SFM consuming too much cpu

[celio_m]

Hi all,
We have an HP-UX 11iv2 and the /var FS is growing. I noticed that the biggest files from that FS are at /var/opt/sfmdb/pgsql/base/…. (running for example:
find /var -size +10000000c 2>/dev/null -type f -exec ls -s {} \; | sort -n –r)
Also, from ‘top’ command, the most consuming processes are:
CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
0 ? 6628 sfmdb 197 20 22688K 1816K run 130:17 29.63 29.58 postgres:
1 ? 8180 sfmdb 134 20 22688K 1684K sleep 129:20 28.81 28.76 postgres:
I have another similar system and that is not happening.
Any idea of what may be causing that overhead?
Thank you all.

[TTr]

> I have another similar system and that is not happening

HP-SIM is using postgres as its database. Either you run SIM on this server and not on the other one or SIM is generating a lot of data on this server which is then stored in the database files in /var/opt/sfmdb. If SIM is generating a lot of data, it could be an indication af a hardware failure or a system inconsistency that SIM is complaining about.

[celio_m]

Thank you TTr, but how can I check your suggestions?

[TTr]

I don't know much about SIM other than it is a large enterprise level monitoring system with many extension agents into the OS and hardware. Maybe somebody that has worked with it can give you an idea where/what to look for. I used to have an agent installed that came packaged by an app vendor. Its default location is under /opt/mx.
Take a look at

http://docs.hp.com/en/5991-0710/5991-0710.pdf

page 44 talks about the running services.

Also general info at
http://h18013.www1.hp.com/products/quickspecs/11824_div/11824_div.html

[David Woodroffe]

Not sure if it related but had a similar problem with SFM on V2 and excessive CPU usage.

To resolve it I put in the latest SFM software, did an upgrade-ux to the latest version of the OS and finally found a duplicate entry in cron for the restart_sfm.sh process. Actually opened a HP case but ended up solving it myself.

Oh and I don't have SIM installed.

[Shinji Teragaito_1]

Hi,

Did you check if a hardware problem is reported or not ?

% evweb eventviewer -L

You may find the root cause of your problem from there.

Out of curiosity, I really appreciate if you could give us your SFM version and diag mode.

% swlist | grep SysFaultMgmt
% sfmcofnig -w -q

Thanks,

Shinji

[Mark McDonald_2]

May not be related, but I had problems with SMF. I had 2 identical systems (Prod and DR)

SMF worked fine on DR, but when administering users in SMF in Production, there was a massive overhead on the system.

Turned out the SMF was checking through the wtmp etc files. Once these were massively cutin size, things improved.

[celio_m]

Thank you all,
Shinji, about evweb records, I have no entries. The version and mode you asked are as follows:
SysFaultMgmt B.03.00.04.01 HPUX System Fault Management
EMS hardware monitors are enabled & SysFaultMgmt is not monitoring devices.
Any way, it seems like the problem comes from a HW problem. I opened a case at support to see what they find. I'll let you know the result...
Thanks.