apache crashing frequently

[noworyz]

for the last few weeks I have been having an issue with apache crashing over 6 times per day.

It just crashed again a few minutes ago and this is the info in the apache error_log file.

[Thu Jul 05 02:17:34 2007] [notice] caught SIGTERM, shutting down
[Thu Jul 05 02:17:34 2007] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec)
[Thu Jul 05 02:17:35 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:35 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:35 2007] [warn] Init: SSL server IP/port conflict: default-204-11-244-234:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:76) vs. webmail:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:131)
[Thu Jul 05 02:17:35 2007] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
[Thu Jul 05 02:17:35 2007] [notice] Digest: generating secret for digest authentication ...
[Thu Jul 05 02:17:35 2007] [notice] Digest: done
[Thu Jul 05 02:17:35 2007] [notice] mod_python: Creating 4 session mutexes based on 10 max processes and 0 max threads.
[Thu Jul 05 02:17:35 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:35 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:35 2007] [warn] Init: SSL server IP/port conflict: default-204-11-244-234:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:76) vs. webmail:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:131)
[Thu Jul 05 02:17:35 2007] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
[Thu Jul 05 02:17:35 2007] [notice] Apache/2.2.3 (Fedora) configured -- resuming normal operations
[Thu Jul 05 02:17:44 2007] [emerg] (22)Invalid argument: mod_fcgid: can't get lock, pid: 19588
[Thu Jul 05 02:17:44 2007] [error] mod_fcgid: fcgid process manager died, restarting the server
[Thu Jul 05 02:17:44 2007] [notice] SIGHUP received. Attempting to restart
[Thu Jul 05 02:17:44 2007] [notice] Digest: generating secret for digest authentication ...
[Thu Jul 05 02:17:44 2007] [notice] Digest: done
[Thu Jul 05 02:17:44 2007] [notice] mod_python: Creating 4 session mutexes based on 10 max processes and 0 max threads.
[Thu Jul 05 02:17:44 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:44 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Thu Jul 05 02:17:44 2007] [warn] Init: SSL server IP/port conflict: default-204-11-244-234:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:76) vs. webmail:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:131)
[Thu Jul 05 02:17:44 2007] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
[Thu Jul 05 02:17:44 2007] [notice] Apache/2.2.3 (Fedora) configured -- resuming normal operations
[Thu Jul 05 02:30:04 2007] [error] server reached MaxClients setting, consider raising the MaxClients setting
[Thu Jul 05 11:11:37 2007] [notice] caught SIGTERM, shutting down


any ideas what I need to do?

Thanks

 

[flupke]

someone must be sending this SIGTERM signal to Apache. Do you have a firewall securing your network? Do have some users shell access, do you trust each of them?

 

[noworyz]

I am the only user and have 13 domains. Is there something I could be doing to send the sigterm?

 

[albans]

Maybe by doing so,

[Thu Jul 05 02:30:04 2007] [error] server reached MaxClients setting, consider raising the MaxClients setting

in your httpd.conf...

 

[noworyz]

ahh, is the httpd.conf file located at

etc/httpd/conf/httpd.conf

also, when I edit it, what do I have to do to "activate" my changes?

Thanks

if that is the correct file, maxclients is set at 10. What should I change it to?

 

[noworyz]

still havn't figured this out yet.

here is the latest crash info.


----------------------

[Tue Jul 10 09:10:53 2007] [notice] caught SIGTERM, shutting down
[Tue Jul 10 09:10:54 2007] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec)
[Tue Jul 10 09:10:54 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Tue Jul 10 09:10:54 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Tue Jul 10 09:10:54 2007] [warn] Init: SSL server IP/port conflict: default-204-11-244-234:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:76) vs. webmail:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:131)
[Tue Jul 10 09:10:54 2007] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
[Tue Jul 10 09:10:54 2007] [notice] Digest: generating secret for digest authentication ...
[Tue Jul 10 09:10:54 2007] [notice] Digest: done
[Tue Jul 10 09:10:54 2007] [notice] mod_python: Creating 4 session mutexes based on 10 max processes and 0 max threads.
[Tue Jul 10 09:10:54 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Tue Jul 10 09:10:54 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Tue Jul 10 09:10:54 2007] [warn] Init: SSL server IP/port conflict: default-204-11-244-234:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:76) vs. webmail:443 (/etc/httpd/conf.d/zz010_psa_httpd.conf:131)
[Tue Jul 10 09:10:54 2007] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
[Tue Jul 10 09:10:54 2007] [notice] Apache/2.2.3 (Fedora) configured -- resuming normal operations

 

[Server Surgeon]

Change MaxClients to be 250 -- you'll be safe at that level.

Then run:

service httpd restart

The change will take effect.

This has nothing to do with the crash though. For that, I think you should go into your server settings, look for "Apache restart interval" and see what is there. It may be that Plesk is restarting apache by itself. If it is, I'd recommend you put a 0 in that field and save.

 

[noworyz]

Originally posted by Server Surgeon
Change MaxClients to be 250 -- you'll be safe at that level.

Then run:

service httpd restart

The change will take effect.

This has nothing to do with the crash though. For that, I think you should go into your server settings, look for "Apache restart interval" and see what is there. It may be that Plesk is restarting apache by itself. If it is, I'd recommend you put a 0 in that field and save.

Thanks for the help. I set the max clients and serverlimit both to 250.

My apache restart interval was already set to 0 seconds. I'll wait for a couple more crashes to see if I get any different information.

 

[Server Surgeon]

On a second look here, your apache is not crashing at all. The "caught SIGTERM" message is coming from Plesk telling apache to restart, and is normal. This happens when you or your users make domain changes (such as adding subdomains, changing ftp login details, etc.). To test, do one of the above tasks then:

grep "caught SIGTERM" /var/log/httpd/error_log

You'll probably see apache was restarted but did not crash. Now that you've upped MaxClients you should be fine here.

 

[noworyz]

i did the command you stated and this is the result.

[Mon Jul 09 04:47:55 2007] [notice] caught SIGTERM, shutting down
[Mon Jul 09 05:08:10 2007] [notice] caught SIGTERM, shutting down
[Mon Jul 09 05:18:21 2007] [notice] caught SIGTERM, shutting down
[Mon Jul 09 06:39:03 2007] [notice] caught SIGTERM, shutting down
[Mon Jul 09 14:52:46 2007] [notice] caught SIGTERM, shutting down
[Tue Jul 10 09:10:53 2007] [notice] caught SIGTERM, shutting down
[Tue Jul 10 12:38:15 2007] [notice] caught SIGTERM, shutting down
[Tue Jul 10 22:32:21 2007] [notice] caught SIGTERM, shutting down
[Tue Jul 10 23:52:00 2007] [notice] caught SIGTERM, shutting down
[Wed Jul 11 00:08:32 2007] [notice] caught SIGTERM, shutting down


I am the only user on the VPS server and I am not making that many changes. When I get an email from my watchdog saying apache crashed, if I go to one of my sites, they are unavailable for a little bit until apache gets restarted.

It seems that I have not had as many crashes in the last few days. It was happening like 6-8 times per day. I will see if this mac clients thing fixes the issue, lets hope!

Thanks again for your help!

 

[teknovision]

Watchdog has been a pain for me and I ended up disabling it, services seem to be running fine without it.

There is some tuning I believe you have to do with regards to the timeout settings, if your server is very busy and watchdog reaches timeout it seems it just kills process - this is hypothesis and not fully investigated though.

 

[noworyz]

i have not had one crash since I changed the max clients to 250 from 10. That seemed to have fixed the issue.

If it occurs again, I will update this thread with the new info.

 

[sim0n]

apache SIGTERM and mod_fcgid "mod_fcgid: can't get lock" problem

Another interesting part nobody seems to have noticed is this error:

[Thu Jul 05 02:17:44 2007] [emerg] (22)Invalid argument: mod_fcgid: can't get lock, pid: 19588
[Thu Jul 05 02:17:44 2007] [error] mod_fcgid: fcgid process manager died, restarting the server

Strangely, some weeks ago I started having the same problems....while running fine for months, and no I haven't done any modifications.
And I'm also getting that fcgid error.

I haven't found a solution for that yet, so if somebody could help me ?
I'm running plesk 8.2, ubuntu 6-LTS.

From my logfile:

[Wed Jul 18 08:33:58 2007] [notice] caught SIGTERM, shutting down
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [notice] suEXEC mechanism enabled (wrapper: /usr/lib/apache2/suexec2)
[Wed Jul 18 08:33:59 2007] [notice] mod_python: Creating 32 session mutexes based on 400 max processes and 0 max threads.
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:33:59 2007] [notice] Apache/2.0.55 (Ubuntu) mod_python/3.1.4 Python/2.4.3 PHP/4.4.2-1build1 mod_ssl/2.0.55 OpenSSL/0.9.8a mo
d_perl/2.0.2 Perl/v5.8.7 configured -- resuming normal operations
[Wed Jul 18 08:34:02 2007] [emerg] (22)Invalid argument: mod_fcgid: can't get lock, pid: 14957
[Wed Jul 18 08:34:02 2007] [error] mod_fcgid: fcgid process manager died, restarting the server
[Wed Jul 18 08:34:02 2007] [notice] SIGHUP received. Attempting to restart
[Wed Jul 18 08:34:02 2007] [notice] mod_python: Creating 32 session mutexes based on 400 max processes and 0 max threads.
[Wed Jul 18 08:34:03 2007] [warn] RSA server certificate CommonName (CN) `plesk' does NOT match server name!?
[Wed Jul 18 08:34:03 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:34:03 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:34:03 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:34:03 2007] [warn] RSA server certificate CommonName (CN) `example.com' does NOT match server name!?
[Wed Jul 18 08:34:03 2007] [notice] Apache/2.0.55 (Ubuntu) mod_python/3.1.4 Python/2.4.3 PHP/4.4.2-1build1 mod_ssl/2.0.55 OpenSSL/0.9.8a mo
d_perl/2.0.2 Perl/v5.8.7 configured -- resuming normal operations

There are two strange things.
1. The same time apache receives the SIGTERM signal, monit (aka watchdog) reports a timeout...maybe monit "thinks" apache is down and sends a SIGTERM whereas apache is working fine ? (will investigate that by turning monit monitoring off for apache)
2. That "emerg" error "mod_fcgid: can't get lock" ... looks like fcgid is calling some functions with wrong parameters ?