cluster lock disk which one ?

[Sammy_2]

cmcld: WARNING: Cluster lock disk /dev/dsk/c8t0d3 has failed: I/O error


hpux 11.11 running MC serviceguard with RAC. (2 node e-rac cluster)

We (pvmmove) moved all the disks from one array (EMC) to another (Hitachi). That changed the device names on new array. c8t0d3 does not really exist.

Now, how do I configure cluster to see the lock disk on new array?
I forget, Do you choose the first disk in the shared vg to be the lock disk even though it has data in it ?


cluster, pkg on both nodes is running fine. I just see lock error.

[Matti_Kurkela]

Yes, you can use the first disk in the (first) shared VG. The "Managing ServiceGuard" manual says:

"The lock disk is not dedicated for use as the cluster lock; the disk can be employed as part of a normal volume group with user data on it. The cluster lock volume group and physical volume names are identified in the cluster configuration file."

To configure the lock disk:

1.) use the cmgetconf command to get an up-to-date cluster configuration ASCII file.
2.) edit the cluster configuration ASCII file to have the correct device names for the lock disk.
3.) use the cmcheckconf command to verify your configuration is valid.
4.) schedule some cluster-wide downtime
5.) at your planned downtime, shutdown the cluster (cmhaltcl), use the cmapplyconf command to apply your new cluster configuration and restart the cluster.

Any changes to cluster lock configuration must be done when the cluster is not running.

MK

[skt_skt]

i would recommend taking an outage to fix diffrence between the current config and the running config.

If it is too difficult to get a down time , then as a work around create the old ctd device with the same major&minor combination as the new device/ctd. So effectivly it is will see as same deice(same M&M number) wit two names/aliases.

[Ramesh S]

Hi

Please go theu this ITRC Doc :UXSGKBRC00017956

Best Regards,
Ramesh

[melvyn burnard]

if you have used pvmove to move the LUN, then this will NOT have copied across the cluster lock flag.
You will have to halt the cluster, then correct the ascii file tgo reflect the correct device, for ALL nodes in the cluster, halt the cluster, then vgchange -c n vgchange -a y and then use cmapplyconf to make the relvant changes to the cluster binary AND mark the LUN as a cluster lock device,