Kernel panic on RHEL 5 and DB2 9

[martynh]

We are testing DB2 9.1 on RHEL 5, on DL380 G4 servers. At random intervals we get a kernel panic. The issue looks like one for Red Hat, but I ask here because the hardware and software combination we are using is very simple and conventional. The only *unusual* aspect is that we are running DB2 9. The problem occurs during periods of high io activity, often during db2 backups, but also during dump. The process that fails is kswapd0.

We have tried configuring kdump. Whilst this traps deliberately triggered crashes, it does not trap the random ones.

We have seen the same failure on 2 difference servers, so we doubt it is faulty hardware.

Any comments are welcome.

 

[sathyarams]

What fixpak are you on? If you are on the base version, install the latest fixpak.

Use db2cos to capture information to send IBM support.

http://blogs.ittoolbox.com/database...nformation-at-first-failure-with-db2cos-15677

Cheers
Sathyaram



For db2 resoruces visit

http://www.db2click.com

More DB2 questions answered at

http://www.dbforums.com/f8

&

http://www.idug.org/user/UserLogin.asp

 

[martynh]

We were using Fix Pack 2. I have just completed a complete rebuild of the server and have applied the interim fix pack 2a. The server crashed again within 2 hours of use.

IBM support pretty well ignored us when I raised the issue. Red Hat have done the same. As requested, we set up kdump and tested it with a controlled kernel panic. The dump was created. However when the server kernel panics 'for real' we don't get a core dump, so kdump isn't working in this case, either.

I think our next stage will be to revert to RHEL 4. If the failure continues to occur then I guess we will switch back to DB2 8.