I'm looking at NSS snapshots as a way to better deal with our backup
problems. Only, we run a cluster and our primary SAN device is running out
of expansion space so I'm pretty sure I don't have enough slack space to
hold the snaps.

So, out comes the creative thinking hat.

It says right there in the Snapshot documentation that snapshots in
clusters need to have the stored-on pool be identical to the original
pool. The reason is pretty clear, in order to handle cluster failover.

So I got to thinking. What if I modified the load/unload script for my
cluster resource to also load and unload a second pool (stored on a
different SAN device)?

nss /poolactivate=CLUPUBLIC
mount FACPUBLIC VOLID=248
CLUSTER CVSBIND ADD CLU-PUB 172.17.246.209
NUDP ADD CLU-PUB 172.17.246.209
add secondary ipaddress 172.17.246.209

And make it:

nss /poolactivate=CLUPUBLIC
nss /poolactivate=CLUPUBLIC-SNAP
mount FACPUBLIC VOLID=248
CLUSTER CVSBIND ADD CLU-PUB 172.17.246.209
NUDP ADD CLU-PUB 172.17.246.209
add secondary ipaddress 172.17.246.209

And for unload script:

del secondary ipaddress 172.17.246.209
CLUSTER CVSBIND DEL CLU-PUB 172.17.246.209
NUDP DEL CLU-PUB 172.17.246.209
nss /pooldeactivate=CLUPUBLIC /overridetype=question
nss /pooldeactivate=CLUPUBLIC-SNAP /overridetype=question

This may ultimately be a question for the Clustering group, but does this
work the way I think it does? Or am I in for a world of woe?

--
Novell, it does a network good