>       Thanks Paul, this is very useful into.

5. We have a monitor that checks the log utilization every 5 minutes (among
other things).  If the log utilization gets above 82%, the monitor sends
pager alerts to the TSM oncall staff.  The oncall staff then logs on to the
server to try to identify what has the log pinned and cancels it.  We also
immediately cancel an inventory expiration if it is in progress, as that is
the most likely thing to make the log utilization skyrocket quickly.  This
used to work pretty well on our old server, but now that we have upgraded
to an RS6000/M80, the log can fill up MUCH more quickly during inventory
expiration.  To address this, we will probably automate the killing of the
expiration process in this situation.

>       You know, I don't think we oughtta be writing monitors and scripts
for this; I think we have a fundamental management problem here that the
developers should address.


6. There are two server options, throughputtimethreshold and
throughputdatathreshold, that were added to the server recently to detect
and automatically kill slow or hung sessions.  We have been experimenting
with this, but have found that it sometimes kills sessions that we don't
think should be killed.  I think the sessions in question might be ones
that have been waiting a long time for a tape to be mounted.

>       FWIW, this happened to us last weekend for the first time in a
while, and it appeared to me also the guilty sessions were ones that were
waiting on tape mounts.

************************************************************************
Wanda Prather
The Johns Hopkins Applied Physics Lab
443-778-8769
wanda_prather AT jhuapl DOT edu

"Intelligence has much less practical application than you'd think" -
Scott Adams/Dilbert
************************************************************************