Apache load problem
I'm operating a small Apache custer of 8 server nodes.
Randomly with all of the nodes, the following effect keeps coming back:
For some time (varies), the backend server operates examplary, serves thousands and thousands of request on an average load of 4,5-6. (The frontend is HAProxy in roundrobin mode.) Then, for no apparent reason, load skyrockets to unbelivably high values, like 115-180 or so, then the only way to save the system from crashing is with killing all apache processes with SIGKILL, wait for 20 seconds, then issue an apache service start...
I've written a script to monitor this kind of screwup, but I'd like to know the reason behind this.
What can cause something like this?
Like I said there's no apparent pattern behind this... It's sometimes happening to the same server, sometimes another... And killing the Apache processes always helps so I figure it must be something with Apache.
Last edited by Sheridan; 02-11-2008 at 04:21 AM.