Disk failure in VG

[David Owens_1]

I have a volume group with three disks. One disk has failed and my hardware support people (not HP unfortunately) won't have a replacement for 24 hours so I am trying a quick fix.

I really only need two of the disks for my storage requirements so I went to SAM and tried to remove the volume group and recreate it with just the two disks and restore from backup. When trying to remove the VG, it gives me an error indicating that it can't find the bad disk. Actually I removed the disk but the system would ne recognize it anyway.

Since SAM would not complete this task, I tried the following with no success.
# vgreduce /dev/vg02 /dev/dsk/c2t1d0
# vgreduce -f /dev/vg02
# vgreduce -l /dev/vg02

Can anyone help me remove this volume group?


Thanks in advance,
David Owens

[Pete Randall]

David,

Try vgexport. I suspect it will fail - complaining about the missing disk as well, but it's worth a shot.


Pete

[Donald Kok]

Hi David,

I would remove the good disks from the vg, and create a new vg with it. You can bother about the bad vg later.

Goodluck
Donald

[A. Clay Stephenson · Accepted solution]

Do a vgchange -a n /dev/vg02 followed by a vgexport /dev/vg02.


Rule 1: ALWAYS mirror then you won't care if a disk fails.

Rule 2: Keep spares on hand for things that fail rather often (e.g. drives) so that you don't have to depend upon anyone for fixes. I typically replace a disk from spares and then call Mr. Goodwrench to ship a replacement spare.

[Patrick Wallek]

I'm with Pete. Try vgexport.

# vgexport /dev/vg02

[S.K. Chan]

Well...
# vgexport /dev/vg02
should do it for you. It removes all your vg devices files and the entry in /etc/lvmtab. But before you do that take note of all you LVs in vg02 since you will need to recreate them later. The entries in /etc/fstab file would still be there.

[Michael Steele_2]

'vgreduce -f /dev/vg02' is a good start. It often works. Need to activate without quorum though. Try this first:

vgchange -a y -q n /dev/vg02
vgreduce -f /dev/vg02

(* -q n = activate w/out quorum *)

Next procedure to try is the following:

a) lvdisplay -v -k /dev/vg02/lvol#

Note the PV1 id number = '1':

LE PV1 PE1 Status1
0000 1 0198 stale
0001 1 0199 stale

NOTE: Since you have 3 disks you should see 3 PV ids. Cross reference by disk, c0t2d0. Then, when you know the PV id number proceed:

b) lvreduce -m 0 -k /dev/vg02/lvol# 1

Note the PV1 id number is at the end of the lvreduce command.

c) repeat a) and b) off of bad disk.

d) vgreduce -f /dev/vg02