vglock disk

[Animesh Chakraborty]

Hi all,
I have lost my vglock disk while creating new volume group using sam as it was showing unused.Now how can I retrieve that vglock?
can vgcfgrestore help??do I need to shutdown the cluster?I have removed the new vg created via SAM.
In syslog I am getting
cmcld: WARNING: Cluster lock on disk /dev/dsk/c5t5d7 is missing
Any help ?
Thanks
Animesh

[eran maor]

Hi

The warning message about "Cluster lock on disk /dev/dsk/... is
missing!" could be result of corrupted SG binary OR if the cluster lock
disk was replaced or moved to a different disk.
Make sure cluster lock VG and cluster lock PV is defined in the
ascii cluster configuration file, then to fix:

1. Halt the packages and cluster.

2. Activate the cluster lock VG on the configuration node ONLY, so the
lock disk can be initialized, for example:

vgchange -a y /dev/vglock

2b. If needed, do the following on the VG:

vgchange -c n /dev/vglock

3. Generate the cluster binary configuration file & distribute
to the nodes with cmapplyconf command, for example:

cmapplyconf -C /etc/cmcluster/cmclconf.ascii
-P /etc/cmcluster/pkg1/pkg1conf.ascii

4. Deactivate cluster lock VG.
vgchange -a n /dev/vglock

5. Restart the cluster

[Vincenzo Restuccia]

See this link very good:
http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0x440ca24d9abcd4118fef0090279cd0f9,00.html

[Carsten Krege]

The cluster lock information is stored in the LVM header of the disk. Therefore a vgcfgrestore will restore the LVM and cluster lock information on the disk (but only if the configuration has been saved after the last cmapplyconf: cmapplyconf writes the cluster lock information to the disk, but does not automatically issue a vgcfgbackup command. Therefore it requires that vgcfgbackup is run after the last cmapplyconf).

Alternatively you can apply Eran's procedure, but this would require a downtime of the cluster.

SG checks the health of the cluster lock hourly and will log a message to syslog if the cluster lock information on disk is missing.

Carsten

[Animesh Chakraborty]

Hi all,
Thanks for your quick response.
In normal situation vglock should be activated or deactivated on configured node ?
In my case, I think I have already corrupted the vglock so will vgchange -a y work ?
Do I need to create a fresh vglock?
vglock entry still there in lvmtab file.
Can any one tell me the correct steps?

Thanks
Animesh

[Carsten Krege]

>In normal situation vglock should be activated or deactivated on configured node ?

During normal operation the cluster lock VG doesn't need to be activated. If you want to (re)apply your cluster configuration the cluster lock must be activated on the node where you execute cmapplyconf.

>In my case, I think I have already corrupted the vglock so will vgchange -a y work ?

No. If you erased the LVM header from the disk (e.g. by pvcreate, dd) , vgchange will not work. You can try it though and if vgchange worked, you obviously did NOT destroy the LVM header.

>Do I need to create a fresh vglock?
vglock entry still there in lvmtab file.

If you still have an entry in the /etc/lvmtab for this VG, you cannot pvcreate the disk (on this node). So at least it is unlikely that you pvcreate'd the disk and destroyed the LVM header by this command.

> Can any one tell me the correct steps?

First find out whether the cluster lock is healthy. If SG doesn't log messages in syslog within 1 hour, it probably is. Your local Response Center has tools to help you verify the status of the cluster lock area, too.

If you are sure that the cluster lock area has been destroyed, you can reapply it with

# vgcfgrestore -n /dev/vglock /dev/rdsk/cXtYdZ
(if you have a current vgcfgbackup on one of the node: check dates of /etc/lvmconf/vglock.conf!)

If you don't have a current vgcfgbackup, you need to follow Eran's procedure:

- Halt the cluster
- Activate the cluster lock VG locally
- Run cmapplyconf

Carsten

[Animesh Chakraborty]

Hi carsten,
Inspite of the entry in lvmtab file SAM did a pvcreate using alternate path and while adding alternate path it showed an error that can not add alternate link as there is entry in lvmtab file.after that I am getting mesg in syslog that cluster lock disk is missing.....

While trying vgcfgrestore I am getting follwing error:
#vgcfgrestore -n /dev/vlock /dev/rdsk/c5t5d7
vgcfgrestore: Couldn't query physical volume "/dev/dsk/c5t5d7":
The specified path does not correspond to physical volume attached to
this volume group
vgcfgrestore: Mismatch between the backup file and the running kernel:
Kernel indicates 4 disks for "/dev/vlock"; /etc/lvmconf/vlock.conf indicates 1 d
isks.
Cannot proceed with the restoration. Deactivate the Volume Group and try again.

Thanks
Animesh

[Animesh Chakraborty]

Hi all,
Do I need any thing to do on adoptive node ?
Will that previous exported vglock will work or I need to do export it again to secondary node?
Or can I make use of this ,if healty ,to do a export from adoptive node to configured node.


Thanks
Animesh

[Carsten Krege]

Sounds like the problem was due to SAM not being smart enough..

I recommend you do the following.

0) Check the availability of the disk.

# ioscan -fnCdisk

1) Make the VG consistent.


# cp -p /etc/lvmtab /etc/lvmtab.before
# vgreduce -f /dev/vglock
# cp -p /etc/lvmtab /etc/lvmtab.after
# vgscan

2) Restore the LVM information to the disk.

# vgcfgrestore -n vglock /dev/rdsk/c5t5d7

3) Update lvmtab

# vgscan

4) Tell the kernel about the new disk

# vgchange -a y vglock


5) vgexport/vgimport vglock on the secondary node to keep lvmtab consistent

6) If this procedure failed, you can copy the old lvmtab back into place

# cp /etc/lvmtab.before /etc/lvmtab

7) If it succeeded, run

# vgcfgbackup

It is likely that you get this config recovered, however, there are some uncertainties what the output of every single command will be. It might happen that a vgscan in step 3) is not sufficient, but a complete vgexport/vgimport (and therefore VG deactivation) is needed.

Good luck: Carsten