MC/serviceguard lock device

[tecnomatix]

I have installed MC/serviceguard on 2 rp7410 machines.
The error message I'm recieving when trying to configure lock device is:
"There are no volume group disks that are shared by all nodes in the cluster, hence no cluster lock disk is available
you must have one cluster lock if you have only two nodes in the cluster."

The lock device is located on A SAN storage, Known as /dev/dsk/c9t9d7 (to one machine) and /dev/dsk/c5t9d7 (To the other machine)

I cannot pvdisplay the disks from both of the machines, because its imported only on one of them...


Thanks in advance,
Ayelet

[Bernhard Mueller · Accepted solution]

Ayelet,

just like the message says, you need at least one shared volume group, to use one of its disk devices as a lock disk.

So why don't you
vgexport -p -s -m /tmp/vgname.map
on the machine where you have the volume group

rcp /tmp/vgname.map other_host:/tmp

and on other_host do
vgimport -s -m /tmp/vgname.tmp

That should solve your problem.

Regards,
Bernhard

[Bernhard Mueller]

Before the vgimport you need to
mkdir /dev/vg_whatever_name
mknod /dev/vg_whatever_name/group c 64 0x##0000
where ## should be the same as for the vg group file on the other node

[Enrico P.]

Hi,
if you created a VG with the disk in the question you need to export the VG then import her in the other node. The VG shoud be shared from the two node.

Enrico.

[tecnomatix]

Thank you very much, it solved my problem...

I have a question, How come i export the VG, and he still acted like it shared on both machines?
Why didn't it remove it from the original machine?

Ayelet

[Enrico P.]

Hi,
the -p (preview) option in the vgexport command allow it.

Enrico.

[Bernhard Mueller]

Hello again,

the vgexport -p (preview) did not do anything but generate the map file for the vgimport on the other host.

I hope you do NOT use the volume group locally on the machine where it existed before (check /etc/fstab and /etc/lvmrc)

Remember that you should start the cluster daemon and do a vgchange -c y to make the vg cluster aware, if you have not yet done so. From then on you can activate the vg only by
vgchange -a e (exclusive access) which is important for cluster vgs!

Next you should deactivate it on the first node, activate it on the other node where you imported it and do a vgcfgbackup to save the config under /etc/lvmconf.

Regards
Bernhard