why vgchange error?

[Thomas Schler_1]

Hi,

trying to start a software package, I get the error message:

vgchange: Couldn't activate volume group "vgseismo_03":
I/O error while reading the VGDA.

What happened? How to resolve this problem???

At the moment, the software package can't be started on any server.

We're running two K460 servers, HP-UX 11.00, MC/SG A.11.14. Disks are mirrored (MirrorDisk/UX B.11.00).

[harry d brown jr]

What does "vgdisplay -v vgseismo_03" return?

Why are you starting the package, did you stop it gracefully or did you have a system crash?

What is the exact vgchange command you are attempting to execute? If it's from the startup script, I'd advise you contact your software people.

live free or die
harry

[Jerome Baron]

Hi,

Can you manually activate the vg ?
Are disk in a external array?

Regards,
Jerome

[Ken Hubnik_2 · Accepted solution]

So if all your volume groups and lvol are correct when you run a vgscan -v -p. The -p tells the command only to preen not to update. very important. READ man pages for vgscan.

[Dietmar Konermann]

Check if you are able to read physically from the VG's devices, e.g. using dd(1).

If that works fine you may hit one of the problems reported as JAGad43153 and JAGad55510. Both are fixed with PHKL_24646 and later (latest is PHKL_27178).

So long...
Dietmar.

[T G Manikandan]

Do a dmesg and check for the messages.
Also check your syslog.log file for any hard disk problems.

#ioscan -fnC disk
# diskinfo /dev/rdsk/cxtydz


Thanks

[Thomas Schler_1]

Thank you very much for your immediate help. Meanwhile I solved this problem with the help of the HP support.

This is for your information.

Harry:
"vgdisplay -v vgseismo_03" did nothing return because the volume group could not be activated (see above).

I did stop the package gracefully.

The command was: vgchange -a e vgseismo_03

Jerome:
1st question: No. 2nd: Yes.

Ken:
Thanks.

Dietmar:
Thanks for the patch hints. I, already, had installed PHKL_27178.

TG:
dmesg did no give hints. syslog indeed has several entries like "LVM: vg[5]: pvnum=8 (dev_t=0x1c062000) is POWERFAILED". Sometimes the specified disk come back, sometimes the server uses more and more resources until the server hangs. Then I have to TOC the server. After that all the hardware works fine, for several days, until, again, similar "POWERFAILED" messages are listed in syslog, and the server, again, has to be rebooted. The HW group of HP will work on that, but, maybe, you have some worthful hints on these strange HW failures?

'ioscan' all the time reports all HW in claimed status. 'diskinfo' also reports normal information on powerfailed disks (after the reboot).

----

At the moment, all works fine. The solution was to recover the VGDA using 'vgcfgrestore'. But, I still don't know what destroyed the VGDA.

[T G Manikandan]

x1c062000

it is your c6t2d0 disk which is creating problems.
If you notice scsi reset messages in the syslog,
When you tell it comes back,
usually there is a scsi hung,followed by a scsi reset.
check the scsi cables and the scsi termination.

If the error is followed by a message about pv returned to vg, then usually be to a low timeout value on the disk driver. By default, this timeout is 30 seconds.

change it by using pvchange.

Also check your I/O on the specific disk.
use sar -d to check it.

Also this might be due to a hard disk going bad.

Thanks