[BlueOnyx:09661] Re: Logwatch Kernel Errors

[Michael Stauber]

Hi Matt,

> I'm not sure how long this error may have been happening.
> 
> WARNING:  Kernel Errors Present
>     [<c011f140>] ? mm_fault_error+0xe0/0xe0 ...:  16 Time(s)
>     [<c06903c6>] ? error_code+0x5a/0x60 ...:  8 Time(s)

This is a generic memory management error. It can happen when the server runs 
out of memory. I've also seen it happen when the memory is defective. All in 
all that error message by itself is not a software issue or OS related. It 
boils down to hardware issues (defects) or usage issues (too much load).

So how is the server load and how much memory is used when the error happens? 
How does the process list look at the time of the error?

Another option would be to boot off the CD and to run a memory check to see if 
that uncovers any problems. However, depending on the amount of RAM this can 
take quite a while to finish.

If the server is taxed to the limit when this problem happens, then adding RAM 
or reducing the load might help. If the problem happens during minimal or 
normal load, then the source of the problem is probably defective memory, or a 
problem with the motherboard.

I have two identical servers and one of them exhibited a similar problem, 
while the other one was running fine. Short of swapping the RAM out, I reduced 
the speed of the frontside bus from 1333MHz to the next lower stepping 
(1066MHz or around that figure). That kind of solved the issues for now, 
although the box is now running a bit slower.

-- 
With best regards

Michael Stauber