I think that the server was processing based upon your 25%
specification. I suspect that the server builds a rather large
internal list of what vols it needs, and in what order, to be able to
create new copypool vols. Offsite reclaimation works by copying data
from onsite tapes to NEW copy copypool vols and then updating its db
to point to the new location of the data. This in turn causes the
original tape(s) to become less utilized, where it could then become
eligible for reclaimation.
How's this:
OFF001 is elible for reclaimation.
Server determines what data on OFF001 is current and where the data
exists in the onsite tapepool. Build list of onsite vols needed to
create new copypool vol, say OFF100. BTW it appears to me that
reclaimation also eliminates SPANNING. So while coming up with the
list of files on OFF001, it is possible that 1 or 2 files span
multiple vols. So therefore the process of reclaimation will also
free up some space on OFF002 and/or OFF003, which *could* fall below
25% utilization, and then also become eligible for reclaimation.
The process begins again for the next vol. I have noted that nothing
happens until this is all worked out.
Factors involed are the number of tapes in each storage pool,
reclaimation %, db size, # of files, expiration processing, frequency
of reclaimations, and your device class mount limit, because the
server will parallelize the reclaimations in order to get them done
asap.
What I do here is keep offsite reclaimaion at 100 (turned off) during
*normal* processing times. Nightly I set copypool reclaimation to
75%. After that completes, I then backup tapepool to copypool,
followed by turn off copypool recaimation, the mark new copypool vols
(access=readw) to offsite. There's a few db backups in there too.
Don't forget setting REUSEDELAY, both for tapepool and copypool, in
consideration of the number of db backups you keep and could
potentially restore.
IMHO if you're intent on setting reclaimation to 25%, it would
probably be faster to DELETE the copypool vols, and then backup
tapepool to copypool.
Hope this helps,
Al Barth
Zurich Kemper Investments
______________________________ Reply Separator _________________________________
Subject: Copy pool volume reclaimation
Author: Naveed Mustafa <Naveed_Mustafa AT MCKINSEY DOT COM> at ~Internet
Date: 9/11/97 10:56 AM
Hi there,
We are running ADSM server version of 2.1.0.12 on AIX. I have a question
regarding the reclaimation of the offsite Copy storage pool volume.
According to IBM and also from the ADSM Admin guide for AIX, in order to
reclaim the offsite tapes belonging to the Copy pool, we need to the
following and I quote from the Admin guide:
"When you do perform offsite reclaimation, it is recommended that you turn
on reclaimation for copy storage pools during your storage pool backup
window and before marking the copy storage pool volumes as OFFSITE. Next,
turn off reclaimation and then mark any newly created volumes as OFFSITE."
We run our primary storage pool backups to the copy pool on Saturdays at
9:00am. Therefore, I created an admin schedule to kick off at 9:30am on
Saturday, during the backup of the primary storage pool to the copy pool,
that will lower the reclaimation threshold of the copy pool volumes to 25%
from 100%.
The admin schedule kicked in at the right time and in the activity log, I
did see that it listed a number of copy pool offsite volumes that will be
reclaimed. Also, "query process" listed SPACE RECLAIMATION process for the
offsite copy pool volumes as one of the active processes, but nothing was
reclaimed and not a single byte was moved to the newly created copy pool
volumes. I left that active process for two days but again not a singly
byte moved. Then, I tried to kill that process but was unsuccessful.
Finally, I change the reclaimation of the Copy Pool back to 100% and change
the access of the new copy pool volumes to offsite and checked them out of
the library. IBM suggested to recycle the ADSM in order to kill the space
reclaimation process of the offsite copy pool volumes which offcourse killed
that process.
Does any body know why the reclaimation of the offsite volumes did not
occur? I know that I should not actually need to have the offsite volumes
in the library in order for them to be reclaimed. The active files that
are on a reclaimable offsite volume should be picked up from the primary
pool and should be put on the newly created copy pool volume. But why did
it not perform the reclaimation? Am I doing something wrong?
All suggestions are welcome and are greatly appreciated.
Thanks,
Naveed.
|
