httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Arnab Ganguly" <>
Subject Re: [users@httpd] too many open files
Date Wed, 27 Aug 2008 09:11:11 GMT
Hi Michael,
Did you tried the following "ulimit -S -n `ulimit -H -n` ?
This will assign the max open file hard limits value to your soft limit.
Can you check the output of ulimit -H -n and ulimit -S -n?Former gives your
hard and the latter gives your soft limits respectively.
Even if the above step is giving problem then increase the hard limit to max
value as per ur system should be something like 999999.So put this value to
ur soft limit by the above procedure in the apachectl script and try out.You
print the value of ulimit -S -n in you apachectl script just to make sure
the change is taking place.Make sure the shell which runs the apachectl gets
the incremented FDS.

On Wed, Aug 27, 2008 at 12:40 AM, Michael Sutter

>  Hello Arnab,
> sorry for answering so late, but it took some time to test it.
> At the beginning of the apachectl script I put
> ulimit -n 50000
> and restarted the daemon with the apachectl script.
> Nevertheless, it don't changed the behavior. The service
> was running for about  7 - 8 hours and after that I error
> message.
> Maybe you have another hint for me?
> Kind regards
> Michael
> Arnab Ganguly wrote:
> Can you try assigning the soft limit value to hard limit value and restart
> the server.You can put this in apachectl script so that it gets affected for
> the shell used for Apache.
> Thanks
> Arnab
> On Mon, Aug 25, 2008 at 6:47 PM, Michael Sutter <
> > wrote:
>> Hello list,
>> I have a strange problem with my httpd daemon and hopefully somebody can
>> help me.
>> I'm running a Apache 2.0.49 on a Suse 9.1 and have the mod_axis2 deployed.
>> Inside
>> Axis2 I'm running a service which is queried every ten seconds.
>> After running for some hours, sometimes 2, sometimes 4, sometimes more I
>> always got
>> a exception: Too many open files. The exception is not written to the
>> error log, it is the
>> return value of my service. I also have no entry at the corresponding time
>> in my access.log,
>> so I think it is thrown before the service is accessed.
>> I searched through the list and found, that normally the solution is  to
>> increase the limit
>> of open files. So I added in /etc/security/limits.conf
>> *  soft    nofile  8192
>> *  hard    nofile  50000
>> logged out and in again.
>> For my understanding this should increase the number of open files for
>> every user.
>> Nevertheless, this don't changed the behaviour. I always got the
>> exception. So I also
>> added ulimit -n 8192 to my init script, which shows the same behaviour.
>> I also monitored the number of open files on the system. It is always
>> about 2000 - much
>> less then I have declared in the configuration. The httpd daemon normally
>> has 10 process
>> and every process has opened about 90 - 95 files. So I'm also not on the
>> configured limit.
>> Has anybody some idea what I'm doing wrong or how I can solve the problem?
>> Kind regards
>> Michael
>> ---------------------------------------------------------------------
>> The official User-To-User support forum of the Apache HTTP Server Project.
>> See <URL:> for more info.
>> To unsubscribe, e-mail:
>>  "   from the digest:
>> For additional commands, e-mail:

View raw message