httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andre Eisenbach" <an...@eisenbach.com>
Subject [users@httpd] Apache is killing me...
Date Thu, 12 Jun 2003 22:58:39 GMT
...well, my server anyway...

I'm running Apache 2 on a RedHat Linux 7.3 server (I've been running 1.3 before with the same
results though).

The server is hosting only 5 small websites with about 50GB of traffic per month. Really nothing
special. The server load is usually about 0.50 or less and very responsive. Which confirms
that the server really is up to the task. There are usually about 15 httpd processes running.

But about once a week (sometimes every two weeks), all of a sudden something goes nuts.

The server load spikes to almost 300+ (!!!), completely grinding the server to a halt. You
can barely login at that point (remotely). The only way to get the server back under control
is stoping apache and restarting it. No need to reboot. During the spike, there are over 120+
apache processes running. The logs don't show anything bad. MRTG shows the bandwith usage
to go to almost 0 during these spikes, but that could be simply because the snmp server is
not able to respond under the system load. Here's what such a spike looks like in MRTG:
http://eisenbach.com/~andre/posted/apache_spike.jpg

I can't login locally since the server is hosted somewhere else...

Any ideas?

Cheers,
  Andre


Mime
View raw message