Frequent Web Server Restarts

[VeZoZ]

According to the cPanel documentation the use of piped logging should prevent the frequent restarts that started to happen in 11.25 during stats processing. In our testing this appears to not always be the case and it is still issuing unnecessary restarts in a lot of it's stats processing:


Restarts thus far today on one machine:

Mon Oct 18 00:19:09 CDT 2010: -DSSL -t -f /usr/local/apache/conf/httpd.conf
Mon Oct 18 03:24:56 CDT 2010: -k graceful
Mon Oct 18 03:25:10 CDT 2010: -k graceful
Mon Oct 18 05:14:34 CDT 2010: -v
Mon Oct 18 07:06:54 CDT 2010: -l
Mon Oct 18 07:06:54 CDT 2010: -V
Mon Oct 18 13:01:01 CDT 2010: -k graceful
Mon Oct 18 13:01:27 CDT 2010: -k graceful
Mon Oct 18 13:01:54 CDT 2010: -k graceful
Mon Oct 18 13:02:05 CDT 2010: -v
Mon Oct 18 13:02:09 CDT 2010: -k graceful
Mon Oct 18 13:02:14 CDT 2010: -k graceful
Mon Oct 18 13:02:43 CDT 2010: -k graceful
Mon Oct 18 13:02:44 CDT 2010: -k graceful
Mon Oct 18 13:02:48 CDT 2010: -k graceful
Mon Oct 18 13:03:14 CDT 2010: -k graceful
Mon Oct 18 13:03:18 CDT 2010: -k graceful
Mon Oct 18 13:03:45 CDT 2010: -k graceful
Mon Oct 18 13:03:49 CDT 2010: -k graceful
Mon Oct 18 13:04:22 CDT 2010: -k graceful
Mon Oct 18 13:04:26 CDT 2010: -k stop -DSSL
Mon Oct 18 13:04:27 CDT 2010: -k start -DSSL
Mon Oct 18 13:04:38 CDT 2010: -k stop -DSSL
Mon Oct 18 13:04:38 CDT 2010: -k start -DSSL
Mon Oct 18 13:04:55 CDT 2010: -v
Mon Oct 18 13:05:39 CDT 2010: -k graceful
Mon Oct 18 13:06:51 CDT 2010: -k graceful
Mon Oct 18 13:06:57 CDT 2010: -k graceful
Mon Oct 18 13:07:28 CDT 2010: -k graceful
Mon Oct 18 13:07:31 CDT 2010: -k stop -DSSL
Mon Oct 18 13:07:32 CDT 2010: -k start -DSSL
Mon Oct 18 13:07:42 CDT 2010: -k stop -DSSL
Mon Oct 18 13:07:42 CDT 2010: -k start -DSSL
Mon Oct 18 13:08:36 CDT 2010: -k graceful
Mon Oct 18 13:08:38 CDT 2010: -k graceful
Mon Oct 18 13:08:43 CDT 2010: -v
Mon Oct 18 13:09:33 CDT 2010: -k graceful
Mon Oct 18 13:09:39 CDT 2010: -k graceful
Mon Oct 18 13:10:16 CDT 2010: -k graceful
Mon Oct 18 13:10:20 CDT 2010: -k graceful
Mon Oct 18 13:10:56 CDT 2010: -k graceful
Mon Oct 18 13:11:00 CDT 2010: -k graceful
Mon Oct 18 13:11:37 CDT 2010: -k graceful
Mon Oct 18 13:11:41 CDT 2010: -k graceful
Mon Oct 18 13:12:17 CDT 2010: -k graceful
Mon Oct 18 13:12:20 CDT 2010: -k graceful

It seems about the same on most machines when it's running stats processing. If you run /scripts/runweblogs username

You can clearly see with some add-on domains it clearly restarts the web server:

Log checker loaded ok..
What user's logs did you want to update? username
==> cPanel Log Daemon version 24.0
==> Loaded RRDs: version 1.3006
Processing username...
Run Logs domain: domain.com BW Limit: 62914560000 Domains: [about domain domains here]
info [safeapacherestart] Restart elapsed seconds: 2
[2010-10-18 13:35:04 -0500] Process bandwidth for zbnet
[2010-10-18 13:35:05 -0500] Processing exim stats for zbnet....
[2010-10-18 13:35:05 -0500] ...Done
[2010-10-18 13:35:05 -0500] WEBLANG english
[2010-10-18 13:35:05 -0500] Archive Status: 0
[2010-10-18 13:35:05 -0500] About to fork for zbnet [ALL DOMAINS: about domain domains here]
[2010-10-18 13:35:05 -0500] Webalizer V2.01-10 (Linux 2.6.18-194.11.4.el5) English

[2010-10-18 13:35:05 -0500] Using logfile /usr/local/apache/domlogs/domain.com.bkup (clf)

You clearly see it do this:

info [safeapacherestart] Restart elapsed seconds: 2

This machine has piped logging enabled! Why would it restart the web server when running /scripts/runweblogs when it's piped.


Any plans to address this? It's extremely frustrating having unnecessary restarts being issued on servers and we're using the suggested solution to the problem already.

 

[cPanelKenneth]

Run Logs domain: domain.com BW Limit: 62914560000 Domains: [about domain domains here]
info [safeapacherestart] Restart elapsed seconds: 2
[2010-10-18 13:35:04 -0500] Process bandwidth for zbnet

The above restart is triggered when the system fails the "using piped logging?" test. The test is comprised of two checks:

1. Does enable_piped_logs=1 exist in /var/cpanel/cpanel.config?
2. Is Apache version 2 or higher?

Does that information help you track down why this is happening on your server(s)?

If not then please open a support ticket so we can examine a system where this consistently happens.