Volume Group Problem

[tech1214]

I have a volume group that has 197 GB of space available (not assigned to a logical volume or file system) but I cannot use the whole of it due to the amount of physical extends the VG currently has. I have onlinejfs installed. I am also aware of vgreduce. It seems vgreduce will allow you to remove physical volumes from the VG. How can I tell which physical volumes I am using and which ones are free?

Is there a way to shrink a volume group without potentially destroying the volume group itself?

[Stefan Farrelly]

You need to use pvdisplay to see which physical disks have no extents in use (ie. all free). Those that are free you can vgreduce out of your VG. Dont worry - if you try to vgreduce a PV which is used it wont let you.

The same output you can get from vgdisplay -v, it lists all physical volumes at the bottom and which have extents used (so you cant vgreduce them) and which are 100% free which means you can vgreduce them on the fly.

You can shrink a volume group by removing physical volumes which arent used at any time without any problems. No danger there.

You only use OnlineJfS to expand or reduce mounted filesystems.

You can also use pvmove to move extents on a physical volume from one to another, so that you can free up a physical volume then vgreduce it. This is also danger free.

[John Poff]

Hi,

You can use vgreduce to remove some physical volumes from your volume group and use them in another VG, but first you'll need to make sure that they are empty. Do a 'vgdisplay -v VGNAME' where VGNAME is the name of your volume group. In the first section, look for the number of PVs in your VG. The sections at the end of the report will show the details for each PV. If you have any that have no PEs used, [Total PEs is equal to Free PEs], you can vgreduce those PVs. If all your PVs have some PEs used, you'll either have to look at using pvmove to move the extents to another PV, or else you'll have to see if you can remove a logical volume to free up a PV.

Post the output of your vgdisplay -v command and we'll take a look at it.

JP

[Sunil Sharma_1]

Hi,
You have to reduce volume group using command vgreduce before doing that just insure that physical volume(disk) is not used by any other volumes using pvdisplay command. if disk is not used by any volume you can remove that disk from volume group.
#pvdisplay /dev/dsk/cxtydz
#vgreduce vgnn /dev/dsk/cxtydz

using this command you can remove disk cxtydz from volume group and use this disk for creating new volume group with more number of Physical extent(PE)

SS

[Bill McNAMARA_1]

shrink a volume group without potentially destroying the volume group itself?


You can use the pvmove command to move an lvol off a PV onto another PV in that VG. This will allow you to free up a PV for removal from the VG via vgreduce.

See the man for pvmove.

Later,
Bill

[Michael Steele_2]

I'll be answering your question in three parts, 1) why you can't use the whole of your vg, 2) identifying the disk to reduce and 3) reducing the disk.

1) Regarding "... cannot use the whole of it due to the amount of physical extends the VG currently has..."

Only 'vgcreate' can change the values of PE_MAX and MAX_PE and make a larger vg. To do those you'll have to make a new vg. You won???t be able to modify the current vg. This includes new lvols, temporary mount points and copying over the data from the old vg into the new. Once copied over you modify /etc/fstab with the new vg information.

Please respond that you have this problem and I'll give you more information on the procedure.

2) Identifying the disk: You???ll need to unmount the file systems residing on the disk first. This is done with vgdisplay, pvdisplay, lvdisplay and by reviewing /etc/fstab or the bdf report. Check all the lv's in the vg and map the disks. Here's an example of lvdisplay. Note the end of the report indicates the disks in use:

--- Logical volumes ---
LV Name /dev/vg00/lvol4
VG Name /dev/vg00
LV Permission read/write
LV Status available/syncd
Mirror copies 1
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 1212
Current LE 303
Allocated PE 606
Stripes 0
Stripe Size (Kbytes) 0
Bad block on
Allocation strict
IO Timeout (Seconds) default

--- Distribution of logical volume ---
PV Name LE on PV PE on PV
/dev/dsk/c3t6d0 303 303
/dev/dsk/c3t5d0 108 108
/dev/dsk/c5t5d0 70 70
/dev/dsk/c5t6d0 125 125

--- Logical extents ---
LE PV1 PE1 Status 1 PV2 PE2 Status 2
00000 /dev/dsk/c3t6d0 00172 current /dev/dsk/c5t6d0 00172 current
00001 /dev/dsk/c3t6d0 00173 current /dev/dsk/c5t6d0 00173 current
00002 /dev/dsk/c3t6d0 00174 current /dev/dsk/c5t6d0 00174 current
00003 /dev/dsk/c3t6d0 00175 current /dev/dsk/c5t6d0 00175 current
00004 /dev/dsk/c3t6d0 00176 current /dev/dsk/c5t6d0 00176 current
00005 /dev/dsk/c3t6d0 00177 current /dev/dsk/c5t6d0 00177 current
00006 /dev/dsk/c3t6d0 00178 current /dev/dsk/c5t6d0 00178 current
00007 /dev/dsk/c3t6d0 00179 current /dev/dsk/c5t6d0 00179 current

3) Reducing the disk out of the vg. Now that you've mapped out the lvols you can first unmount the file system or systems before vgreducing. See above responses:

bdf | grep -i vg##
umount /mount_point
vgreduce (* see above responses *)