We've got an RS/6000 41T running the 1.2.0.8 ADSM server and AIX
3.2.5.  The ADSM server also runs a nifty monitor tool called xmconsole
that displays bar graphs that get updated in real time and shows cpu
usage, context switches, and system calls, among other things.
  Last night while waiting for a suspect tape drive to fail, we watched
a reclaim process.  In half an hour, it managed to move only four
colocation clusters, 21 files in all.  Each time ADSM started on a new
cluster, we could see the cpu, context switch and system calls graphs
shoot up for 5-10 minutes while ADSM did some sort of processing for
the cluster.  After the wait, the graphs would dive down, and ADSM
would transfer the cluster from one tape to the other in a few seconds;
then the usage would shoot up again for processing the next cluster.
  This seems particularly slow and overly busy.  Any idea what ADSM is
doing so energetically with the cpu during reclaims?

Wendy Alberg, Computer Resources Technical Support Group, 315 CCC,
Cornell Information Technologies, Cornell Univ., Ithaca, NY 14853
wa10 AT cornell DOT edu    607/255-2672