httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michel Jouvin <jou...@lal.in2p3.fr>
Subject Re: [users@httpd] 2.0.48/49 : httpd children multiplying and eating 100% of CPU
Date Tue, 27 Apr 2004 21:21:56 GMT
Hi,

I got a new occurrence of the problem this afternoon. Attached is the 
output of the script checking regularly the number of httpd processes and 
taking care of the problem in case some httpd children look suspect. This 
include a thread traceback with dbx for each suspect process.

I also attach the mod-status output on this server at the time of the 
problem (in fact 2 hours after the run of the script as a bug in the script 
prevented it to fix the problem).

Thanks in advance for any advice.

Michel

--On lundi 26 avril 2004 15:17 +0200 Michel Jouvin <jouvin@lal.in2p3.fr> 
wrote:

> Hi Joshua,
>
> I have not been able to send you the information requested still. The
> problem has not occurred for 3 days...
>
> In fact, the only change I made just before posting the problem to the
> list was to review worker MPM settings and lower the 'StartServers' to 1
> as there is no need to start more at Apache startup (else other children
> are started and immediately killed). I don't think it explains the
> disparition of the problem. May be I just need to wait more...
>
> I attached my httpd.conf file and I'll let you know as soon as the
> problem occurs again, with the material requested.
>
> Thanks for your help.
>
> Michel
>
> --On vendredi 23 avril 2004 12:57 -0400 Joshua Slive <joshua@slive.ca>
> wrote:
>
>>
>> On Fri, 23 Apr 2004, Michel Jouvin wrote:
>>> From time to time (between 1 time/day and 1 time every 3 days, not at
>>> the same hour or interval), we see the number of httpd children process
>>> increasing to 20 or more in one hour. All these children except one
>>> compete to eat 100% of the idle CPU and have in common a strange
>>> characteristic : parent of these processes is not the master httpd but
>>> the only slave which is not consuming the CPU and has the master httpd
>>> as parent.
>>>
>>> These processes seems to be no longer under the control of Apache. First
>>> they are much more than allowed by ServerLimit (16). And they are not
>>> killed (even there is no message they couldn't be killed in main
>>> error_log) if stopping Apache. You need to kill -9 them manually.
>>
>> What modules are you using?
>>
>> Can you attch to the spinning processes with strace and/or a debugger and
>> see what they are up to?
>>
>> Joshua.
>>
>> ---------------------------------------------------------------------
>> The official User-To-User support forum of the Apache HTTP Server
>> Project. See <URL:http://httpd.apache.org/userslist.html> for more info.
>> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
>>    "   from the digest: users-digest-unsubscribe@httpd.apache.org
>> For additional commands, e-mail: users-help@httpd.apache.org
>>
>
>
>
>      *************************************************************
>      * Michel Jouvin                 Email : jouvin@lal.in2p3.fr *
>      * LAL / CNRS                    Tel : +33 1 64468932        *
>      * B.P. 34                       Fax : +33 1 69079404        *
>      * 91898 Orsay Cedex                                         *
>      * France                                                    *
>      *************************************************************
>



     *************************************************************
     * Michel Jouvin                 Email : jouvin@lal.in2p3.fr *
     * LAL / CNRS                    Tel : +33 1 64468932        *
     * B.P. 34                       Fax : +33 1 69079404        *
     * 91898 Orsay Cedex                                         *
     * France                                                    *
     *************************************************************


Mime
View raw message