vgdisplay -v unsuccessful

[J.D._3]

I am trying to run a make_tape_recovery -Av -x inc_entire=vg00 and it would not complete. It normally takes about 35 mins or so but this time it just hangs. This is part of a SG cluster and I was able to create the ignite tape last month. Attached is the recovery.log and below is a listing of swlist for ignite versions. This is an HP-UX 11.00 and is of March 2003 patch set. Thanks in advance.

B5725AA B.4.4.30 HP-UX Installation Utilities (Ignite-UX)
IUX-Recovery B.4.4.30 Ignite-UX network recovery tool subset
Ignite-UX-11-00 B.4.4.30 HP-UX Installation Utilities for Installing 11.00 Systems
Ignite-UX-11-11 B.4.4.30 HP-UX Installation Utilities for Installing 11.11 Systems

[Steven E. Protter · Accepted solution]

A common issue here is a disk in /etc/lvmtab thats not responding.

I'm attaching a script that checks for that and can let you figure out if its lvm or a bad disk.

Later version of ignite require all disks in lvmtab to be working.

SEP

[Patrick Wallek]

They problem is:
* Backing Up Volume Group /dev/vglock
* /usr/sbin/vgcfgbackup /dev/vglock
vgcfgbackup: Warning: couldn't query physical volume "/dev/dsk/c51t0d1":

It appears that you have a bad disk or LUN or do not have access to it. The error is telling you that it can't query the LUN at /dev/dsk/c51t0d1. You need to figure out what the problem there is.

This could also be bad news because this appears to be your MC/SG lock disk in the vglock VG. If it is and you have a problem with one of the machines in the cluster, you could have real problems like "split brain syndrome" where each machine tries to form the cluster.

Take care of the disk/LUN problem and your make_tape_recovery problem will probably go away.

[J.D._3]

Steven,

I ran your script but it indicating that my SG lock disk is not responding. That should be true because right now the primary node has it activated. The recovery log does give Warning messages about the /dev/vglock VG but it continues until ignite creates the Archive File List. Then it just hangs with a NOTE: vgdisplay -v unsuccessful. Thanks.

[J.D._3]

Steven,

my mistake. The lock disk is not activated in either node. Neither node can do a vgdisplay on the VG or a pvdisplay on the LUN. I did get a successful ignite on the primary node on the same night.

[J.D._3]

Patrick,

Same reply as above. The LUN is inactive on both node but the primary node was able to get a successful ignite image. Thanks.

[Patrick Wallek]

If that is truly your MC/SG lock disk, then that LUN *must* be active on both systems. Both systems have to be able to get to the lock disk in order for MC/SG to function correctly.

[Patrick Wallek]

Lock disk info:
http://tinyurl.com/2pqfg

[Vincent Abad]

Hi J.D

That care that you are in a cluster environment.
If you vglock is part of a Serviceguard,
you surely put it under cluster control,
with command "vgchange -c y vglock".
The only way to activate a vg in a cluster
is to activate it with exclusive option to discard data corruption.
You could perform this with the command
"vgchange -a e y vglock".

If the command works, you will need to run this command before launching ignite backup.
and then deactivate it with command
"vgchange -a n vgsglock" and performeing vgdisplay

For your information SG doesn't use public part of vg for lock mechanisn, put the private structure part.

Hope this help!

--Vincent