Debugging a system freeze?

[Jeff Gipson]

On Sun, Jun 17, 2012 at 02:10:35PM +0200, Christopher Svanefalk wrote:
>On Sun, Jun 17, 2012 at 2:04 PM, Jeff Gipson <jeffagipson at gmail.com> wrote:
>
>> On Sun, Jun 17, 2012 at 01:44:47PM +0200, Christopher Svanefalk wrote:
>>
>>> Problem has been solved for the record - followed a recommendation and
>>> notched vcore up to 1.280, misread one of the original graphs.
>>>
>>> --
>>> Best,
>>>
>>> Christopher Svanefalk
>>>
>>
>>
>>        ### End of Message from Christopher Svanefalk ###
>>
>> To address your (now moot) request, the Kernel ring buffer is probably the
>> first place I would go, if I suspected a hardware or hardware driver issue.
>> You can see the current contents of the ring buffer by running the 'dmesg'
>> command, or you can see more history is /var/log/messages for general logs.
>>
>> I should note from experience, that often when there's an obsure or
>> intermittant hardware problem sometimes the problem manifests itself
>> different ways each time, or side effects of the problem can misinform the
>> troubleshooting process. In an environment where you have >1 identical
>> pieces of hardware with the same (Exact Same) Kernel version, Kernel
>> modules, and drivers, If one system had this problem and the other didn't,
>> then I would suspect either the h/w or environment first. If I could
>> reproduce the problem on both systems, then I would suspect s/w, but not
>> rule out the possibility of environment or h/w design flaw.
>>
>> HTH
>> --
>> ×-×-×-×-×-×-×-×-×-×-×   Rule of Economy: Programmer time is expensive;
>>  Jeffrey A. Gipson     conserve it in preference to machine time.
>> ×-×-×-×-×-×-×-×-×-×-×   -- Eric S. Raymond, The Art of Unix Programming
>>
>> --
>> users mailing list
>> users at lists.fedoraproject.org
>> To unsubscribe or change subscription options:
>> https://admin.fedoraproject.org/mailman/listinfo/users
>> Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
>> Have a question? Ask away: http://ask.fedoraproject.org
>>
>>
>Thanks for the replies Heinz and Jeff (though the issue is not entirely
>moot just yet Jeff, presumptuous wording on my part), I will run those
>through the next time it happens.
>
>-- 
>Best,
>
>Christopher Svanefalk

	### End of Message from Christopher Svanefalk ###

And this is a huge PITA, but if your system becomes to unusable to use 
dmesg, *and* you have a another computer (preferably laptop) then you 
can either use a null-modem cable (with a terminal emulator on the 
laptop, such as 'screen' or 'minicom') or a network cable (with syslogd 
on the laptop configure to accept logs from your ailing system over the 
network). What you would be going for, is sending all the kernel logs to 
the serial port (if using a null-modem) or remote laptop (if using 
syslog/networking). That way if your system crashes hard, hopefully 
enough logs will have escaped to the connected system to provide some 
hints.

Unfortunately, I cannot give the specifics on how to configure this 
either way... I could tell you if it were an RHEL5 system, but so much 
of the logging has changed on recent Fedoras and I haven't had time to 
examine how it is working.

-- 
×-×-×-×-×-×-×-×-×-×-× 	Rule of Composition:
   Jeffrey A. Gipson   	Design programs to be connected to other programs.
×-×-×-×-×-×-×-×-×-×-× 	-- Eric S. Raymond, The Art of Unix Programming
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 4147 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/users/attachments/20120617/43661497/attachment.bin>