SD32000 logical volume stale

[stephen peng]

dear all,
the vg00 on SD32000 is consisted of 4 pv, using PVG to produced a mirror-disk enviroment, when i run vgdisplay -v, the following messages appear:
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t0d0":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
and /dev/dsk/c0t0d0 is one of the four PVs, it seems that there is hardware problem of c0t0d0, though it is CLAIMED in ioscan output; when i run diskinfo /dev/rdsk/c0t0d0, it shows the following messages after a long period of time:
SCSI describe of /dev/rdsk/c0t0d0:
vendor: HP 36.4G
product id: ST336753LC
type: direct access
size: 0 Kbytes
bytes per sector: 0
so i consider that i should replace c0t0d0, do you think it is a correct way?
now i want to retrieve the lvm enviroment, what should i do? is that following way right?
1, replace the hard disk
2, vgcfgrestore
3, vgsync
is there any other step i should pay attention to?

thanks a lot for your reply!

[Tim D Fulford · Accepted solution]

Hi

You disk is on vg00. This means that whilst the above is what you might be told to do, it wont work (or at least I've had problems with replacing vg00). - it is a root disk, so you will need to crate the new disk as a bootable device.

My advice is
1 - lvreduce all LVs on /dev/dsk/c0t0d0
2 - vgreduce vg00 /dev/dsk/c0t0d0
3 - replace new disk
4 - pvcreate -B /dev/rdsk/c0t0d0
5 - vgextend -g vg00 /dev/dsk/c0t0d0
6 - lvchange -m 1 /dev/vg00/lvol1
7 - lvchange -m 1 /dev/vg00/lvol2
8 - lvchange -m 2 /dev/vg00/lvol3
9 - .. as above for rest of LVs ..

It is important lvol1, 2 & 3 are done in order else you wont be able to boot from this device.

You will also want to update the LIF area too... but that is another question..

Regards

Tim

[Tim D Fulford]

oops ... i put lvchange above.. I think it is lvextend .. I think you might get a few replies pointing this error out...

Regards

Tim

[stephen peng]

hi Tim,
because of the strange state of the c0t0d0( CLAIMED in ioscan output, and the following is the output of lvdisplay:
--- Logical extents ---
LE PV1 PE1 Status 1 PV2 PE2 Status 2
00000 ??? 00000 stale /dev/dsk/c5t8d0 00000 current
00001 ??? 00001 stale /dev/dsk/c5t8d0 00001 current
00002 ??? 00002 stale /dev/dsk/c5t8d0 00002 current
00003 ??? 00003 stale /dev/dsk/c5t8d0 00003 current
00004 ??? 00004 stale /dev/dsk/c5t8d0 00004 current)
i assume that the command will hang if i run lvreduce or vgreduce, so could i just omit these two steps in your advice?
i have never deal with some PVG circumstance, is there any different or what should i concern?

after all, thanks a lot!

stephen

[Tim D Fulford]

Hi

I'm only giving my advice here. As it happens you can try replacing the disk in a number of ways and see what problems you face. You can judge from this what works. Basically you can do it any way you like, but I'd do it as above..

IMHO you have to run the lvreduce & vgreduce commands, other wise the system will continually try to use /dev/dsk/c0t0d0, and so you will not be able to replace it.

If you want to

Regards

Tim

[Pete Randall]

This is an excellent document that details the required steps:

http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad.pdf


Pete