I have had my file/networking/virtual machine server crash twice this week so I added a graphics card back so I can see whats going on as ssh stops.

It then crashed again and I got another error. I don't really understand what is going on about or how to search my server to see what happened. When I look though the /var/log/syslog I don't see anything unusual.

Seemed to work it work though all the cores taking about running for (61s) each then displaced the call trace. I would love to get the error up again but no idea when it will happen again.

It Does seem to mention:

Comm:  <Service Name> Tainted. 

So my question is, how do I find what cased this error?

Unexpected Server Crash

  • 1
    To me looks like an error in the filesystem (syscall fastpath, RIP in spin_lock), but I suspect an error in the system's RAM. Can you run memtest for some hours? – ott-- Oct 1 '13 at 21:09
  • This is not the first oops of the session, and will probably be useless. Check the system logs for the fault which preceded it. The correct one will be marked as Untainted. – Michael Hampton Oct 1 '13 at 21:10
  • I will run memtest overnight and see if it sparks up any errors – Simon Bennett Oct 1 '13 at 22:26
  • I agree with Michael, the D flag to Tainted means that there was a previous problem with the kernel that should be logged, I would guess in /var/log/kern.log (I don't exactly have an oops to look at here) unless it really is a filesystem problem in which case it probably couldn't be written anywhere. – DerfK Oct 1 '13 at 22:37
  • I have looked though the kern.log nothing in there that seems to be unusual. I did run the memtest which failed. I then loaded safe defaults on the bios and ran again and it passed fine. I think I clocked it when I was running a lot of virtual machines. – Simon Bennett Oct 2 '13 at 8:50

That image looks like the tail end of a kernel panic. That combined with the "Tainted" service suggest to me a poorly functioning third-party kernel driver for that service. That's about all I can say with the information you've posted, but it might get you started on isolating the issue.

  • I don't know much about kernel's. Would there be any way to see a log of the event or add a log for the next time? – Simon Bennett Oct 1 '13 at 19:57
  • On RHEL, I would set up kdump and/or netdump to get the full dump and log. On Debian, I don't know how to do so, so all I can tell you is "Yes, there is a way." - I just can't tell you what that way is. – John Oct 1 '13 at 20:00
  • You could also redirect the console output to a serial port. Link the serial port to another computer and redirect the output to a file or something via screen. – Zoredache Oct 1 '13 at 20:38
  • When I ran a Apt-Get update/upgrade it updated the kernel for vboxhost. I hope this might solve things but will try @John idea thanks :) – Simon Bennett Oct 1 '13 at 21:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.