Service guard Errors when cmapplyconf

[likid0]

Hy, i am adding a new member to the cluster, and during cmviewcheckconf -v -k -C cluster i get this errors:

Warning: The volume group /dev/vglock on node centauro2 has inconsistent cluster configuration data
This can be cleared using vgchange with "-c".

Error: Volume group /dev/vglock has mismatched cluster ids. (1048677792 != 18196 36596)

any idea how can i fix this?

vgchange -c n vg
vgchange -c y vg

doesn't help, get the same one.

its 11.15 and 11.11
any idea?

[skt_skt]

run "cmgetconf" for the cluster and collect it in a file from both the node and compare . Give sepcial attention tot he cluster lock PV disk details.

Also check the cluster lock VG/disk on bot the nodes and make the cluster ascii file is updated prperly on both the nodes

[Steven E. Protter]

Shalom,

I think the cluster configuration files, the ascii ones are not the same on both nodes.

As noted it probably relates to the lock disk.

This message would come up if node 1 and 2 were using different lock disks.

That won't work.

SEP

[Rita C Workman · Accepted solution]

I'm going with the belief that your cluster is UP currently and all your trying to do is add a new node. So with your existing cluster up and running....

My guess is that whatever got imported into the new node - it was not a solid mapfile of the original /dev/vglock.

Make sure that the new node is mapped/masked to see the /dev/vglock disks.

On new node get rid of what is there using:
vgexport /dev/vglock

On original node recreate your mapfile.
vgexport -pvs /etc/lvmconf/vglock.mapxxxx /dev/vglock
Now rcp that mapfile to the new node and re-import

mkdir /dev/vglock
mknod /dev/vglock c 64 0x<##0000> where this minor is the same as the original node's
vgimport -vs -m /etc/lvmconf/vglock.mapxxxx /dev/vglock

Recopy over your pkg.ascii & pkg.cntl files (just in case) and then retry the checkconf

cmcheckconf -k -v -C

Just a thought,
Rita

[likid0]

is there a clear case, on when this happens?

i exported, imported vglock on all nodes and then it worked ok.

But i am not really shore in what cases you get this error:

volume group has mismatched cluster ids


out of date Vgs is one of them

[Rita C Workman]

My guess....and it's just a guess

You were working on some disk on the server that had ownership of the /dev/vglock at the time. You were probably adding/extending some disk and you ran pvcreate on some new disk you going to add to a volume group.
But you 'fat fingered' something and by mistake pvcreated existing disk.

Now your up when you do this, so things just go humming along and you think all is fine. Until normal operations, like a reboot or a cmcheckconf of your cluster, and now you discover that now something is wrong.

Like I said, just a guess..
Rgrds,
Rita