Did a Mistake with HACMP

azzed27 · October 20, 2008, 11:26am

Hi,

I needed space on a FS, and when I've added the space on the filesystem, I did it trough the regular smitty fs inteface and not with smitty cl_lvm.
Can someone help me to repair the situat before a faileover happen ?
Thanks for your help,

shockneck · October 20, 2008, 11:39am

which AIX version, which HACMP version, ECM vg or standard vg, jfs or jfs2?

azzed27 · October 20, 2008, 2:36pm

Hi,

Thank you for your reply, here is the infos :
AIX 5.3 TL07
HACMP 5.4.1
Jfs2 and ECM do you mean EMC, if so yes the VG's are on EMC disks.

Rgds

shockneck · October 20, 2008, 2:58pm

thx for the info so far. But no, I did not mean EMC (the disk vendor) but if the VG was Enhanced Concurrent Mode ECM or a standard vg?

azzed27 · October 20, 2008, 3:34pm

How can I check this ? I'm not expert on HACMP

Thanks

shockneck · October 20, 2008, 3:36pm

try
# lsvg <yourvg>

azzed27 · October 20, 2008, 4:17pm

Concurrent : Enhanced-Capable
VG Mode : Concurrent

is that ok ?

Thanks

shockneck · October 20, 2008, 4:44pm

So it's ECM ok. Coming to think about that you extended the FS on a SAN - did you add a hdisk device to the VG in question or did you grow the FS on disk that was already known in the cluster? If you did add a disk, was its PVID visible (known) to all cluster nodes before you extended the FS?

cchiang12 · October 21, 2008, 7:49am

have you tried sync?

azzed27 · October 21, 2008, 9:46am

Hi,
Thanks for all reply,
I just grow the FS, I've not added disks on the cluster.
I didn't try the sync command, do you mean a syncvg ?.

I'm just preventing the case of a failover, and want to know if I can check that my HACMP is not broken.

Thanks again.

cchiang12 · October 21, 2008, 9:58am

sync as in hacmp sync.
"accident" happens all the time such as in your case.
if sync while hacmp is active not possible, you might want to think about bring the hacmp down and then sync it.

azzed27 · October 21, 2008, 10:01am

If I run a sync, is there an impact on the production ?
Thanks

cchiang12 · October 21, 2008, 10:03am

only if you bring the hacmp down.
you could sync it while it's active.
It will probably give you errors. Save the errors.

But if you want to be catious, do it when it is not busy.

shockneck · October 21, 2008, 1:39pm

Assuming that there are no more unpropagated changes to your cluster environment I think you don't need to worry. Probably the easiest way to react to what you did would be: do nothing. HACMP has a function called lazy update which would detect the change in the FS during failover/takeover and can synchronise the cluster nodes without manual intervention. There are two possible drawbacks to keep in mind though. Firstly: the next takeover will take somewhat longer. Depending on your Resource Group that might or might not be a problem. Secondly: there is the possibility to overload the lazy update if there are several unpropagated changes. In such a case a takeover can fail. So when I said "do nothing" this also means: do not change anything in the cluster again before it is in sync again. Hopefully there weren't any unmanaged changes already before you extended the FS. After it is in sync use CSPOC only to configure the cluster's environment. It might be a good idea to edit the cluster node's herald message to inform root to not use standard AIX configuration commands.

Should you feel unhappy with this, there is also the possibility to update the LVM information on secondary nodes manually.