Itanium server getting restarted

[bijoy k]

Dear all,

we have an Itanium server. which is
running on Redhat AS 2.1 version. for the last 6 days the server is getting restarted (Hardware restart), automatically once in a day. we have changed the hotswap SMPS in the server.and checked the cooling fans inside the server.
i am attaching the dmesg log of the server,while restart. and i would like to know whatit means by the following line.

Jul 30 14:18:53 srfifbtps kernel: oracle(16200): unaligned access to 0x2000000000f2ebc8, ip=0x40000000032822f0

Responses highly appreciated in advance....

thanks and regards...

bijoy. k

[Steven E. Protter]

Shalom,

You are the victim of a hardware failure. If the machine is under hardware contract the best thing to do is let HP hardware come out and fix the problem.

SEP

[Alan_152]

do you have an iLo/mp installed in this unit? If so, you can look in the system logs for any incidents of alert level 3 or higher. In text mode, the system log is fairly descriptive.

[Ivan Ferreira]

Connect to the management processor and check the logs. I had a similar problem and was a fan failure. Server restarted to avoid overheating (I think). But in our case, the shutdown was gracefull.

[bijoy k]

dear experts..

the server gets restarted only once in a day..some times twice..( very rare )..

i am sending the logs of the server
dmesg and messages..

i just want to resolve the line
srfifbtps kernel: oracle(16200): unaligned access to 0x2000000000f2ebc8, ip=0x40000000032822f0 , which is coming frequently in the server..

your responses highly appreciated.........

bijoy.

[Alan_152]

The only references I can find to this involve memory issues (i.e. http://www.unixguide.net/compaq/faq/Programming7.shtml, http://www.google.com/search?hl=en&lr=&q=%22unaligned+access%22+oracle&btnG=Search)

Since I don't read hex natively, I stick by my original suggestion of getting into the iLo and reading the system logs in text mode (sl>f>a>3>t>l). You might get lucky and have it tell you exactly what is wrong (I've had it even tell me, on occasion, which part to replace) Or, get yourself a memory testing program and run that against the box.

[Jorge Cocomess]

Is this part of a cluster setup of some kind? What else running on this server?