Greetings, all.  I'm dipping my toes into the simultaneous-write
regime with server-to-server volumes.  I'm seeing a behavior I'd like
to confirm.

I've got a few boxes which do small backups every few minutes of the
day.  I'd been hoping to get them all doing simultaneous-write to my
remote copypool.

But I'm seeing an odd behavior:  even when there's a remote volume
mounted and IDLE, the simultaneous-write process opens a new one each
time.  This means I end up with (for each small incremental session) a
~100MB and a ~5MB volume.  For servers with TB scale total occupation,
I expect bajillions of tiny volumes, and I expect that reclamation
processing would become a nightmare.


I can understand why we don't try to re-mount server volumes, there's
no telling where the rest of the data for a given volume might be.
But I'm kind of surprised that I'm preferring a scratch mount to an
existing idle mount.  I'm guessing that the fact that the mounted
volumes are deemed "full" is what's making the difference.


So has anybody piddled around with this sort of thing?  Confirm or
deny on the mount crierion?


- Allen S. Rout