Hi all,

need expert advice on this. I don't know whether this is a correct
method or not. The plan is to replace 2 SSA adapter in a server running
in a hacmp environment.

There are 2 ssa adapters (node1) which are faulty and it is connected
to a SSA disk. Both SSA adapter has fast write cache turn on. The steps
to replace both cards are :

- varyoffvg sharevg
- exportvg sharevg
- rmdev -dl ssa0
- rmdev -dl ssa1
- shutdown server
- Take out ssa0 and take out fast write cache and put into replacement
ssa.
- Take out ssa1 and take out fast write cache and put into replacement
ssa.
- Power up server.

Is that it? Do I need to do something from the hacmp level? I can't
find anything that related with HACMP in the SSA adapters documents. I
really appreciate any help from anyone.

Regards

Move any resources from the node1 you are bringing down to another node
in the cluster and stop cluster services on node1. You'll have to
remove the pdisk and hdisk devices before rmdev'ing the ssa adapters.
Maybe make sure your pdisk and hdisks are in the same relationship
after the power up of the server.

LABEL ALL CABLES!!!!!

How many VG's involved?
if just the one just
>move the resources to the other HA box
>stop the cluster services on the system with the bad SSA adapters
>shutdown and replace the adapters
>fire it back up... if the cables are all in the original places, then the pdisk hdisk mapping should come out the same.

you really don't need to remove all that stuff... if the pdisk/hdisks
do get all screwed up then rmdev them and cfgmgr... they should come
back in with PVID's as well as the associated VG's...

fire up the cluster services and failback the stuff you want on this
system...
hope it helps