Hello everybody,

Can anybody explain what could be the worst scenario/situation which could happen if we do not setup superstrict mirror for a shared logical volume in HACMP cluster (the logical volume is mirrored and configured on two disks in different datacenters).

Hello,

the worst thing will be you may not be able to remove a failed disk from the VG.

Example:
disk1 disk2
LP1/1 LP1/2
LP2/2 LP2/1

Now if one disk fails you have to remove all LPs from that disk before you can remove the disk from the VG. Which mirror would you like to remove? In any case you will lose you data, so the removal of a mirror is not possible.
Starting with AIX 6.1 you can use the Mirror Pools to ensure that the different mirrors are located on different storage subsystems, but HACMP still expects the strictness set to superstrict.

Regards
Claus

Assuming both are SAN disks, and one of them fails (and they are mirrored) shouldn;t application continue to work without any problem ? Even if there are mixed copy1 and copy2 on the same disks, if one copy is not accessible, lvm will use the 2nd copy, isn;t it true ?

Yes

OK, so the only issue (and not negligible !) is when you need to replace a failed disk which has mixed copies.
Thank you for explanation.