Greetings, all.
I'm trying to work out exactly how IBM thinks virtual volume
reclamation is supposed to proceed, and also how we think it does in
fact proceed. ;)
Noodling around in the docs and QuickFacts, I find that volumes of a
devclass of type SERVER "may not be set to access=offsite", and
experimentation confirms this.
Such would lead me to the conclusion that SERVER devclass volumes
would be reclaimed in a manner logically equivalent to local volumes,
to wit a source virtual volume and a target virtual volume would both
be mounted, the source read from and the target written to. This is
also what Dave said in Oxford, and what I see happening on some of the
servers.
However, on one of them servers I observe reclamation proceeding in
the 'offsite' manner, with a new virtual volume being built from
(many!) mounts of primary volumes. I'm not sure how I can control
this, but the difference is giving me the aggravations something
fierce. I can see circumstances in which I might want to do things in
either way, but optimizations in favor of one are a pain for the
other.
Any ideas?
- Allen S. Rout
|