Moving resource group in HACMP/Power HA is slow

[holdahl]

I'm testing on of our HACMP clusters by moving one resource group between 2 nodes.

The moving takes a long time and I get this messages in the log:
PollAliasEvents: State not STABLE/RP_RUNNING or ibcasts, return

Anyone seen this message and what might cause it?


-holdahl

 

[MoreFeo]

Found this:

http://www.notesofanerd.com/index.php?topic=167.0

"Turns out that the error is related to LAN traffic "
"my rsct.core.* and rsct.basic.* filesets were at different levels"
"when the two nodes did not have identical MTU sizes"

So I'd check rsct levels, and communications.

 

[holdahl]

Have looked at the things suggested in the article you refer to above but all those are ok.

rsct.core and rsct.basic filesets are all at the same level.
MTU size is the same.

Got explained (by an employee at IBM working with hacmp for many years) that the PollAliasEvents message is saying that the movement of the resource group has not finished. (cluster is not stable before the movement is complete)

After investigating the HACMP logs it might look like an unmount of some filesystems is the cause of my problem.