I’ve been migrating from an AIX 5.2 HACMP cluster (Cloverleaf 5.2P2) onto an AIX 5.3 T08 HACMP cluster (Cloverleaf 5.6R2.) and have both working cureently.
As I read your email I got the impression you are talking about doing the migration in place.
I was not brave enough to do that even if that was possible for me which it wasn’t.
My version of Cloverleaf 5.2 would not run on AIX 5.3 so that is the first thing you want to check if you are brave enough to do the upgrade in place.
What I did was create another fresh HACMP cluster with 2 fresh LPARs and layed down everything fresh (AIX 5.3 T08, Cloverleaf 5.6R2) so I had that going parallel with my old HACMP cluster (AIX 5.2 & Cloverleaf 5.2).
Once that was done, I began migrating each site one at a time with no noticeable down time ( 5 minutes or less) to our customers.
This also reduces my risk but is more work for me and takes longer but I have almost no stress doing it this way.
With regards to your question, I don’t recall having to modify my HACMP scripts much.
I did switch to using ether channel and had to modify the network card aliasing in my HACMP scripts so my service address IP is always the primary address used instead of the hostname IP.
Prior to using ether channel, I simply had my HACMP script detach and reattach the NICs to set the order of the IPs used by the active node.
This is how I got the service address IP to fail over transparently regardless of what node I’m are running on.
Russ Ross
RussRoss318@gmail.com