OS and Software > All Things Unix > Diagnosing ubuntu 12.04 LTS critical system errors

Diagnosing ubuntu 12.04 LTS critical system errors

What's /var/log/messages and 'dmesg' say?

reply to freakout9903
I am running 12.04 on 3 systems without any issues at all like this. In my experience things like what you describe are almost always hardware problems which can often be diagnosed from the kernel messages they generate. So as mentioned above check dmesg and post the specific error text
--
Aesthetics should be an inspiration not a pair of handcuffs

reply to freakout9903
While it sounds like hardware and no kernel pointers, heck, see how it works without any updates. Select updates carefully for awhile.

But for sure, I've seen flaky power supplies cause something like that. And I used to have a lot more trouble with motherboards than recent times - I think solder joints and capacitors are better these days.

reply to freakout9903
after going through a lot of my logs and trying to pinpoint what doesn't look normal as the errors occuring I came across this.

Jun 16 19:15:40 Godzilla-HTPC kernel: [ 497.474920] gksu[14497]: segfault at 2043588 ip 00007f12ee296e87 sp 00007fff8de70c10 error 4$
Jun 16 19:15:40 Godzilla-HTPC kernel: [ 497.550210] apport[14501]: segfault at 7f97e6bcee18 ip 000000000050d949 sp 00007ffff9a78bb0 $
Jun 16 19:15:40 Godzilla-HTPC kernel: [ 497.550232] Process 14501(apport) has RLIMIT_CORE set to 1
Jun 16 19:15:40 Godzilla-HTPC kernel: [ 497.550234] Aborting core
--
Join The Fight against Media Censorship www.blackthisout.com

reply to freakout9903
Are you seeing any Machine Check Exceptions anywhere? If so, install mcelog and see if it gives you any extra information (you must be using a 64 bit kernel to use this).

reply to freakout9903
Sounds like this bugreport

»bugs.launchpad.net/ubuntu/+sourc···g/887591

If you read the last post you will see it was due to faulty RAM
--
Aesthetics should be an inspiration not a pair of handcuffs

Well it seems the ram i thought would never go bad, has. My corsair 1gb stick seemed to be the problem...this is my first stick of corsair to go bad in 10+years. I only swapped out my Patriot ram earlier because I had RMA'd those sticks once before, a few weeks into having them they started spitting bsod's on another machine, and by the time the new sticks came I had bought different memory for the machine and stuck them into this one.

The weird thing is the sticks pass memtest, but I've been good since removing them.
Thanks for the help all.....learned my lesson.
--
Join The Fight against Media Censorship www.blackthisout.com

Glad you found it. I had this problem on an AMD opteron compute cluster node very recently. Spat out several Gigs of dmesg messages before the RAM was replaced.
--
Aesthetics should be an inspiration not a pair of handcuffs