Here's a stab at intuitive problem determination.Maybe the object count
you got was after an ovtopofix, and you have oid entries in a seedfile to
omit certain oids. The actual object count goes quite a bit higher between
ovtopofix's and you exeed the cache setting. The extra objects are just
stub objects to prevent actual discovery, but they still take up space and
it may be my imagination, but I believe I have seen the need to keep the
cache setting high enough to accomodate most or all of the max count.
Also, I would increase your paging space to double the memory. The
percent utilization is not really a reliable indicator of how much you
need.
I just double it, unless you have 1G.
Take a look at the nettl log. It is in /usr/OV/log. To view it while
formatting it,
say netfmt -F -f /usr/OV/log/nettl.LOG00 or whichever log it is on.
If it is
big, clean it out first: nettl -stop, rm nettl.LOG*, nettl -start, then the
netfmt.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
==========================================================================
=======
We recently upgraded to Netview 5.1 from Netview 5.0 on AIX. After the
upgrade we began to see these processes (ovtopmd and netmon) using heavy
amounts of CPU (25% -> 75%) using top or monitor. Has anyone else seen
such
a problem? I figured I would try here before opening a PMR.
Our environment
===========
* Netview server is a 7015-R50/2 (multiprocessor) with 512M real
memory, 832M of page space (81% used), AIX4.2.1.0
* 8 maps (all local, all practically identical)
* ovwdb runs with the -O -n7000 -t
* typically 2 to 3 users use netview simultaneously
* ovmapcount resulted in 1186 objects in object database
* ovtopofix -a (requires an ovstop netmon) seems to slow them down.
But after awhile the CPU goes back up
Where would I look for error logging? Is my topology database corrupted?
Any help would be appreciated.
Bryan
|