Apache up & down constantly

[alphawolf]

I'm at wits end here.

On June 8th we upgraded to the latest apache version on all servers. One server failed miserably after the upgrade, but after apache was rebuilt, all seemed well. SEEMED is the operative word.

Almost immediately two things happened

1) httpd would fail anywhere from 2-10 times daily. Most times it would restart itself, other times it would need to be manually restarted.

2) ftp socket response would time out for long enough that at least 4-5 times a day SiteUptime reported FTP down.


Now, server is a dual xeon with 2 GB RAM & the server load, except when running stats or backups, rarely even gets to 2.0. There have been no slowdowns to speak of ever.

The data center did complete hardware diagnostics and everything checked out fine. Their only other two suggestions, remove APF and mod_python, didn't help a bit.

If anyone has any ideas, please let me know. The log files do not show anything consistent that we can 'latch on to' as the culprit.

peace

Wolf

 

[AndyReed]

One server failed miserably after the upgrade, but after apache was rebuilt, all seemed well. SEEMED is the operative word.

Almost immediately two things happened

1) httpd would fail anywhere from 2-10 times daily. Most times it would restart itself, other times it would need to be manually restarted.

2) ftp socket response would time out for long enough that at least 4-5 times a day SiteUptime reported FTP down.

Did you check the syntax of your httpd.conf? Did you also check on all Apache modules and make sure they are working? When recompiling Apache, did you watch for any errors? Check on the size of the log file of your accounts. You also need to check out mod_rewrite and see if the log file grews too large in size.

 

[LiNUxG0d]

Hey there,

Well, for one, let's diagnose the problem shall we?

For starters, try tailing the Apache logs. When something crashes or segfaults, usually you'll find errors there:

tail -f /usr/local/apache/logs/error_log

If this shows nothing, check if you're running PRM and insure that httpd is in the ignore list. You can add it there. Sometimes the load might spike and PRM will go "DIE HTTPD!" or something similar... I mean it won't actually TALK but hey, you know what I mean.

Additionally, try rebuilding Apache using /scripts/easyapache and remember to check - which is usually default - the large filesize limits (from memory FD_SETSIZE)?

The fact that your admin would even dismiss you with mod_python/APF sounds like a brush off to me. SO frustrating.

What OS are you running on that puppy? The box is relatively monstous so I don't think load would be a cause for Apache crashing. I have P4's running loads of 50+ for the last... wow, has it been 7 months already? Hehehehe.

If a P4 can keep Apache up for months without crashing at a load of 50, then man, a Dual-XE should be able to for sure!

Hopefully my information will be helpful, I'll leave you with the following:

service httpd configtest

Try that and let me know if things are ok. Don't forget to run easyapache because it will rule out miscompiled modules in the httpd.conf.

Take care!

Jamie