Logical Volume Manager (LVM) Commands for AIX

Command Summary

First a quick review on some of the terminology that AIX LVM uses

Short	Definition	Description
PV	PHYSICAL VOLUME

Command	Definition
Physical Volumes related
lspv [-l] [-M|p-] [-L] PV	Lists the disks on the server, including the physical volume will give details about that disk. -l option will list the details of how the filesystems are distributed on the disk. -M | -p option lists logical volumes details on physical disk. -L displays the physical volume characteristics.
chdev -l PV -a pv=yes	assign a PVID to a new hdisk (PV)
lsvpcfg	Lists each vpath and the hdisks that make up the vpath
Logical Volume related
lslv [-l|-m] [-L] LV	Lists information about the logical volumes. -l | -m option lists disks in the logical volume. -L displays the logical volume characteristics.
chlv -n newname oldname	Change the name of a logical volume (it must be inactive)
mklv -y LV_name VG	Makes a logical volume in a volume group
rmlv LV	Removes a logical volume (it must be inactive)
Volume Group related
lsvg [-l] [-L] [-p] VG	Lists the volume groups on the server, including the volume group name will give details about that vg. -l option list the logical volumes in the volume group. -L displays the logical volume characteristics. -p option lists disks in the volume group
mkvg -y VG PV . . . PV	Makes a volume group out of one or more physical volumes
varyonvg VG	take the VG online
varyoffvg VG	take the VG offline
exportvg VG	removes a volume group from a machine
extendvg VG PV ... PV	Adds a new physical volume to an existing volume group
importvg -y VG PV	add a volume group to another machine
savevg -l -f device VG	makes a backup copy of another volume group
reducevg VG PV ... PV	Removes a physical volume from a volume group
Filesystem related
df -k	Shows the disk usage of logical volumes on the server.
df -Pm	Shows the disk usage of logical volumes in portability modus and size in MB
crfs -v jfs -m filesystem -g VG -a logname=LV_Name -a size="# of 512 byte blocks"	This command makes a logical volume LVName, with size=#, mount point with a journaled file system
crfs -v jfs2 -A yes -p rw -d LV -m MP -a logname=INLINE	creates a read write jfs2 file system on a logical volume LV and for Mount Point MP and logging on the same LV
crfs -v jfs2 -m filesystem -d LV	creates a jfs2 file system on a logical volume
chfs -a size="#512 byte blocks" filesystem	Increases the size of a journaled file system to the total number of 512 byte blocks specified
chfs -a size="+512 byte blocks" filesystem	Increases the size of a journaled file system by the addional number of 512 byte blocks specified. Beispiel:chfs -a size=+393216 /usr
rmfs filesystem	removes a file system and it's logical volume
mount LV filesystem	mount logical volume on files system mount point
mount filesystem	mount filesystem if it is already in /etc/filesystems
umount filesystem	unmount the file system
Filesystem backup
mksysb -l -f device	makes a bootable backup of rootvg

Glossary

Term	Definition
Journaled File System (JFS)	File system that uses a journaled log for faster, more reliable data recovery
Logical Partition (LP)	The LV is made up of LPs. The LP corresponds to 1 or more (in the case of mirroring) PPs.
Logical Volume (LV)	The VG is subdivided into logical volumes and each LV can have a file system on it.
Physical Partition (PP)	All physical volumes are subdivided into pps. PPs are all the same size.
Physical Volume (PV)	Disk that is being managed by LVM.
rootvg	Default volume group created during installation. The vg holds the OS filesystems ( /,/usr, /home, /proc /opt, /tmp, /var and swap space )
Volume Group (VG)	Area of storage that consists of one or more PVs

Sample LVM Procedures:

move Logical Volume to new assigned disks and remove old disk

Some people always like to use mirroring. Others always use migratepv, because it seems a lot easier, as the data can be moved in a single command. But there are advantages for using the mirroring option with mklvcopy.
mklvcopy lets you create the copy without having to synchronize the data straight away. That means you can synchronize the copies later on using syncvg, perhaps when I/O activity isn’t so intense. The syncvg command allows you to synchronize Physical Partitions (PPs) in parallel using the -P flag or by setting the NUM_PARALLEL_LPS environment variable. Neither migratepv nor mklvcopy allow this.
You can also monitor progress of your synchronization by checking the stale partition count with lsvg or lspv. In contrast, if you try to check how migratepv is going, you'll get a message saying there's a lock on the volume group.
Another benefit of using mklvcopy is you can keep two current copies of your data before you commit to using the target PV. Although this doubles the writes while both copies are in use, you can easily remove the copy from the source PV later or from the target if you want to revert to the source PV quickly.
If you did't synchronize the LV, rmlvcopy gives an error, when you try to remove the last good copy while there are still some stale partitions.

direct migration from old to new disk (migratepv)

The migratepv command is the easiest way of moving data. You can migrate a logical volume to one or more PVs with one command. As a minimum, you need a source PV and a target PV, and both of them must be in the same volume group.

• add new disk to VG
  

  extendvg [VG] [hdisk##]

• migrate whole disk
  

  migratepv -l lv02 hdisk_old hdisk_new1 ... hdisk_new#

• or migrate one logical volume after another
  

  migratepv -l lv02 hdisk_old hdisk_new1 ... hdisk_new#

• The above command will move all of the physical partitions. The command may take a while to complete. That's because it creates a mirror of the data, synchronizes the copy and then removes the original. If the command is successful, the old disk will have 0 physical partitions in use, but still remain in the volume group.You can remove it from the volume group using reducevg.
  

  reducevg  [Volume Group] [physical Volume]

sometimes you need to move only single PE. You can do it with the migratelp command. The below command will copy partition 63 from second mirror from logical volume httplv to hdisk27 to partition 100.

migratelp httplv/63/2 hdisk27/100

use additonal mirror for the migration (mklvcopy)

If, instead of moving data in one hit, create a mirror of each LV using the mklvcopy command. You can synchronize the two copies when you run the mklvcopy command with option -k , or do it afterward. It makes sense to create the mirror copy without synchronizing immediately. This allows you to arrange the synchronization of the copy to run at a quiet time.

• add new disk to VG
  

  extendvg [VG] [hdisk##]

• mirror Logical Volume. With mklvcopy, you need to indicate how many copies of the LV you want to end up with. You can have two or three copies.
  

  mklvcopy -s y [Logical Volume] [Number of mirrors 2|3] [physical Volume]

• synchronize Volume Group
  

  syncvg -v [Volume Group]

• or synchronize Physical Volumes at a quiet time when I/O activity isn’t so intense.
  

  syncvg -p [Physical Volume(s)]

• verify successful mirroring (pps will appear "stale" until synchronization is complete).
  

  lsvg -l [Volume Group]

• remove mirror on old disk
  

  rmlvcopy [Logical Volume] 1 [OLD physical Volume]

• the old disk can be removed from the Volume Group, if all Logical Volumes are moved to the new disk
  

  reducevg  [Volume Group] [physical Volume]

Mirror rootvg

Make sure you have an empty disk, in this example its hdisk1
Add the disk to the rootvg

extendvg rootvg hdisk1

Mirror the rootvg via

mirrorvg rootvg

Adapt the bootlist to add the current disk, the system will then fail to hdisk1 is hdisk0 fails during startup

bootlist -o -m normal

this will list currently 1 disk, in this exmaple hdisk0, run command below to add the second disk

bootlist -m normal hdisk0 hdisk1

Run a bosboot on both new disks, this will install all software needed for boot on the disk

bosboot -ad hdisk0
bosboot -ad hdisk1

---