Is there any way to tell what is going on when a server constantly locks or freezes after about 24-48 hrs of operation?
The server was upgraded from v9 - v11 some time ago. I was asked to look into why it is locking, I’ve asked for new memory chips, but I was wondering if there is a particular log file that can be read after the lock or freeze that may be able to shine some light on the subject. Or better still is there a recommended utility that I can run to test the hardware and memory.
Have you applied all update packages, especially the kernel? There was a serious bug in the kernel in the 11.1 release. A kernel update around March fixed that.
Here the same - a server (AMD Opteron dual core, Apache with PHP and MySQL) freezing once or twice a day without any hint in the logfiles immediatedly before freezing.
OpenSuSE 11.1 with Kernel 2.6.27.37-0.1-default x86_64
Any ideas?
Are you not able to login via ssh from another machine?
Do you have keyboard connected to this machine? If yes, are you able to see if the key indicators are working at this time - CapsLock, NumLock etc.
When the server freezes, everything freezes - no logging anymore (syslog is directed to another server, but from this moment no data is passed any more), no port is reachable, not even ping is possible.
There is no keyboard connected to this server, but a serial console - which in these moments does not show any reaction. Only a hard reset is possible.
Without load this happened only once within a week, under a very moderate load (0.1) is happens once or twice per day.
have you tried disabling APIC with nolapic (though it may also prevent the kernel seeing your other core)?
As microchip8 suggested, try disabilng APIC.
Also, see if you can check the memory.
This seems to help - Now 22 hours have passed without any problems.
Also, see if you can check the memory.
The memory has been checked several times without any negative result.
I think you may have to live with APIC disabled for some time. The motherboard combined with the driver is causing the problem. The problem is not fully identified and rectified even in the latest kernels.
Anyway, you can use the machine with APIC disabled, but it will be a bit slow (not very noticeable).