RAID 5 Volume

If the volume uses a RAID 5 plex layout or if you cannot add a mirror to a volume with a single striped or concatenated data plex, you must either stop all applications from using the volume while the backup is in process or allow the backup to occur while the volume is in use.

If the volume remains in use during the backup, the volume data might change before the backup completes, and therefore the backup data will not be an exact copy of the volume's contents. The following procedure stops the volume to eliminate the risk of data corruption in the backup.

To back up a nonredundant or RAID 5 volume:

If applicable, select a convenient time and inform users to save files and refrain from using the volume, the application, or file system that uses the volume while you back it up.

Stop the volume:

# volume stop volume

For example:

# volume stop r5_vol

Back up the volume using your preferred backup method. For more information, see Section 5.4.2.6.

When the backup is complete, restart the volume:
```
# volume start volume
```
For example:
```
# volume start r5_vol
```

If applicable, inform users that the volume is available.

5.4.2.6 Performing a Backup

You can use any of the following methods to back up the data in an LSM volume:

If the volume contains a UFS file system, you can use the dump command; for example:
```
# dump -0u /dev/rvol/rootdg/r5_vol
```
For more information, see dump(8).

If the volume contains raw data such as for a database application, use the application's built-in backup utility. Be careful to select the temporary volume name as the object to back up.

If the application has no backup utility, you can use the dd command to make an image copy of the volume's contents; for example:
```
# dd if=/dev/rvol/rootdg/3wayvol_bk of=/dev/tape/tape0_d0
```
For more information, , including options for specifying input and output block sizes, see dd(8).

5.4.2.7 Special Case: Backing Up LSM Volumes in an AdvFS Domain

Because AdvFS domains can use storage on several devices, including several LSM volumes, it is critical for backups to capture the state of the domain, filesets, and the AdvFS metadata at the same point in time.

Caution

Before you begin, read the information on backing up and restoring data in the AdvFS Administration manual.

The following procedure involves temporarily freezing the domain to ensure the metadata is in a consistent state until after you create the backup LSM volumes. To minimize the time the domain must remain frozen, review Section 5.4.2.2, Section 5.4.2.3, Section 5.4.2.4, and Section 5.4.2.5, as appropriate, to determine what you must do to create each backup volume. You can perform some tasks ahead of time, such as finding free space in a disk group and adding a mirror to a volume.

To back up an AdvFS domain that uses several LSM mirrored volumes:

Freeze the domain. (The default freeze period is 60 seconds.) Freezing any mount point in the domain freezes the entire domain.
```
# /usr/sbin/freezefs /mount_point
```

For each LSM volume in the domain, create a backup volume using the appropriate procedure.

If the freeze time has not already elapsed, thaw the domain:
```
# /usr/sbin/thawfs /mount_point
```

Create a new domain directory and link the newly created LSM backup volumes to the temporary domain; for example:
```
# mkdir /etc/fdmns/my_dom_BK
# ln -s /dev/vol/rootdg/vol_1_backup /etc/fdmns/my_dom_BK
# ln -s /dev/vol/rootdg/vol_2_backup /etc/fdmns/my_dom_BK
```
Do not use the mkfdmn command to create the domain on the LSM volumes; this initializes the backup LSM volumes and destroys the existing data in the volumes.

Create a temporary backup directory at the root directory level:
```
# mkdir /backup
```

Display the filesets in the domain you are backing up:

# showfsets /etc/fdmns/my_dom

my_files
        Id           : 3caa0e34.000ef531.1.8001
        Files        :      0,  SLim=      0,  HLim=     0
        Blocks (512) :      0,  SLim=      0,  HLim=     0
        Quota Status : user=off group=off
        Object Safety: off
        Fragging     : on
        DMAPI        : off

Dual-mount the filesets in the new domain to the temporary backup mount point:

# mount -o dual temp_domain#parent_fileset /backup

For example:

# mount -o dual my_dom_BK#my_files /backup

Perform the backup on the temporary domain:

# vdump [options] [backup_device] /backup

When the backup is complete, unmount the temporary domain, and stop and remove the LSM volumes:
```
# umount /backup
# volume stop backup_volume...
# voledit -fr rm backup_volume...
```

The following example shows the entire process:

The original domain is called data_dmn with a fileset called data_files, mounted on /data.

The domain uses two mirrored volumes called vol_1 and vol_2.

The backup LSM volumes are called vol_1_backup and vol_2_backup.

The backup domain is called data_dmn_bk, mounted on /backup.

# volprint -vht vol_1 vol_2
Disk group: rootdg
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  vol_1        fsgen        ENABLED  ACTIVE   262144   SELECT    -
pl vol_1-01     vol_1        ENABLED  ACTIVE   262144   CONCAT    -        RW
sd dsk0-01      vol_1-01     dsk0     0        262144   0         dsk0     ENA
pl vol_1-02     vol_1        ENABLED  ACTIVE   262144   CONCAT    -        RW
sd dsk2-01      vol_1-02     dsk2     0        262144   0         dsk2     ENA
pl vol_1-03     vol_1        ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk1-01      vol_1-03     dsk1     0        65       LOG       dsk1     ENA
 
v  vol_2        fsgen        ENABLED  ACTIVE   262144   SELECT    -
pl vol_2-01     vol_2        ENABLED  ACTIVE   262144   CONCAT    -        RW
sd dsk3-01      vol_2-01     dsk3     0        262144   0         dsk3     ENA
pl vol_2-02     vol_2        ENABLED  ACTIVE   262144   CONCAT    -        RW
sd dsk5-01      vol_2-02     dsk5     0        262144   0         dsk5     ENA
pl vol_2-03     vol_2        ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
sd dsk1-02      vol_2-03     dsk1     65       65       LOG       dsk1     ENA
 
# /usr/sbin/freezefs /data_1
# volassist -f snapfast vol_1 vol_1_backup
# volassist -f snapfast vol_2 vol_2_backup
# /usr/sbin/thawfs /ka
# cd /etc/fdmns
# mkdir data_dmn_bk
# ln -s /dev/vol/rootdg/vol_1_backup /etc/fdmns/data_dmn_bk
# ln -s /dev/vol/rootdg/vol_2_backup /etc/fdmns/data_dmn_bk
# mkdir /backup
# showfsets ka_dom
data_files1
        Id           : 3cadbee2.000aca16.1.8001
        Files        :        5,  SLim=        0,  HLim=        0
        Blocks (512) :    29340,  SLim=        0,  HLim=        0
        Quota Status : user=off group=off
        Object Safety: off
        Fragging     : on
        DMAPI        : off
 
# mount -o dual data_dmn_bk#data_files1 /backup
# vdump [options] [backup_device] /backup

5.4.3 Restoring LSM Volumes from Backup

The way you restore an LSM volume depends on what the volume is used for, whether the volume is configured and active, and the method you used to back up volume data.

If you used the vdump command to back up the volume (used by UFS file systems or AdvFS domains), use the vrestore command to restore the data.

If you used the rvdump command, use the rvrestore command to restore it.

If the volume is used for an application such as a database, see that application's documentation for the recommended method for restoring backed-up data.

To restore a volume for a UFS file system from a backup that you created with the dump command:

# restore -Yf backup_volume

Note

Both the original volume and the backup volume must be mounted.

You can restore a volume and its contents even if the volume no longer exists, if you have a backup copy of the configuration database. (See Section 5.3.1.)

To restore a volume that no longer exists:

Recreate the volume:
```
# volrestore [-g disk_group] -v volume
```
This recreates the structure of the volume on the same disks it was using at the time you saved the configuration with the volsave command. You cannot assume the data on the disks is valid at this point.

Recreate the file system:
```
# newfs /dev/rvol/disk_group/volume
```

Mount the file system:

# mount /dev/vol/disk_group/volume /mount_point

Change directory to the recreated file system:
```
# cd /mount_point
```

Restore the volume data:
```
# restore -Yrf backup_volume
```

For more information, see restore(8), rrestore(8), and vrestore(8).

5.4.4 Starting LSM Volumes

LSM automatically starts all startable volumes when the system boots. You can manually start an LSM volume that:

You manually stopped

Belongs to a disk group that you manually imported

Stopped because of a disk failure or other problem that you have since resolved

To start an LSM volume:

# volume [-g disk_group] start volume

To start all volumes in a disk group (for example, after importing the disk group):

# volume -g disk_group startall

5.4.5 Stopping LSM Volumes

LSM automatically stops LSM volumes when the system shuts down. When you no longer need an LSM volume, you can stop it then remove it. You cannot stop an LSM volume if a file system is using it.

To stop an LSM volume:

If applicable, stop a file system from using the LSM volume.
- For AdvFS, dissociate the volume from the domain:
```
# rmvol LSM_volume domain
```
  Data on the volume is automatically migrated to other volumes in the domain, if available. For more information, see the AdvFS Administration manual and rmvol(8).
- For UFS, unmount the file system:
```
# umount /dev/rvol/volume
```

Stop the LSM volume:
```
# volume [-g disk_group] stop volume
```
For example, to stop an LSM volume named vol01 in the dg1 disk group:
```
# volume -g dg1 stop vol01
```
To stop all volumes in a disk group:
```
# volume -g disk_group stopall
```

5.4.6 Removing LSM Volumes

Removing an LSM volume destroys the data in that volume. Remove an LSM volume only if you are sure that you do not need the data in the LSM volume or the data is backed up elsewhere. When an LSM volume is removed, the space it occupied is returned to the free space pool.

Note

To remove a volume that was created by encapsulating an AdvFS domain, see Section 5.4.6.1.

To remove a swap volume for a cluster member, see Section 7.6.

The following procedure also unencapsulates UFS file systems.

To remove an LSM volume:

If applicable, stop a file system from using the LSM volume.
- If the volume is part of an AdvFS domain, dissociate the volume from the domain:
```
# rmvol LSM_volume domain
```
  Data on the volume is automatically migrated to other volumes in the domain. For more information, see the AdvFS Administration manual and rmvol(8).
- If the volume is used by a UFS file system, unmount the file system:
```
# umount /dev/rvol/volume
```

Edit the necessary system files as follows:
- If the volume was configured as secondary swap (for a standalone system), remove references to the LSM volume from the vm:swapdevice entry in the sysconfigtab file.
- If the swap space was configured using the /etc/fstab file, update this file to change the swap entries back to disk partitions instead of LSM volumes.
These changes are effective the next time the system restarts.
For more information, see the System Administration manual and swapon(8).

Stop the LSM volume:
```
# volume [-g disk_group] stop volume
```

Remove the LSM volume:
```
# voledit -r rm volume
```
This step removes the plexes and subdisks and the volume itself.

If the volume contained an encapsulated UFS file system, edit the /etc/fstab file to change the volume name to the disk name. For example, change /dev/vol/rootdg/vol-dsk4g to /dev/dsk4g.

5.4.6.1 Unencapsulating AdvFS Domains

You can stop using LSM volumes for an AdvFS domain and revert to using the physical disks or disk partitions directly. This is called unencapsulating a domain.

To unencapsulate the storage in an AdvFS domain, you can use the script that LSM created when you originally encapsulated the domain. The script contains both LSM commands and general operating system commands and performs all the steps necessary to remove the LSM volumes, remove the disks from LSM control, and update the links in the /etc/fdmns directory.

One script for each disk or storage device in the domain is saved in a subdirectory named after the disk access name in the following format:

/etc/vol/reconfig.d/disk.d/dsknp.encapdone/recover.sh

Make sure you run the correct script to unencapsulate the storage in the appropriate domain.

Note

Unencapsulating storage in an AdvFS domain requires that you unmount the filesets.

To unencapsulate an AdvFS domain:

Identify the name of the LSM volume for the domain (which is typically derived from the disk name):

# showfdmn domain

For example:

               Id              Date Created  LogPgs  Version  Domain Name
3a65b2a9.0004cb3f  Wed Jan 17 09:56:41 2001     512        4  dom_1
 
  Vol   512-Blks        Free  % Used  Cmode  Rblks  Wblks  Vol Name
   1L    8380080     8371248      0%     on    256    256  /dev/vol/tempdg/vol-dsk26c

Display and unmount all the filesets in the domain. For example, to unmount the filesets in the dom_1 domain:

# mount
root_domain#root on / type advfs (rw)
/proc on /proc type procfs (rw)
usr_domain#usr on /usr type advfs (rw)
var_domain#var on /var type advfs (rw)
mhs:/work on /work type nfs (v3, rw, udp, hard, intr)
dom_1#junk on /junk type advfs (rw)
dom_1#stuff on /stuff type advfs (rw)
# umount /junk /stuff

Stop the LSM volume:
```
# volume stop volume
```

Find the name of the appropriate unencapsulation script, if one exists:

# ls /etc/vol/reconfig.d/disk.d/
dsk23c.encapdone  dsk24c.encapdone  dsk26c.encapdone  dsk27g.encapdone

Run the appropriate unencapsulation script.
For example, to run the script for the dom_1 domain on disk dsk26c:
```
# sh /etc/vol/reconfig.d/disk.d/dsk26c.encapdone/recover.sh
```

If the script is not available, do the following:
1. Change directory to the domain directory:
```
# cd /etc/fdmns/domain
```
2. Remove the link to the volume:
```
# rm disk_group.volume
```
3. Replace the link to the disk device file:
```
# ln -s /dev/disk/dsknp
```
4. Remove the LSM volume:
```
# voledit [g disk_group] -r rm volume
```
5. Remove the disk media name from the disk group:
```
# voldg -g disk_group rmdisk dm_name
```
6. Remove the disk access name from LSM:
```
# voldisk rm da_name
```

Remount the filesets to the domain:

# mount dom_1#junk /junk
# mount dom_1#stuff /stuff

The domain is available for use. I/O to the domain goes through the disk device path instead of the LSM volume. You can confirm this by running the showfdmn command again:

# showfdmn dom_1

               Id              Date Created  LogPgs  Version  Domain Name
3a65b2a9.0004cb3f  Wed Jan 17 09:56:41 2001     512        4  dom_1
 
  Vol   512-Blks        Free  % Used  Cmode  Rblks  Wblks  Vol Name
   1L    8380080     8371248      0%     on    256    256  /dev/disk/dsk26c

5.4.7 Disabling the Fast Plex Attach Feature on LSM Volumes

If you no longer want to use FPA logging on a volume, you can turn the feature off for that volume. Turning off FPA logging does not remove the FPA log from the primary volume. You can turn the FPA feature back on later if you choose.

If the volume is actively logging to the FPA log plex (if a secondary volume exists), turning off the FPA feature stops logging. If you then return the migrant plex to the primary volume (with the volassist snapback command), the plex undergoes complete resynchronization.

To turn off the FPA feature for a volume:

# volume set fpa=off volume

To remove an FPA log plex, see Section 5.5.6.

5.4.8 Renaming LSM Volumes

You can rename an LSM volume. The new LSM volume name must be unique within the disk group. If the LSM volume supports a file system or is part of an AdvFS domain, you must also update the /etc/fstab file and the /etc/fdmns directory entry.

To rename an LSM volume:

# voledit rename old_volume new_volume

Note

If you do not update the relevant files in the /etc directory before the system is restarted, subsequent commands using a volume's previous name will fail.

5.4.9 Growing LSM Volumes

You can increase the size of a volume (grow a volume) by specifying either an amount to grow by or a size to grow to. For example, you can increase the size of the primary swap space volume. The size of any log plexes remains unchanged.

Note

After you grow a volume that is used by an AdvFS file system, use the mount -o extend command to update the domain to include the additional space in the LSM volume. For more information on increasing the size of a domain, see the AdvFS Administration manual and mount(8).
If the volume is used by a file system other than AdvFS, you must perform additional steps specific to the file system type for the file system to take advantage of increased space. For more information, see the System Administration manual, extendfs(8), and mount(8).

If an application other than a file system uses the volume, make any necessary application modifications after the grow operation is complete.

5.4.9.1 Growing LSM Volumes by a Specific Amount

Use the -f option to grow a primary or secondary volume that is actively logging to an FPA log plex.

Note

Growing a primary or secondary volume with an active FPA log disables the FPA log. When you reattach a migrant plex to a primary volume in this case, regardless of which volume grew, a full resynchronization will occur as though no FPA log ever existed.

You can use the -b option to perform the operation in the background. This is helpful if the growby length specified is substantial and if the volume uses mirror plexes or a RAID5 plex, because it will undergo resynchronization as a result of the grow operation.

To grow a volume by a specific amount:

# volassist [-g disk_group] [-f] [-b] growby volume length_change

For example, to grow a volume by 100K bytes:

# volassist -g dg1 growby myVol 100k

The volume looks similar to the following, before and after the growby operation:

# volprint -g dg1 -vht dataVol

V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  dataVol      fsgen        ENABLED  ACTIVE   400      SELECT    dataVol-01
pl dataVol-01   dataVol      ENABLED  ACTIVE   512      STRIPE    2/128    RW
sd dsk4-01      dataVol-01   dsk4     0        256      0/0       dsk4     ENA
sd dsk5-01      dataVol-01   dsk5     0        256      1/0       dsk5     ENA

# volassist -g dg1 growby dataVol 100k
# volprint -vht -g dg1 dataVol

V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  dataVol      fsgen        ENABLED  ACTIVE   600      SELECT    dataVol-01
pl dataVol-01   dataVol      ENABLED  ACTIVE   768      STRIPE    2/128    RW
sd dsk4-01      dataVol-01   dsk4     0        384      0/0       dsk4     ENA
sd dsk5-01      dataVol-01   dsk5     0        384      1/0       dsk5     ENA

In this case, LSM was able to grow the subdisks using contiguous space on the same disks. If the subdisks in a volume map to the end of the public region on the disks, LSM uses available space in the disk group to create and associate new subdisks to the volume's plex or plexes.

5.4.9.2 Growing LSM Volumes to a Specific Size

Use the -f option to grow a primary or secondary volume that is actively logging to an FPA log plex.

Note

Growing a primary or secondary volume with an active FPA log disables the FPA log. When you reattach a migrant plex to a primary volume in this case, regardless of which volume grew, a full resynchronization will occur as though no FPA log ever existed.

You can use the -b option to perform the operation in the background. This is helpful if the growto length specified is substantial and if the volume uses mirror plexes or a RAID5 plex, because it will undergo resynchronization as a result of the grow operation.

To grow a volume to a specific size:

# volassist [-g disk_group] [-f] [-b] growto volume new_length

For example, to grow a volume from 1MB to 2MB:

# volassist -g dg1 growto vol_A 2m

The volume looks similar to the following, before and after the growto operation:

# volprint -vht -g dg1 vol_A

V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  vol_A        fsgen        ENABLED  ACTIVE   2048     SELECT    vol_A-01
pl vol_A-01     vol_A        ENABLED  ACTIVE   2048     STRIPE    2/128    RW
sd dsk6-01      vol_A-01     dsk6     0        1024     0/0       dsk6     ENA
sd dsk7-01      vol_A-01     dsk7     0        1024     1/0       dsk7     ENA

# volassist -g dg1 growto vol_A 2m
# volprint -vht -g dg1 vol_A

V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  vol_A        fsgen        ENABLED  ACTIVE   4096     SELECT    vol_A-01
pl vol_A-01     vol_A        ENABLED  ACTIVE   4096     STRIPE    2/128    RW
sd dsk6-01      vol_A-01     dsk6     0        2048     0/0       dsk6     ENA
sd dsk7-01      vol_A-01     dsk7     0        2048     1/0       dsk7     ENA

In this case, LSM was able to grow the subdisks using contiguous space on the same disks. If the subdisks in a volume already map to the end of the public region on the disks, LSM uses available space in the disk group to create and associate new subdisks to the volume's plex or plexes. Alternatively, you can specify which disks LSM can use to create the new subdisks.

The following example shows what happens when you specify disks for LSM to use:

# volassist -g dg1 growto dataVol 100k dsk9 dsk10
# volprint -vht -g dg1 dataVol

V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  dataVol      fsgen        ENABLED  ACTIVE   800      SELECT    dataVol-01
pl dataVol-01   dataVol      ENABLED  ACTIVE   1024     STRIPE    2/128    RW
sd dsk4-01      dataVol-01   dsk4     0        384      0/0       dsk4     ENA
sd dsk9-01      dataVol-01   dsk9     0        128      0/384     dsk9     ENA
sd dsk5-01      dataVol-01   dsk5     0        384      1/0       dsk5     ENA
sd dsk10-01     dataVol-01   dsk10    0        128      1/384     dsk10    ENA

Notice that LSM added the subdisk it created on dsk9 to the end of the first column (column 0) and the subdisk it created on dsk10 to the second column (column 1).

5.4.10 Shrinking LSM Volumes

You can decrease the size of a volume by specifying either an amount to shrink by or a size to shrink to. The size of any log plexes remains unchanged.

Cautions

If the volume is used for an AdvFS file system, do not decrease the space in the domain by shrinking an underlying LSM volume. Instead, remove a volume from the domain (in AdvFS, a volume can be a disk, disk partition, or an LSM volume). For more information on removing volumes from a domain, see the AdvFS Administration manual.

If the volume is used for a file system other than AdvFS, you must perform additional steps specific to the file system type before shrinking the volume, so that the file system can recognize and safely adjust to the decreased space.
There is no direct way to shrink a UFS file system other than backing up the data, destroying the original file system, creating a new file system of the smaller size, and restoring the data in the new file system.

For more information, see the System Administration manual.

If an application other than a file system uses the volume, make any necessary application modifications before shrinking the LSM volume.

Note

Shrinking a (primary or secondary) volume with an active FPA log disables the FPA log. When you reattach a migrant plex to a primary volume in this case, regardless of which volume shrank, a full resynchronization will occur as though no FPA log ever existed.

5.4.10.1 Shrinking LSM Volumes by a Specific Amount

To shrink a volume by a specific amount:

# volassist [-g disk_group] -f shrinkby volume length_change

For example, to shrink a volume by 100K bytes:

# volassist -g dg1 -f shrinkby dataVol 100k

5.4.10.2 Shrinking LSM Volumes to a Specific Size

To shrink a volume to a specific size:

# volassist [-g disk_group] -f shrinkto volume new_length

For example, to shrink a volume from 2MB to 1MB:

# volassist -g dg1 -f shrinkto vol_A 1m

5.4.11 Changing LSM Volume Permission, User, and Group Attributes

By default, the device special files for LSM volumes are created with read and write permissions granted only to the owner. Databases or other applications that perform raw I/O might require device special files to have other settings for the permission, user, and group attributes.

Note

Use LSM commands instead of the chmod, chown, or chgrp commands to change the permission, user, and group attributes for LSM volumes. The LSM commands ensure that settings for these attributes are stored in the LSM database, which keeps track of all settings for LSM objects.

To change Tru64 UNIX user, group, and permission attributes:

# voledit [-g disk_group] set \
user=username group=groupname mode=permission volume

The following example changes the user, group, and permission attributes for an LSM volume named vol01 in the rootdg disk group:

# voledit set user=new_user group=admin mode=0600 vol01

5.5 Managing Plexes

The following sections describe how to use LSM commands to manage plexes.

5.5.1 Displaying Plex Information

To display general information for all plexes:

# volprint -pt

Disk group: rootdg
 
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
 
pl ka1-01       ka1          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl ka2-01       ka2          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl ka3-01       ka3          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl ka4-01       ka4          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl rootvol-01   rootvol      ENABLED  ACTIVE   524288   CONCAT    -        RW
pl swapvol-01   swapvol      ENABLED  ACTIVE   520192   CONCAT    -        RW
pl tst-01       tst          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl tst-02       tst          ENABLED  ACTIVE   2097152  CONCAT    -        RW
pl tst-03       tst          ENABLED  ACTIVE   LOGONLY  CONCAT    -        RW
pl vol-dsk25g-01 vol-dsk25g  ENABLED  ACTIVE   2296428  CONCAT    -        RW
pl vol-dsk25h-01 vol-dsk25h  ENABLED  ACTIVE   765476   CONCAT    -        RW

To display detailed information about a specific plex:

# volprint -lp plex

Disk group: rootdg
 
Plex:     tst-01
info:     len=2097152
type:     layout=CONCAT
state:    state=ACTIVE kernel=ENABLED io=read-write
assoc:    vol=tst sd=dsk0-01
flags:    complete

5.5.2 Adding a Data Plex (Mirroring LSM Volumes)

You can add a data plex to a volume to mirror the data in the volume. You cannot use a disk that is already used by the volume to create the mirror. A volume can have up to 32 plexes, which can be any combination of data and log plexes.

Adding a plex is one way to move the LSM volume to disks with better performance, if you specify the disks for LSM to use. You can also add a plex temporarily, so you can repair or replace disks in the original plex.

When you add a plex to a volume, the volume data is copied to the new plex. This process can take several minutes to several hours depending on the size of the volume.

Caution

To mirror the volumes for the boot disk and primary swap space (rootvol and swapvol) on a standalone system, you must use the volrootmir command (Section 3.4.1). The volrootmir command performs special operations that ensure the system can boot from either mirror in the root file system volumes.

To add a data plex:

# volassist mirror volume [disk]

Note

Adding a data plex does not add a DRL plex to the volume. Volumes with mirror plexes should have a DRL plex, except volumes for swap space. To add a DRL plex to a volume, see Section 5.5.3.

5.5.3 Adding a Log Plex

You can add a log plex (DRL or RAID5 log) to a volume that has mirrored data plexes or a RAID5 data plex. LSM automatically creates the log of a size appropriate to the size of the volume (65 blocks per gigabyte of volume size, by default).

Note

A DRL plex is not supported on rootvol (on a standalone system), cluster_rootvol (in a cluster), or any swap volumes (standalone systems or clusters).

To improve performance and avoid the risk of losing both data and the log if the disk fails, create the log plex on a disk that is not in use by one of the volume's data plexes and, if possible, on a disk that does not already support the log for other volumes.

To add a log plex to a volume using any available space in the disk group:

# volassist [-g disk_group] addlog volume [disk]

To add a log to a volume by specifying a disk:

Make sure the disk is not in use by the same volume:

# volprint -vht volume

For example:

# volprint -vht genvol

Disk group: rootdg
 
TY NAME         ASSOC        KSTATE   LENGTH   PLOFFS   STATE    TUTIL0  PUTIL0
v  genvol       gen          ENABLED  204800   -        ACTIVE   -       -
pl genvol-01    genvol       ENABLED  204800   -        ACTIVE   -       -
sd dsk19-02     genvol-01    ENABLED  204800   0        -        -       -
pl genvol-02    genvol       ENABLED  204800   -        ACTIVE   -       -
sd dsk5-01      genvol-02    ENABLED  204800   0        -        -       -

In the previous output, the volume genvol uses space for its data plexes on disks dsk19 and dsk5.

Optionally but recommended, make sure the disk does not already support log plexes for other volumes:
```
# volprint -ht | grep disk
```
For example:
```
# volprint -ht | grep dsk5
```
```
sd dsk5-03      fsvol-03     dsk5     322912   245760   0         dsk5     ENA
sd dsk5-01      genvol-01    dsk5     118112   204800   0         dsk5     ENA
sd dsk5-02      vol_r5-01    dsk5     32768    85344    4/0       dsk5     ENA
 
```
In the previous output, space on dsk5 is used by volumes fsvol, genvol, and vol_r5, but not for their logs. (However, dsk5 is already heavily used and would not be a good candidate for the new log plex.)

Add the log to the volume, specifying an eligible disk:
```
# volassist [-g disk_group] addlog volume disk
```
For example, to add a log plex to volume genvol on dsk11, which is not used in any other volume in the rootdg disk group:
```
# volassist addlog genvol dsk11
```

5.5.4 Adding an FPA Plex

You can add one or more Fast Plex Attach (FPA) logs to a mirrored volume to provide redundancy for the FPA logs, just as you can add multiple dirty-region logs (DRLs) to a volume.

FPA logging is supported on any mirrored volume on a standalone system or a cluster, including the rootvol and cluster_rootvol volumes, but excluding volumes used for swap space.

Note

You cannot add an FPA log to a volume while a migrant plex is detached from the volume (attached to a secondary volume).

If the volume has a DRL log, the FPA log length will be the same as the DRL log length. If the volume has no DRL log, when you add the first FPA log to a mirrored volume, you can specify the length of the log with the loglen=length attribute, the number of FPA logs with the nfpalog=count attribute, and which disks it can or cannot use. To exclude storage, use the ! prefix (or \! in the C shell).

To add an FPA log plex to a volume, optionally specifying the number of FPA log plexes or the disk or disks to use:

# volassist addfpa volume [nfpalog=count] [disk...]

For example, to add two FPA log plexes on disks dsk5 and dsk6 to a volume named dvol_01:

# volassist addfpa dvol_1 [nfpalog=2] dsk5 dsk6

The volume looks like the following:

Disk group: rootdg
 
TY NAME         ASSOC        KSTATE   LENGTH   PLOFFS   STATE    TUTIL0  PUTIL0
v  dvol_1       fsgen        ENABLED  245760   -        ACTIVE   -       -
pl dvol_1-01    dvol_1       ENABLED  245760   -        ACTIVE   -       -
sd dsk1-01      dvol_1-01    ENABLED  245760   0        -        -       -
pl dvol_1-02    dvol_1       ENABLED  245760   -        ACTIVE   -       -
sd dsk2-01      dvol_1-02    ENABLED  245760   0        -        -       -
pl dvol_1-03    dvol_1       ENABLED  LOGONLY  -        ACTIVE   -       -
sd dsk1-02      dvol_1-03    ENABLED  65       LOG      -        -       -
pl dvol_1-04    dvol_1       ENABLED  FPAONLY  -        ACTIVE   -       -
sd dsk5-01      dvol_1-04    ENABLED  65       FPA      -        -       -
pl dvol_1-05    dvol_1       ENABLED  FPAONLY  -        ACTIVE   -       -
sd dsk6-01      dvol_1-05    ENABLED  65       FPA      -        -       -

When you create a backup volume using the volassist snapfast command (Section 5.4.2.2 ), LSM tracks the changes to the primary volume in all the FPA logs associated with the volume.

5.5.5 Detaching a Plex

Detaching a plex preserves the plex's association to a volume. You can detach a plex to perform temporary operations with it, with the intent of reattaching the plex to the same volume later.

Caution

If you detach all but one data plex from a mirrored volume, the volume's data is no longer redundant.

Note the following:

If you detach the last log plex from a mirrored volume and the system fails, LSM resynchronizes the contents of all the plexes when the system restarts. The volume remains usable during this operation, but its performance might be significantly degraded.

If you detach the last RAID 5 log plex from a RAID5 volume and the system fails, LSM recalculates the parity for the entire volume; this involves reading back all the volume data, regenerating the parity for each stripe, and rewriting each stripe in the plex. The volume remains usable during this operation, but its performance might be significantly degraded.

If you detach the last FPA log plex from a primary volume, FPA logging is disabled. When you return the migrant plex to the primary volume, a full resynchronization occurs, as if FPA were never enabled. To disable FPA logging on a volume, see Section 5.4.7.

To detach a plex from a volume:

# volplex [-f] det plex

Use the force (-f) option to detach a volume's last complete, enabled data plex or the FPA log plex from a primary volume.

5.5.6 Dissociating a Plex

Dissociating a plex removes its association to a volume. You can dissociate a plex from a volume to unmirror a volume (remove all but one data plex) or reduce the number of mirrors or logs in a volume, usually with the intent of completely removing the plex and reusing its disk space for another purpose. To support this, you can dissociate and recursively remove a plex and its components in one command. Alternatively, you can dissociate a plex and use it to create another volume; for example, to back up a volume.

Note

The recommended method for backing up a mirrored volume is with the volassist snapfast command (to use FPA logging) or the volassist snapshot command. For more information, see Section 5.4.2.

Note the following:

If you dissociate the last log plex from a mirrored volume and the system fails, LSM resynchronizes the contents of all the plexes when the system restarts. The volume remains usable during this operation, but its performance might be significantly degraded.

If you dissociate the last RAID 5 log plex from a RAID5 volume and the system fails, LSM recalculates the parity for the entire volume; this involves reading back all the volume data, regenerating the parity for each stripe, and rewriting each stripe in the plex. The volume remains usable during this operation, but its performance might be significantly degraded.

If you dissociate the last FPA log plex from a primary volume, FPA logging is disabled. When you return the migrant plex to the primary volume, a full resynchronization occurs, as if FPA were never enabled. To disable FPA logging on a volume, see Section 5.4.7.

Caution

If you dissociate all but one data plex from a mirrored volume, the volume's data is no longer redundant.

To remove the last data plex from a volume (remove a volume completely), see Section 5.4.6.

To dissociate a plex from a volume:

# volplex [-f] dis plex

Use the force (-f) option to dissociate a volume's last complete, enabled data plex or the FPA log plex from a primary volume.

To dissociate and recursively remove a plex from a volume:

# volplex [-f] -o rm dis plex

Recursively dissociating a plex removes both the plex and its subdisks. The disks remain under LSM control.

5.5.7 Reattaching Plexes

If you detached or dissociated (but did not recursively remove) a data plex or log plex from a volume, you can reattach it to the volume. You can attach a detached plex only to its original volume.

To reattach a detached data or log plex to a volume:

# volrecover volume

To reattach a dissociated data or log plex to a volume:

# volplex att volume plex

5.5.8 Changing the Plex Layout of LSM Volumes

For volumes that use one or more concatenated or striped plexes, you can change the plex layout from concatenated to striped or vice versa. For example, you can change from a concatenated plex to a striped plex to improve performance.

The steps involved include adding a plex to the volume and then dissociating the original plex or plexes from the volume.

5.5.8.1 Changing the Plex Layout from Concatenated to Striped

To change the plex layout of a volume from concatenated to striped:

Display the size of the volume whose plex layout you want to change:

# volprint [-g diskgroup] -ht volume

Disk group: rootdg
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  volC         fsgen        ENABLED  ACTIVE   204800   SELECT    -
pl volC-01      volC         ENABLED  ACTIVE   204800   CONCAT    -        RW
sd dsk2-01      volC-01      dsk2     0        204800   0         dsk2     ENA

In this example, the volume volC has one concatenated data plex of 204800 sectors (100 MB).

Verify there is enough space in the same disk group to mirror the volume:
```
# voldg [-g disk_group] free
```

Add a new plex with the characteristics you want to the volume.

For example, to convert the volume volC to a striped volume, add a striped plex to the volume, specifying the number of columns and, optionally, the disks you want LSM to use:

# volassist -g dg1 mirror volC layout=stripe \
ncolumn=2 [disk...]

The volume now looks similar to the following, with the original concatenated plex and the new striped plex:

# volprint -g dg1 -ht volC

Disk group: dg1
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  volC         fsgen        ENABLED  ACTIVE   204800   SELECT    volC-02
pl volC-01      volC         ENABLED  ACTIVE   204800   CONCAT    -        RW
sd dsk2-01      volC-01      dsk2     0        204800   0         dsk2     ENA
pl volC-02      volC         ENABLED  ACTIVE   204800   STRIPE    2/128    RW
sd dsk3-01      volC-02      dsk3     0        102400   0/0       dsk3     ENA
sd dsk5-01      volC-02      dsk5     0        102400   1/0       dsk5     ENA

Remove and dissociate the original plex from the volume:
```
# volplex [-g diskgroup] -o rm dis old_plex
```
For example, to remove the original concatenated plex volC-01 from volC:
```
# volplex -g dg1 -o rm dis volC-01
```

If the volume has only one plex and you add another plex with the volassist mirror command, the new plex will have the same layout as the current plex (subject to the free disk space in the disk group and other restrictions).

If the volume still has plexes with the original layout type, repeat the process of adding new plexes with the layout you want and removing the original plexes until all plexes use the new layout.

You can also add a Dirty Region Log (Section 5.5.3) to the volume if it does not have one.

5.5.8.2 Changing the Plex Layout from Striped to Concatenated

To change the plex layout of a volume from striped to concatenated:

Display the size of the volume whose plex layout you want to change:

# volprint [-g disk_group] -ht volume

Disk group: dg1
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  volS         fsgen        ENABLED  ACTIVE   204800   SELECT    volS-01
pl volS-01      volS         ENABLED  ACTIVE   204800   STRIPE    4/128    RW
sd dsk10-01     volS-01      dsk10    0        51200    0/0       dsk10    ENA
sd dsk11-01     volS-01      dsk11    0        51200    1/0       dsk11    ENA
sd dsk12-01     volS-01      dsk12    0        51200    2/0       dsk12    ENA
sd dsk14-01     volS-01      dsk14    0        51200    3/0       dsk14    ENA

In this example, the volume volS has one striped data plex of 204800 sectors (100 MB).

Verify there is enough space in the same disk group to mirror the volume:
```
# voldg [-g disk_group] free
```

Add a new plex with the characteristics you want to the volume.

For example, to convert the volume volS to a concatenated volume, add a concatenated mirror to the volume, optionally specifying the disks you want LSM to use:

# volassist -g dg1 mirror volS layout=nostripe [disk...]

The volume now looks similar to the following, with the original striped plex and the new concatenated plex:

# volprint -g dg1 -ht volS

Disk group: dg1
 
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL   PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
 
v  volS         fsgen        ENABLED  ACTIVE   204800   SELECT    volS-01
pl volS-01      volS         ENABLED  ACTIVE   204800   STRIPE    4/128    RW
sd dsk10-01     volS-01      dsk10    0        51200    0/0       dsk10    ENA
sd dsk11-01     volS-01      dsk11    0        51200    1/0       dsk11    ENA
sd dsk12-01     volS-01      dsk12    0        51200    2/0       dsk12    ENA
sd dsk14-01     volS-01      dsk14    0        51200    3/0       dsk14    ENA
pl volS-02      volS         ENABLED  ACTIVE   204800   CONCAT    -        RW
sd dsk19-01     volS-02      dsk19    0        204800   0         dsk19    ENA

Remove and dissociate the original plex from the volume:
```
# volplex [-g disk_group] -o rm dis old_plex
```
For example, to remove the original striped plex volS-01 from volS:
```
# volplex -g dg1 -o rm dis volS-01
```

If the volume has only one plex and you add another plex with the volassist mirror command, the new plex will have the same layout as the current plex (subject to the free disk space in the disk group and other restrictions).

If the volume still has plexes with the original layout type, repeat the process of adding new plexes with the layout you want and removing the original plexes until all plexes use the new layout.

You can also add a Dirty Region Log (Section 5.5.3) to the volume if it does not have one.

5.6 Managing Subdisks

The following sections describe how to use LSM commands to manage subdisks.

5.6.1 Displaying Subdisk Information

To display general information for all subdisks:

# volprint -st

Disk group: rootdg
 
SD NAME      PLEX        DISK   DISKOFFS    LENGTH  [COL/]OFF DEVICE  MODE
 
sd dsk2-01   vol_mir-01  dsk2   0           256        0      dsk2    ENA
sd dsk3-02   vol_mir-03  dsk3   0            65      LOG      dsk3    ENA
sd dsk3-01   vol_mir-02  dsk3   65          256        0      dsk3    ENA
sd dsk4-01   p1          dsk4   17          500        0      dsk4    ENA
sd dsk4-02   p2          dsk4   518        1000        0      dsk4    ENA

To display detailed information about a specific subdisk:

# volprint -l subdisk

For example:

# volprint -l dsk12-01

Disk group: rootdg
 
Subdisk:  dsk12-01
info:     disk=dsk12 offset=0 len=2560
assoc:    vol=vol5 plex=vol5-02 (offset=0)
flags:    enabled
device:   device=dsk12 path=/dev/disk/dsk12g diskdev=82/838

5.6.2 Joining Subdisks

You can join two or more subdisks to form a single, larger subdisk. Subdisks can be joined only if they belong to the same plex and occupy adjacent regions of the same disk. For a volume with striped plexes, the subdisks must be in the same column. The joined subdisk can have a new subdisk name or retain the name of one of the subdisks being joined.

To join subdisks:

# volsd join subdisk1 subdisk2 new_subdisk

5.6.3 Splitting Subdisks

You can divide a subdisk into two smaller subdisks. After the subdisk is split, you can move the data in the smaller subdisks to different disks. This is useful for reorganizing volumes or for improving performance. The new, smaller subdisks occupy adjacent regions within the same region of the disk that the original subdisk occupied.

You must specify a size for the first subdisk; the second subdisk consists of the rest of the space in the original subdisk.

If the subdisk to be split is associated with a plex, both of the resultant subdisks are associated with the same plex. You cannot split a log subdisk.

To split a subdisk and assign each subdisk a new name:

# volsd -s size split original_subdisk \
new_subdisk1 new_subdisk2

To split a subdisk and retain the original name for the first subdisk and assign a new name to the second subdisk:
```
# volsd -s size split original_subdisk new_subdisk
```

5.6.4 Moving Subdisks to a Different Disk

You can move the data in subdisks to a different disk to improve performance. The disk space occupied by the data in the original subdisk is returned to the free space pool.

Ensure that the following conditions are met before you move data in a subdisk:

Both source and destination subdisks must be the same size.

The source subdisk must be part of an active plex on an active volume.

The destination subdisk must not be associated with any other plex.

To move data from one subdisk to another:

Display the size of the subdisk you want to move:

# volprint subdisk

For example:

# volprint -l dsk20-01

Disk group: dg1
 
Subdisk:  dsk20-01
info:     disk=dsk20 offset=0 len=204800
assoc:    vol=volS plex=volS-01 (offset=0)
flags:    enabled
device:   device=dsk20 path=/dev/disk/dsk20g diskdev=81/1350

Create a new subdisk of the same size on a different disk:

# volmake [-g diskgroup] sd subdisk_name disk len=length

For example:

# volmake -g dg1 sd dsk11-01 dsk11 len=204800

Move the data in the old subdisk to the new subdisk:
```
# volsd mv source_subdisk target_subdisk
```
For example:
```
# volsd -g dg1 mv dsk20-01 dsk11-01
```
This leaves the old subdisk unassociated with any volume. If it is not needed, you can remove the subdisk (Section 5.6.5). Removing the subdisk returns that space to the free pool.

5.6.5 Removing Subdisks

You can remove a subdisk that is not associated with or needed by an LSM volume. Removing a subdisk returns the disk space to the free space pool in the disk group. To remove a subdisk, you must dissociate the subdisk from a plex or volume, then remove it.

To remove a subdisk:

Display information about the subdisk to identify any volume or plex associations:

# volprint -l subdisk

If the subdisk is associated with a volume, the following information is displayed:

Disk group: rootdg
 
Subdisk:  dsk9-01
info:     disk=dsk9 offset=0 len=2048
assoc:    vol=newVol plex=myplex (column=1 offset=0)
flags:    enabled
device:   device=dsk9 path=/dev/disk/dsk9g diskdev=82/646

If the subdisk has no associations to any plex or volume, the following information is displayed:

Disk group: dg1
 
Subdisk:  dsk20-01
info:     disk=dsk20 offset=0 len=204800
assoc:    vol=(dissoc) plex=(dissoc)
flags:    enabled
device:   device=dsk20 path=/dev/disk/dsk20g diskdev=81/1350

Do one of the following to remove the subdisk:
- If the subdisk is associated with a volume:
```
# volsd [-g disk_group] -o rm dis subdisk
```
- If the subdisk is not part of a volume and has no associations:
```
# voledit [-g disk_group] rm subdisk
```