[Systems] Apache is hogging the CPU on solarsail

[Bernie Innocenti]

On 06/25/09 11:59, Bernie Innocenti wrote:
> On 06/25/09 11:53, Bernie Innocenti wrote:
>> On 06/25/09 11:44, Bernie Innocenti wrote:
>>>  5686 www-data  20   0  313m  50m  22m R  103  0.2   0:56.39 apache2
>>>  9877 www-data  20   0  313m  49m  21m R  103  0.2   0:15.48 apache2
>>>  2009 www-data  20   0  313m  53m  25m R  100  0.2   1:22.23 apache2
>>>  2442 www-data  20   0  314m  72m  43m R   96  0.2   1:28.54 apache2
>>>  5472 www-data  20   0  314m  53m  23m R   96  0.2   0:31.36 apache2
>>>  5675 www-data  20   0  313m  58m  30m R   96  0.2   0:49.86 apache2
>>>  5681 www-data  20   0  308m  44m  22m R   96  0.1   0:01.06 apache2
>>>  9845 www-data  20   0  314m  50m  21m R   96  0.2   0:23.26 apache2
>>>  9876 www-data  20   0  312m  49m  21m R   96  0.2   0:15.46 apache2
>>> 10271 www-data  20   0  308m  45m  21m R   96  0.1   0:01.10 apache2
>>>
>>> I'm investigating
>>
>> This is no bug: we're really being hit by several connections per
>> second.  The load comes from the wiki.
>>
>> One of the main referrers is:
>>   http://news.bbc.co.uk/2/hi/technology/8117064.stm
> 
> What could we do to speed up the wiki a bit more?  mod_cache?
> memcached?  We're already using APC, right?

I enabled Memcached for the main wiki...  Responsiveness seem
to have slightly improved.

However, I've noticed lots of segfaults in the apache error log (also before installing memcached):

[Thu Jun 25 06:30:53 2009] [notice] child pid 11280 exit signal Bus error (10)
[Thu Jun 25 06:31:44 2009] [notice] child pid 27352 exit signal Segmentation fault (11)
zend_mm_heap corrupted
[Thu Jun 25 06:34:40 2009] [notice] child pid 27327 exit signal Bus error (10)
[Thu Jun 25 06:36:12 2009] [notice] child pid 5972 exit signal Illegal instruction (4)
[Thu Jun 25 06:36:14 2009] [notice] child pid 29106 exit signal Segmentation fault (11)
[Thu Jun 25 06:39:31 2009] [notice] child pid 29114 exit signal Illegal instruction (4)
[Thu Jun 25 06:40:11 2009] [notice] child pid 17198 exit signal Segmentation fault (11)
[Thu Jun 25 06:41:03 2009] [notice] child pid 29786 exit signal Segmentation fault (11)
[Thu Jun 25 06:43:51 2009] [notice] child pid 23083 exit signal Segmentation fault (11)
[Thu Jun 25 06:45:19 2009] [notice] child pid 23225 exit signal Segmentation fault (11)

Given the diversity of these errors, I have quite a scary theory: CPU overheating?

-- 
   // Bernie Innocenti - http://codewiz.org/
 \X/  Sugar Labs       - http://sugarlabs.org/