Topic: Kernel Panic - not syncing: Fatal Execption in interrupt (Screen Shot)

[steve288]

Periodically we have been getting this error and the system crashes. eg stops working, with the message on the screen in my screen capture Im providing. This occures about once a week. Last time I reported this I was asked to provide a screen shot. I am now providing this.

Can anyone provide any help on this and how to stop my system from crashing. Im using SME 7.5.1.
Other than this the system works fine.
The picture is not perfect but it was shot with a cell phone. However I think it give a good shot of the screen.
This is a link.

http://www.picamatic.com/view/8067075_sme_server_crash/

this is the same picture above via inserted into Newsgroup. (if it works)


Thank you.
 

[Stefano]

99.999999% an hw problem

you should test the ram with memtest; if everything is ok with it, mobo is the next candidate..

[steve288]

I will try memtest today and see what the results are.
Although I never know how good these tests are eg. If it only happens once a week then, presumably the problem is erratic and I never know if these tests are very thorough? Anyone have any comments on this ??
Regards

[Stefano]

I had similar problem some years ago: one server started to crash (kernel panic) randomly.. one or twice a day or once a week.. or a month..

the only way to discover the problem was to run memtest for 4 hours, just to discover that one memory bank was broken..

linux uses ram in a (for us) "strange" way.. and errors occur only when you use the corrupt address..

[MSmith]

Concur that it's more than likely a hardware issue. I'd also check the motherboard for bulging or leaking capacitors and run any motherboard and/or hard drive diagnostics available. I've been using SME for quite a number of years now on varied hardware and have *never*, repeat *never*, seen a problem with a default install (i.e. no contribs) that wasn't hardware related.

Edit:  don't know if I'd be certain to "eight nines" however

[CharlieBrady]

I suspect that this is a kernel bug (fixed in later kernels) in the appletalk kernel module. I suggest that you disable appletalk, and see if the problem goes away:

http://bugs.contribs.org/show_bug.cgi?id=3500#c7

[steve288]

Two things.
Im not sure how to run memtest, do I have to use an install cd and boot from it.
I tried rebooting the sytem but could not see a place to change to this test. Im thinking its on the install cd. Could someone clarify. I have read the SME manual but I dont find it spelled out in my read if it is on cd, although perhaps its infered.  However , I have booted using one of the ubcd - CD's and it has a memtest. I asume this is the same it is version 4.20. Is this going to do the same thing. Also how long should I run it. I mean if this happens only a once a week doesnt that mean that I man need to run it for a week before it sees anything. Again as I think I mentioned before Im not sure how these tests work. Of couse it could be that the mem test, tests every area of the memory that normally would not be used but a few times a week and that may crash it earlier.  I dono?

Thanks for the apple talk advice. I was thinking same sort of thing. I mean it is giving an error some software, I just didnt know how how to read the error. Thanks CharlieBrady I will try that also.

Regards

[Frank VB]

Im not sure how to run memtest, do I have to use an install cd and boot from it.

Insert the SME 7.5.1 installer CD in the CD-ROM drive. Reboot your server from the CD. At the command prompt type

Code: [Select]

memtest

As posted earlier by Stefano run memtest for a couple of hours. He ran it for 4 hours before an error occurred. Make at least 1 complete pass of your memory (indicated in the next to last column "Pass" of memtest).

[steve288]

Thank you Frank VB for your clear answer on using memtest.

I'm going to for the moment try the option provided by link CharilieBrady. My screenshot and the screen shot from the bug tracker link that CharlieBrady provided above looks suspiciously similar in Almost every way. So Im going to run ....

config setprop atalk status disabled
signal-event reboot

Because this server is being used all the time its hard to do the test even for 4 hours. I'm going to see if the disabling of the atalk fixes it. If it does not then I'm gong to have to do memtest at night or on weekends when it is being used less.
I will get back to the group with my results, which may help others and provide feedback. This could take a few weeks. As I now need to wait and see if it crashes and if so then find time to do memtest.
Thanks for help.

[byte]

I suspect that this is a kernel bug (fixed in later kernels) in the appletalk kernel module. I suggest that you disable appletalk, and see if the problem goes away:

http://bugs.contribs.org/show_bug.cgi?id=3500#c7

I can pretty much say this is/was a kernel bug as once I disabled appletalk I have not since I report that screen shot say I've encountered this issue.

[steve288]

Just wanted to get back to the group.
I believe the problem is resolved. It has been two weeks since I disabled Netatalk and the problem has not been back. So thanks for the sound advice CharlieBrady. This seems to have resolved the crashes.

Regards.