Hey Guys, I have a VPS (with Slicehost), and it has been going down every few days in the middle of the night. It looks like apache is overloading the box and the kernal is trying to kill it. Any ideas as to how to track it down? In apache logs, the only thing that happens before then is a bunch of "internal dummy connections" about 23 in 25 seconds and a few minutes later a number of "Unable to fork new process". Thanks all in advanced! Justin
Hello. Around midnight could start a cron job to update statistical data (webalizer, awstats) or a data backup. Give more info for debug (operating sistem, kernel version, apache version). Also could be a bug in Apache (search for "apache2 invoked oom-killer" on Google), depending on apache version. Have a nice day.
Hey there, oom-killer normally means openVZ is being used inside of a XEN? right? If so, go into the VPS and do cat /proc/user_beancounters and see if it's running out of anything. If it is, you can increase it with vzctl set VZID Best of luck
wow where are you guys located (or why are you up?), 3:30AM and 7:30 AM PST, I'm still sleeping. @Dexter - OS: Ubuntu 9.04 Jaunty Kernal Release: 2.6.24-23-xen Kernal Version: #1 SMP Mon Jan 26 03:09:12 UTC 2009 Apache Server version: Apache/2.2.11 (Ubuntu) Apache Server built: Jul 10 2009 18:37:31 @Frantech - I tried running your command. Apparently, I don't have /user_beancounters. What's a good monitoring solution? I installed Cacti yesterday, but I don't think that's what I'm really looking for. The mod_status seems like it's in the right direction, but I don't know if it's detailed enough nor know how to log the information that it provides. Thanks again for all your help.
Me ? Romania In here is evening ... 2.6.24-23-xen ... probably you are running a VPS. Try: cat /proc/user_beancounters Code (markup):