Re: How to locate immediate spikes in Disk Usage

Mike,


>A few times now, I have noticed that the system disk usage will jump from
5% ini a matter of a an hour or 2. I check fairly often during the day. >Is
there a way to immediatly pinpoint what has happened? Once it turned out to
be a restore of a large file that someone cancelled in the middle. >The
space is already allocated however. It was returned either during a disk
report that I ran or by the IPL. It has happened again, but in this >case,
I did restore a library, but it was not 5 % for sure, and I did not cancel,
yet this is a similar activity. Any ideas and help would be >greatly
appreciated. Mike

in the most cases that happens here, it is some (interactive) Query
building an complicated index. We had queries building up to
20 (twenty) GB of index. But this is only temporary storage space and
should show up on the wrksyssts screen (right upper half, unprotected
storage). On our system, this value is between 5GB and 10GB, so when it
goes above 10GB, I know something is going on.

Somebody posted a little program that shows which jobs use the most temp
storage. Search the archive..

After you have found and killed the query job (and probably the user) the
temp storage should be freed again (at least most of it).

Other storage killers are undeleted (and unneeded) journal receivers, but
that is more long-term.

HTH,

Oliver

(Quote of the day:  HIT ANY USER TO CONTINUE!)