Cloning configured PowerHA node with mksysb for standalone system.

[foobar13]

I put this question to AIX support and received a "that's not supported" answer.

Anyone got a workaround for this issue? Seems pretty basic.

Product or Service: POWERHA SYSTEMMIRROR V7.1 Technology Level1 7.1.1
.
Operating System: AIX
.
Problem title
How to prevent accidental cluster deletion from mksysb clones?
.
Problem description
The scenario: We have a working PowerHA 7.1 cluster running under AIX
7.1. We create a mksysb of one of the nodes and clone another separate
standalone node using NIM. The new standalone node gets a new IP
address, and a new name. We delete the cluster config from the
standalone node and it reaches out and deletes the cluster config from
what it thinks was its partner cluster node, which is still running in
the cluster. Ouch.

One workaround I see is to pollute the /etc/hosts file on the
standalone before attempting the cluster delete to redirect to a non-
existent node; perhaps 127.0.0.1.

I would prefer to exclude something on the mksysb creation. Can you
suggest a workaround?

Thanks.

 

[ogniemi]

can't you delete the configuraion from standalone before IP is configured on it?

 

[foobar13]

Well yes, I could. I was hoping to stop this at the source to prevent accidents.

 

[mrn]

Could you not just use the -e flag on mksysb to exclude the config and/or hosts file?

-e Excludes files listed in the /etc/exclude.rootvg file from being backed up.

Mike

"Whenever I dwell for any length of time on my own shortcomings, they gradually begin to seem mild, harmless, rather engaging little things, not at all like the staring defects in other people's characters."

 

[foobar13]

AIX 7.1 is cluster aware, and it appears that the node configuration, including IP address is in the ODM, probably somewhere under /etc/obj/ es. If I exclude the ODM configuration, it might make things more complicated. Another thought I had was to encrypt the cluster communication and then exclude the private key.

 

[DukeSSD]

How about deleting the cluster from the console before you put it on the network...

 

[foobar13]

I would prefer to handle this at the source to prevent accidents. We have several team members and if we could devise a way to do this that's relatively foolproof, that'd be a good thing.