Out Of Memory problems. My server is down. Memory logs attached.
Hi.
Using Fedora Core 2 '2.6.9-1.667', I'm suffering memory problems. This night I had to restart this server at 22:45h because most of daemons were killed. The problem started at 21:03h. From my '/var/log/message': Code:
Oct 16 21:01:16 www crond(pam_unix)[995]: session closed for user root I'm monitorizing memory uses each 10 minutes, and these are the outputs at 21:00h. Code:
[]# uptime |
You seam to have plenty of swap space free... The thing that concerns me is the load of 22.53. Can you run top while this is happening and tell us what process(es) are hogging all that CPU time?
|
Hi.
Thank you very much, jtshaw, for your answer. I didn't notice the figures you point. It's a very high load, but the unique log I've got it's the 'ps aux' output I reported :( Isn't it the same information as 'top'? Regards. |
just looking at that log and specifically looking at the mysql and httpd makes me suspicious that perhaps you have a php script querying the database that is not closing. I have experienced this myself and what it does it leaves the query open. You have 38% use by mysql. Have you added any new query scripts recently?
|
Hi TigerOC. Thank you very much for your answer.
I'm not very sure about how to interpret the memory use. If I add up every 'mysql' processes at the time of the incident, I get 36-38%, and if I add up every proccess, I get 149%. So I consider it's the reason of my problem, right? But If I add up every proccess just now, I get 228%, but my server runs ok right now. :? Could anybody explain it to me? |
This article may help you pinpoint the problem. I see from the log that you are running the memmon script so have you checked the memmon.txt file?
|
Hi TigerOC.
I'm actually checking that document to find out any explanation. Thank you very much!!! |
If I add up every 'mysql' processes at the time of the incident, I get 36-38%, and if I add up every proccess, I
get 149%. So I consider it's the reason of my problem, right? But If I add up every proccess just now, I get 228%, but my server runs ok right now. :? This could be because, individual process memory stats show the shared libraries also that are loaded. Where as in the total memory usage memory used by shared libraries is counted only once.... You can probably check how much memory exactly a process is using with pmap <pid> command. (my guessing) Thanks, Sudheendra |
All times are GMT -5. The time now is 02:01 AM. |