[11.25 Current] apache restarting every hour

[cPanelDon]

I am glad a fix is on the way, but I should not have to update to an edge build to solve a bug. In the event that a bug makes it down to the release branch, shouldnt the fix then also be applied to that next release version? I suspect I will just introduce more issues by going with an edge build.

I apologize as I was not meaning to suggest EDGE as the only solution.

To help clarify, the message was intended to help keep everyone on this thread informed that additional refinements and corrections have been developed, extensively tested in BETA, then later pushed to EDGE and tested further, and subsequently will be available in CURRENT when they are ready for that build tree. The same applies to RELEASE and STABLE; new fixes and enhancements make their way into each applicable build tree when they are ready for that build tree.

[Snowman30]

We have been recommended to bring out 11.25 CURRENT installs up to Edge in a ticket we have open but this unfortunately hasnt solved the issues for us

PHP processes appear to use up all the servers memory and are not clearing out once the process has completed, so apache loads up the server to massively high loads and all php based sites return a 500 error

An apache restart corrects the issue but given each of our servers require this 3 or 4 times a day at random intervals is a right pain

and we would prefer not to downgrade back to 11.24

The apache build we use is complex but it has worked well ever since ea3 came out

so far no one has been able to identify the cause of the issue, but we are looking closely at suhosin and suPHP ourselves as the cause of the issue

[cPanelDon]

We have been recommended to bring out 11.25 CURRENT installs up to Edge in a ticket we have open but this unfortunately hasnt solved the issues for us ...

Please PM me the support request ID (i.e., ticket number).

[sebby]

We are testing a testing a redesign of the log processing mechanism that will significantly reduce the number of restarts. This will be in place before publishing cPanel 11.25 RELEASE.

As WHM-11.25 is moving into the RELEASE tree (R-42213) I can see that this issue was addressed:

Adjusted message displayed when Apache restarts are disabled by EasyApache
Fixed warnings related to Apache restarts
Queue Apache restarts in TweakSettings
Reduce the number of apache restarts caused by cpanellogd
Reduced Apache restarts triggered by log processing to no more than 3
Updated /scripts/restartsrv_httpd to observe the apache_update_no_restart setting

Are those issues all related to this bug? What does the "Reduced Apache restarts triggered by log processing to no more than 3 " mean?

Can someone confirm that this issue was definitely fixed? What was causing apache to restart so often (how/why was the log processing causing apache to restart)?

Thanks.

/sebby

[ffeingol]

We just got the 11.25 version last night also (Release). We're seeing a lot more Apache restarts than normal on one server. They are not people adding domains/sub-domains/add-on's etc.

[VeZoZ]

What is the technical reason for restarting at all when doing stats processing? This did not happen before 11.25 so I'm curious why the sudden change. I'm also curious if that's restarts per account or for an entire run through. If that's per account wow that's a lot of restarts when you start dealing with higher density servers that could have 1000 accounts on them or more.

[cPanelDon]

What is the technical reason for restarting at all when doing stats processing? This did not happen before 11.25 so I'm curious why the sudden change. I'm also curious if that's restarts per account or for an entire run through. If that's per account wow that's a lot of restarts when you start dealing with higher density servers that could have 1000 accounts on them or more.

Apache (httpd) must be restarted when logs are rotated to ensure new data is logged to the correct file and not logged to the older, just-rotated log file with data that has already been processed. In the ideal scenario cpanellogd will restart Apache only once after processing logs for all users. The frequency of log processing depends on your statistics schedule configuration as defined via WHM.

The time intervals between log and bandwidth processing runs is defined by a number of hours and both may be modified independently via the following menu path in WHM:
WHM: Main >> Server Configuration >> Statistics Software Configuration >> Schedule Configuration
* Log Processing Frequency "Process log files every # hours."
* Bandwidth Processing Frequency "Process bandwidth every # hours."
Documentation: Statistics Software Configuration

[Bono_]

Haha this looks like 1st april joke, restarting apache every 15 minutes are you insane?

All my cron jobs are also messed up...is there any way to return to previous state where I can process logs once a day and I restart web server when I want to restart mostly once a month.

You can advertise with brand new cpanel none opcode caching doesn't work anymore.

I thought most stupid thing cpanel did was trying to control urchin6 even though it doesn't support it. But it also tried to control it like this shutting down it's scheduler.

I'm only sorry I didn't read this topic before I would switch to stable release because this is insane.

[cPanelDon]

As WHM-11.25 is moving into the RELEASE tree (R-42213) I can see that this issue was addressed:

Are those issues all related to this bug?

The change log entries are not all for the same exact issue, but it is possible for a correction to have a positive impact on or be related to more than one issue.

What does the "Reduced Apache restarts triggered by log processing to no more than 3" mean?

In an ideal situation only one restart will occur, but it is possible that a maximum of 3 could be triggered during a stats run to ensure Apache (httpd) is logging to the correct files after rotating logs.

Can someone confirm that this issue was definitely fixed?

In testing the issue is fully resolved. If persistent, frequently recurring symptoms are experienced they should be reported via a support request so the specific details and circumstances may be fully investigated, properly documented, and tracked through resolution.

What was causing apache to restart so often (how/why was the log processing causing apache to restart)?

Thanks.

/sebby

This is answered in detail via my last reply: cPanel Forums - View Single Post - [11.25 Current] apache restarting every hour

[cPanelDon]

All my cron jobs are also messed up...is there any way to return to previous state where I can process logs once a day and I restart web server when I want to restart mostly once a month.

This is answered by the following post: cPanel Forums - View Single Post - [11.25 Current] apache restarting every hour

In addition to being able to customize the status schedule configuration via WHM it is also possible to customize the upcp and cpbackup cron tab entries via an optional WHM plug-in as detailed here: cPanel Forums - View Single Post - WHM Plugin to Configure cPanel Cron Times

[Bono_]

This is answered by the following post: cPanel Forums - View Single Post - [11.25 Current] apache restarting every hour

In addition to being able to customize the status schedule configuration via WHM it is also possible to customize the upcp and cpbackup cron tab entries via an optional WHM plug-in as detailed here: cPanel Forums - View Single Post - WHM Plugin to Configure cPanel Cron Times

I fixed cron tabs manually, and how come it worked perfectly for lets say 10 years and from now on web server needs to be restarted so logs are processed properly?
So my question is how to get it to previous state to process files like it used to without apache restart.
I'm sorry to say but this is more annoying that Vista UAC notices.

[VeZoZ]

Apache (httpd) must be restarted when logs are rotated to ensure new data is logged to the correct file and not logged to the older, just-rotated log file with data that has already been processed. In the ideal scenario cpanellogd will restart Apache only once after processing logs for all users. The frequency of log processing depends on your statistics schedule configuration as defined via WHM.

I'm just wondering if there can be a tweak settings option to disable this restart functionality? It was not there before and in some cases it might not be desired and the potential loss of some stats data is not worth the restarts.

The way it sounds though you'd want to be doing this every time you stats process a site otherwise I don't see much point in what it's doing now. You have 1000 accounts you're going to ignore the first 999 accounts so that the 1000th account you actually restart once rotated to stats going to the wrong log file.

[m0rgulvale]

i am seeing random restarts too now and i can't have this

what can i do to stop this from happening other than downgrading?

is possible to just disable logging?

my settings are changed now to:
Bandwidth Processing Frequency 24hrs

Log Processing Frequency 24hrs

[m0rgulvale]

will disabling all stats programs stop apache from randomly restarting?

randomly restarting apache just doesn't make sense.. what if user happens to be doing something really important at that time?

[ffeingol]

Are you actually seeing anything bad happen? It's issuing a 'gradeful' restart. That means the current process is allowed to complete, log files etc. get closed. Depending on how you have Apache configured this happens all the time.