Software Channels

Ok everybody, this gets technical and it needs to be. This blog posting is for a specific problem with Apache 2.0 I have not found any solution for online:

Yesterday night, one of our webservers died again. While everything SEEMED to be working, it was not. The node did not deliver any pages and requests timed out. I did an immediate restart, because the monitoring software was ringing the bells. After 30 minutes, the node died again. Ok, it had a solid load with 3-5 requests per second, but not really a HUUUUGE one. I found that line in the error log:

child process XYZXYZ still did not exit, sending a SIGTERM

Where XYZ is a random number from the system. I looked into the Apache bug reports and searched my eyes out in Google to find people with the same problems. I finally got one guy posting a simliar problem and I am testing his suggestion right now:

The configuration entry for “KeepAlive” was switched off. That worked well for over 2 years and I did not suspect that to be a problem, but he thinks it is! I changed it to “KeepAlive on” and lowered the values (like TimeOut to 3 seconds and MaxRequests to 50) and since then it is stable again. I have no idea, if that was the reason for the blackout, but we will see. Until now, no further outages occured and I cross my fingers.

In hope this posting helped someone… have a good day!

This entry was posted on Tuesday, May 15th, 2007 at 6:17 pm and is filed under Software. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.