The Plesk Web Server service on host ... is down.

[stefan rinderle]

Hi,

since I installed the watchdog module and activated the check of plesk web service. I get 4-5 times a day the following mail:
The Plesk Web Server service on host is down.
and shortly after that a mail that it has been started.
The Plesk Web Server service on host has been started on Mär 9, 2010 01:35 .

I search in several log-files but could not find the file or information why its been stopped or started. Has anyone also this problem or fixed it already? Where is the log-file I have to look?

I'm working with plesk 9.3 on an ubuntu 8.10 system.

thanks for your ideas

 

[IgorG]

Try to find reason of stopping in /var/log/sw-cp-server/error_log file.

 

[stefan rinderle]

log-file

I already looked in this file:
2010-03-09 02:40:10: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-03-09 05:19:39: (log.c.135) server stopped
2010-03-09 05:19:40: (log.c.75) server started

There is no additional info beside the ssl failure, but it seems that ihe ssl failure has nothing to do with my problem. Any other ideas?

Thanks

 

[stefan rinderle]

logfile

Hi,

the logfile has no significant entries. Beside some ssl failure at other times, theres ionly the message, that the plesk web server hast been startet and stopped:
2010-03-09 13:51:33: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-03-09 15:06:16: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-03-09 15:31:34: (log.c.135) server stopped
2010-03-09 15:31:35: (log.c.75) server started
2010-03-09 15:36:42: (log.c.135) server stopped
2010-03-09 15:36:44: (log.c.75) server started
2010-03-09 15:41:52: (log.c.135) server stopped
2010-03-09 15:41:53: (log.c.75) server started
2010-03-09 15:47:00: (log.c.135) server stopped

Are there any other log-files?

Thanks

 

[IgorG]

Is there something related in /var/log/messages ?

 

[stefan rinderle]

/var/log/messages

Hi,

no, there are only some entries like this:
Mar 9 13:53:42 h1365957 -- MARK --
Mar 9 14:13:42 h1365957 -- MARK --
Mar 9 14:33:42 h1365957 -- MARK --
Mar 9 14:53:42 h1365957 -- MARK --
Mar 9 15:13:42 h1365957 -- MARK --
Mar 9 15:33:42 h1365957 -- MARK --
Mar 9 15:53:42 h1365957 -- MARK --

But the funny thing is, since the watchdog decided to stop monitoring the plesk web server, theres no more message in /var/log/sw-cp-server/error_log. So it seems that the error is in watchdog. Or any other ideas?

 

[SGT]

I am having this trouble too.

Nothing in /var/log/messages

Same issues in /var/log/sw-cp-server/error_log file:

2010-03-11 07:24:17: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-03-11 07:40:35: (log.c.135) server stopped
2010-03-11 07:40:36: (log.c.75) server started
2010-03-11 07:40:36: (mod_fastcgi.c.1337) Dynamic spawning with max_procs > 1 is not supported; setting min_procs = max_procs
2010-03-11 07:40:36: (mod_fastcgi.c.1337) Dynamic spawning with max_procs > 1 is not supported; setting

I wonder...

1) What does "log.c.135" mean?
2) Why should we be getting SSL handshake errors?
3) What's going on with "Dynamic spawning with max_procs > 1 is not supported; setting min_procs = max_procs"?

 

[IgorG]

http://kb.odin.com/en/6875

 

[SGT]

I have 9.3 - your patch refers to 9.2.

Also it refers to log messages with "error "Cannot allocate memory":

"php (pre-forking): Cannot allocate memory"

But that's not what I reported from my logs.

So are you saying it is safe and relevant to apply this hotfix?

 

[IgorG]

So are you saying it is safe and relevant to apply this hotfix?

You can try it and revert all back if it not help.

 

[AnsgarS]

Webserver down issue as well

but i am not sure if it is SSL related since the time between the SSL error and the de server going down is significant.

Yesterday i reinstalled PLESK intirely; so i am working with a clean install from Plesk template. I had to downgrade openssl-0.9.8e-12.el5_4.6 to version 4.1 (Known bug for plesk 9.3), as well as mod_ssl and httpd as they are related.

I created one domain; and then got these errors over night.

Same issues in /var/log/sw-cp-server/error_log file:

2010-04-05 20:31:19: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-04-06 00:13:56: (log.c.135) server stopped
2010-04-06 00:13:58: (log.c.75) server started
2010-04-06 00:19:08: (log.c.135) server stopped
2010-04-06 00:19:09: (log.c.75) server started
2010-04-06 01:04:44: (log.c.135) server stopped
2010-04-06 01:04:44: (log.c.75) server started
2010-04-06 01:09:53: (log.c.135) server stopped
2010-04-06 01:09:54: (log.c.75) server started
2010-04-06 01:20:11: (log.c.135) server stopped
2010-04-06 01:20:13: (log.c.75) server started
2010-04-06 01:40:28: (log.c.135) server stopped
2010-04-06 01:40:28: (log.c.75) server started
2010-04-06 01:45:41: (log.c.135) server stopped
2010-04-06 01:45:42: (log.c.75) server started
2010-04-06 04:12:12: (log.c.135) server stopped
2010-04-06 04:12:15: (log.c.75) server started
2010-04-06 04:17:25: (log.c.135) server stopped
2010-04-06 04:17:26: (log.c.75) server started
2010-04-06 04:27:38: (log.c.135) server stopped
2010-04-06 04:27:39: (log.c.75) server started
2010-04-06 05:43:26: (log.c.135) server stopped
2010-04-06 05:43:27: (log.c.75) server started
2010-04-06 09:05:20: (log.c.135) server stopped
2010-04-06 09:05:20: (log.c.75) server started

2010-04-06 12:24:14: (connections.c.299) SSL: 1 error:140780E5:SSL routines:SSL23_READ:ssl handshake failure
2010-04-06 12:29:53: (connections.c.281) SSL: -1 5 104 Connection reset by peer
2010-04-06 12:29:53: (connections.c.281) SSL: -1 5 104 Connection reset by peer
2010-04-06 13:02:44: (log.c.135) server stopped
2010-04-06 13:02:45: (log.c.75) server started

Could these SSL errors be created by the SSH connection?

Anyway, I am more concerned with the server stopping and restarting.

Anyone got ideas?

 

[IgorG]

Look at Workarounds thread. I have posted fix for openssl related problem a few days ago.

 

[AnsgarS]

SSL issue is not the cause

Hi,
I installed the CP server fix. And I dont have ssl errors anymore. But as I wrote in my previous post: the stopping and starting is unrelated to the SSL issue. I still get the watchdog messages all the time:

2010-04-07 00:13:30: (log.c.135) server stopped
2010-04-07 00:13:33: (log.c.75) server started
2010-04-07 00:18:44: (log.c.135) server stopped
2010-04-07 00:18:45: (log.c.75) server started
2010-04-07 00:23:54: (log.c.135) server stopped
2010-04-07 00:23:55: (log.c.75) server started
2010-04-07 01:04:23: (log.c.135) server stopped
2010-04-07 01:04:24: (log.c.75) server started
2010-04-07 01:44:59: (log.c.135) server stopped
2010-04-07 01:45:00: (log.c.75) server started
2010-04-07 01:50:09: (log.c.135) server stopped
2010-04-07 01:50:10: (log.c.75) server started
2010-04-07 03:05:57: (log.c.135) server stopped
2010-04-07 03:05:58: (log.c.75) server started
2010-04-07 04:06:42: (log.c.135) server stopped
2010-04-07 04:06:44: (log.c.75) server started
2010-04-07 04:11:54: (log.c.135) server stopped
2010-04-07 04:11:54: (log.c.75) server started
2010-04-07 04:17:05: (log.c.135) server stopped
2010-04-07 04:17:06: (log.c.75) server started
2010-04-07 04:27:18: (log.c.135) server stopped
2010-04-07 04:27:19: (log.c.75) server started
2010-04-07 05:43:06: (log.c.135) server stopped
2010-04-07 05:43:07: (log.c.75) server started
2010-04-07 09:04:58: (log.c.135) server stopped
2010-04-07 09:04:59: (log.c.75) server started

During daytime, when there is server load I do not get these messages as often as during nighttime

Any thoughts on this?

 

[IgorG]

Do you have installed all updates (openssl too) and this patch? Mentioned workaround should fix this problem too.

 

[AnsgarS]

Which fixes do you mean exactly?

I am working with openssl 4.6 at the moment and installed the CP-SW-server fix dated april 1st.

I did not do: http://kb.odin.com/en/6875 yet, as it was rated for plesk 9.2, while I run 9.3

Would be great if you can summarize which fixes need to be applied exactly. Thanks

 

[AnsgarS]

I just found in one of the other threads the following:

----------------------
it is known problem caused by small connection timeout in watchdog settings (5 seconds) which may cause Watchdog detect apche down on busy systems.

By default Watchdog expects services to respond within 5 seconds. So I suggest you to go to
Modules > Watchdog > apache

and change 'Connection timeout' parameter from 5 to a larger value, e.g. 20.

That usually eliminates such problems. If it will not help, try to disable watchdog (because apache itself works fine).
------------------------

I will be trying this change now and see what happens.

 

[Mike Burn]

Don't know if this helps, but we noticed that the latest version of openssl provided by our OS (in this case RHEL 5.5) is incompatible with that required by plesk. We had to forcefully downgrade openssl back to a compatible version.

Matt

 

[SacAutos]

It's been happening to our 9.3 server as well

We've been experiencing this problem as well ever since we upgraded our "lead" server from 9.2 to 9.3 . Not experiencing it on another server that's also running 9.3 but that was a fresh build. Very annoying...

 

[AnsgarS]

Solution seems to work

Hi all, just to confirm,

Since I changed the timeout of the watchdog as described earlier in this thread, I only get one watchdog message a day instead of 20. So that's an improvement.

cheers, Ansgar

 

[AnsgarS]

problem solved

just to re-confirm: i increased time-out to 30secs, and in three days i haven't seen a watch-dog notification anymore. Problem solved