Apache segfaults + high load after running a week
Dear Maintainers,
several times now, we experienced strange behaviour with the apache. For
weeks everything works fine. Then, suddenly the load raises exessively
like above 100 and stays there. It is hard to log in and to do a "ps" or a
"kill". There are a more than 50 apache and mysql processes running, many
of the apache processes are in dead state.
Within the apache-log I could not find that the web requests raised. So it
does not look like a denial of service attack. We have just one
VirtualHost running Typo3. During the high load period there are about as
many apache processes running as mysql processes.
We have installed debian testing:
Package: apache
Status: install ok installed
[...]
Architecture: i386
Version: 1.3.31-3
[...]
in our httpd.conf we have set:
MinSpareServers 5
MaxSpareServers 10
StartServers 5
MaxRequestsPerChild 100
The latter one we tuned down to 30.
Kernel is 2.4.22
/var/log/apache/error.log:
[Tue Oct 19 11:09:54 2004] [error] server reached MaxClients setting, consider raising the MaxClients setting
[Tue Oct 19 11:15:21 2004] [notice] child pid 23533 exit signal Segmentation fault (11)
Segmentation fault occurs about 20 times in a period of 20 mins.
Can it be that this load problem is caused by the apache? If yes, is it a
known problem? We now update from testing to unstable and hope that the
problems are already fixed in unstable.
Thanks for your assistance and your work!
Marc Daniel Haunschild
--
-----------------------------------------------------------------------
cid - christliche internet dienst GmbH
Fritz-Erler-Allee 79, 12351 Berlin
Tel: (+49)(0)30 - 4 61 61 46
Fax: (+49)(0)30 - 46 50 79 20 http://www.cid.net/
Reply to: