httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Lukas Kolbe <lu...@knup.de>
Subject RE: [users@httpd] Apache 1.3.26/Debian 3.0 Process spawn runaway
Date Fri, 31 Jan 2003 14:33:31 GMT
Am Fre, 2003-01-31 um 15.16 schrieb Boyle Owen:
> - What is the load average when it locks up? (use top to find out).
> - Do you have a lot of CGI operations, DB operations etc.? Anything
> which is CPU hungry.
> - From the logs, what is your hit-rate? Output data rate? 
> - What is your bandwidth to the Web? (If you are serving data faster
> than your pipe can drain it...)

It looks like we have about the same problem here. Box is a P4 1.6Ghz,
1GB RAM, Linux 2.4.18, Debian 3.0.

After a while, apache needs more and more RAM, begins to suck up swap,
load gets up to 200 and more. 

We're using eZPublish (ez.no), so the site very massivly uses MySQL and
PHP. 

Uplink to the World is 100mbit ethernet, and as far I can tell, to some
servers we have about 1Mbyte/Second.

When apache eats RAM, the hits are the same as the rest of the time. No
special spikes.

> >
> >The following other config parameters may be of interest:
> >Timeout 300
> >MinSpareServers 15
> >MaxSpareServers 35
> >StartServers 30
> >MaxClients 300
> >MaxRequestsPerChild 50
> >KeepAlive On
> >MaxKeepAliveRequests 100
> >KeepAliveTimeout 15

We have Keepalive turned off, because there have always been apache's
that locked up because of that.

-- 
Lukas Kolbe <lucky@knup.de>

Mime
View raw message