BUG: soft lockup - CPU#0 stuck for 22s! after recent update of Fedora 20

[Max]

On 03/15/2014 09:45 AM, The Walters wrote:
>
> On 3/15/2014 2:12 AM, D. Hugh Redelmeier wrote:
>> | From: The Walters <sbwalters0731 at gmail.com>
>>
>> | After a recent update started seeing messages like what is listed 
>> below and
>> | now complete lock up. Tried going back 1 or 2 kernels but same issue.
>>
>> I can't help you with the specific problem.
>>
>> It is always good to google an error message.  Have you?  I'm not going
>> to try -- I'll leave that to you.
> I have googled it and have tried the suggestions but none of them have 
> worked. I also searched the list back quite a few months and nothing 
> showed up.
>>
>> It would be a good idea to mention the actual kernel versions you've
>> tried.
>>
>> Red Hat's bugzilla is a useful place to report errors.  I always like
>> to be a little confident before I report them.  So asking on the list
>> may be a good idea.
> I have tried the last 2 kernels and the 1st kernel that that was with 
> the original release of Fedora 20 (I know this was a long shot) but no 
> go. I also have the last Fedora 19 kernel install from before I did an 
> upgrade to 20 and that starts okay with no issues but of course RPC 
> Bind and the IPv4 firewall won't run with this after the upgraded (the 
> system is an i386 system).
> Last 2
> 3.13.6-200
> 3.13.5-200
> Old/1st
> 3.11.10-301
> Fedora 19 Kernel that works with no errors but can not use because of 
> issues mentioned above
> 3.9.9-302
>
> I agree with you about asking on the list first before posting a 
> bugzilla report. In fact I check a couple of bugzilla reports that 
> were close to my issue but they were either closed due to no followup 
> information or just hanging open.
>>
>> | Wondering if this is a hardware problem?
>>
>> Why?  Because old kernels show the same problem yet you only noticed
>> it recently?  That's a good thought.  But Fedora kernels come out
>> frequently -- check the age of the oldest one you've tried.
>>
>> Whenever I wonder about a hardware error, with non-specific symptoms,
>> I start memtest86 going overnight.  That makes me more confident about
>> the RAM at least.
>>
>> Consider trying a really old kernel as another test.
>>
>> Good luck!
>>
> Just wondering because it is an old system but it has been rock solid. 
> No issues with the ram as I upgraded it not too long ago but running a 
> memtest can't hurt.
>
> Thanks for the suggestions.
You could try booting to the F19 kernel and running a stress test on the 
CPU then do the same with the F20 kernels and see what happens.

Is this server dedicated to some specific task? Like a webserver? You 
might try to check service logs, if available, for errors and see if any 
correlate with the times reported for this problem.

-- 
maximilianobianco.deviantart.com/gallery