The Plesk Web Server service on host as.xxxxx.com is down.

[galbus]

I tried to reply to the following post on this forum which has the same unresolved issue, but it is not accepting replies:

http://forum.swsoft.com/showthread.php?threadid=39991

I am experiencing the same problem in that I occasionally receive emails from watchdog stating

The Plesk Web Server service on host as.xxxxx.com is down.

and then 5 mins later another stating

The Plesk Web Server service on host as.xxxxx.com has been started on [date & time]

I have asked my web host (Media Temple) if they have any ideas on what is going on and they replied:

Do you have any cron jobs running during that time? Perhaps your script forced watchdog to restart your server. If it was a cron job, try scheduling it less frequently or during a time that sees less traffic or will have less resources used.

I then looked at the crontab of every domain and found no scheduled tasks.

Then I went through every system user listed in 'Server > Scheduled Tasks'. The only one that had tasks listed was 'root'.

The tasks were:


Min H DM M DW Command
0 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/secur-check
0 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/send-report weekly
0 3 * * 7 /usr/local/psa/libexec/modules/watchdog/cp/clean-reports
10 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/clean-sysstats
11,26,41,56 * * * * /usr/local/psa/admin/sbin/backupmng >/dev/null 2>&1
15 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats day
15 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats week
15 1 1 * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats month
15 1 1 * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats year
20 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/clean-events


Recent times and dates of the emails from watchdog are:

• The problem was discovered on Mar 5, 2007 12:09 PM.
• The problem was discovered on Mar 3, 2007 12:10 PM.
• The problem was discovered on Feb 25, 2007 12:15 PM.
• The problem was discovered on Feb 25, 2007 12:10 PM.
• The problem was discovered on Feb 3, 2007 12:09 PM.
• The problem was discovered on Jan 25, 2007 02:52 AM.
• The problem was discovered on Jan 25, 2007 02:32 AM.
• The problem was discovered on Jan 25, 2007 01:51 AM.

Each one of these emails was followed by another exactly 5 mins later saying that the Plesk Web Server service has been started.

Any ideas? I'm stuck and can't find any more info myself.

 

[albans]

Don't you think it's the proof that your watchdog is doing its job?

- when apache stops, it sends an alert
- then it tries to start apache again
- and if it works, it sends an alert again


-----UPDATE-----
Haaa, sorry fot this "form of" stupid answer, I didn't understand at first that the problem was happening every morning...

 

[galbus]

No problem albans, I should have made it more clear:

The issue that is concerning me is that the web server is repeatedly being reported as being down. It is not that watchdog might not be doing its job properly.

Unless this is a problem with watchdog? I suppose that's something else to consider.

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

Edit:

OK I've just thought of something. It seems recently that the reports have all been around 10mins past the hour and the only job running at xx:10 is

/usr/local/psa/libexec/modules/watchdog/cp/clean-sysstats

Is that likely to be the problem?

 

[ianotr]

We are having the exact same problem on a server that was recently re-imaged to 8.1. It runs along fine and then gets real slow and we get a message that the Plesk Web Server service is down from Watchdog.

We reinstalled SA and have also looked for problem cron tasks but cannot find anything that could be causing the problem. If we turn off SA then the server seems to run fine without issues but after we turn it back on a few hours later we get a series of emails from Watchdog until it finally quits monitoring.

Suggestions would be appreciated.

 

[Nicolas Bussiere]

Hi !

I am not in 8.1, but 9.0, I upgraded last weekend. Since that time, 4 to 6 times per day, Watchdog sends me an email saying the same thing (down) then UP 5 minutes later. The Web server seems to work fine. So it's not related to 8.1.

Here are some clues -- I did not find the problem. I thought it was related to backup (checkbox selected with "site is not available during backup"), it's not that. Cron jobs? No. My next hypothesis is it's related to a severe load on Apache, which appear "down" to the watchdog.

- WatchDog: Apache, are you alive?
- Apache: ....
- WatchDog: Apache?
- Apache: ...
WatchDog sends an email to administrator
- Apache: Yes, I am fine, it's just I was serverly loaded...
- Watchdog: OK
Watchdog sends an email to administrator

As my Watchdog make calls every 300 seconds (default), that explains the delay. I found no places to adjust the sensibility of the Watchdog. So what does mean "Down" for the WatchDog? Not responding for how many seconds?

Thanks to share !

 

[Roel Mulder]

Same issue here on Plesk 9.0.1 on a server that is not doing very much. Certainly no Apache load in the middle of the night. Yet, the message is sent out often enough to annoy me into this research...

The one thing is, it specifically mentions the "Plesk Web Server service" to be down, not Apache. Are there two systems running? Probably, since Watchdog has a nice graph of resource usage. Apache (orange) is near nil, just a few MB of memory, whilst the Plesk Web Service (dark blue) permanently uses between 1GB and 1.5GB of memory...

! Correction apr 21; the 1 - 1.5GB memory is the dark blue line 'Overall statistics' in the diagram, whilst 'Plesk Web Server' which is also a dark blue, doesn't seem to use any resources at all.

Apache started an incremental line of permanent system usage over the past few weeks, reached a 100MB, and the service was restarted manually. It's line is nicely back down, but messages that PWS or Plesk SpamAssassin is down keep coming in.

 

[shall]

I'm experiencing something similar to this, and I also am troubleshooting something on the same server where the backup tmp folder keeps filling forever. I deleted 74gb of **** data that was all created within a 3 hour period in it last week during "backups" (there's only about 15gb of data TOTAL on this server). I'm guessing that backups are getting stuck in an infinite loop backing up the backup files to themselves - and that's killing performance, thus effecting the "web server down" issue.

 

[Peter Vogensen]

Any more news on this? I've been having this problem for a while now - but am unable to find a solution or answer.

 

[shall]

Since disabling the auto-backup and running it manually outside of Plesk, we haven't had the same issue reappear.

 

[parwatcha]

Hi !
- WatchDog: Apache, are you alive?
- Apache: ....
- WatchDog: Apache?
- Apache: ...
WatchDog sends an email to administrator
- Apache: Yes, I am fine, it's just I was serverly loaded...
- Watchdog: OK
Watchdog sends an email to administrator
Thanks to share !

Were u ever able to check if ur websites were still reachable during that "unreal down time" and have u maybe checked ur statistics if there was a high amount of users during that time?

 

[bskrakes]

Hi there, I to have the same issue... What I'd like to know is where this gets logged in the system. I have checked the /var/log/messages and can find nothing logged regarding this error. Does Plesk use a different directory to log system issues specific to Plesk? Perhaps the log file (if there is one) will provide more clues.

Thank you in advance!

 

[JanosS]

check /usr/local/psa/var/modules/watchdog/log/monit.log
I have the same issue here.

I found this in monit.log:

[CEST Aug 19 10:20:48] HTTP: error receiving data -- Resource temporarily unavailable
[CEST Aug 19 10:20:51] 'plesk_apache' failed protocol test [HTTP] at INET[localhost:8443].
[CEST Aug 19 10:20:51] 'plesk_apache' trying to restart
[CEST Aug 19 10:20:51] 'plesk_apache' stop: /etc/init.d/sw-cp-server
[CEST Aug 19 10:20:52] 'plesk_apache' start: /etc/init.d/sw-cp-server
[CEST Aug 19 10:20:59] Collector: error receiving data from http://127.0.0.1:2912/ -- Resource temporarily unavailable
[CEST Aug 19 10:20:59] Collector: communication failure -- aborting message
[CEST Aug 19 10:26:00] 'plesk_apache' connection passed

So watchdog polled plesk_apache, and when it found it inavaible, restarted it, then emailed me.

But I still don't know the root cause, why was it unavaible.

Other question: what watchdog wants on port 2912?

 

[AthurA]

I tried to reply to the following post on this forum which has the same unresolved issue, but it is not accepting replies:

http://forum.swsoft.com/showthread.php?threadid=39991

I am experiencing the same problem in that I occasionally receive emails from watchdog stating



and then 5 mins later another stating



I have asked my web host (Media Temple) if they have any ideas on what is going on and they replied:



I then looked at the crontab of every domain and found no scheduled tasks.

Then I went through every system user listed in 'Server > Scheduled Tasks'. The only one that had tasks listed was 'root'.

The tasks were:


Min H DM M DW Command
0 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/secur-check
0 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/send-report weekly
0 3 * * 7 /usr/local/psa/libexec/modules/watchdog/cp/clean-reports
10 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/clean-sysstats
11,26,41,56 * * * * /usr/local/psa/admin/sbin/backupmng >/dev/null 2>&1
15 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats day
15 1 * * 1 /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats week
15 1 1 * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats month
15 1 1 * * /usr/local/psa/libexec/modules/watchdog/cp/pack-sysstats year
20 1 * * * /usr/local/psa/libexec/modules/watchdog/cp/clean-events


Recent times and dates of the emails from watchdog are:

• The problem was discovered on Mar 5, 2007 12:09 PM.
• The problem was discovered on Mar 3, 2007 12:10 PM.
• The problem was discovered on Feb 25, 2007 12:15 PM.
• The problem was discovered on Feb 25, 2007 12:10 PM.
• The problem was discovered on Feb 3, 2007 12:09 PM.
• The problem was discovered on Jan 25, 2007 02:52 AM.
• The problem was discovered on Jan 25, 2007 02:32 AM.
• The problem was discovered on Jan 25, 2007 01:51 AM.

Each one of these emails was followed by another exactly 5 mins later saying that the Plesk Web Server service has been started.

Any ideas? I'm stuck and can't find any more info myself.

Thanks for sharing.

 

[Justin Clarke]

Anyone have any ideas about this "Plesk Web Server" issue. If I monitor the service in watchdog I get emails every 5 minutes saying it's down. Also on the watchdog page it has the red mark by it, where as all the other services have green marks to show they are "up".

I have tried setting the connection timeout to 120 seconds, and also the polling to 120 but this makes no difference.

I don't think it's an issue with the "Plesk Web Server" as before I started using watchdog this was constantly "up". Only since I enabled watchdog it's being reported as "down". What could be causing the watchdog to think that the "Plesk Web Server" is down?

Thank you.

 

[bskrakes]

I have been unable to trouble shoot this issue any further... I did notice that the latest version of Plesk 9.5.3 for Linux/UNIX will disable the auto notifications if the service crashes to much... That would be handy IF IT ACTUALLY TOLD US WHY IT CRASHED SO MUCH... logging seems weak or else I am missing something which is probably very simple?

 

[Justin Clarke]

I have been unable to trouble shoot this issue any further... I did notice that the latest version of Plesk 9.5.3 for Linux/UNIX will disable the auto notifications if the service crashes to much... That would be handy IF IT ACTUALLY TOLD US WHY IT CRASHED SO MUCH... logging seems weak or else I am missing something which is probably very simple?

They seem to have a tendancy to disable notifications rather than fix issues. This happened with 451 SMTP queue file write errors. They just disabled email alerts (or made them appear much less frequently) to make that go away.

I remember when I first got my new server, think it was Plesk v9.0.1 (possibly) and I added Watchdog and it kept alerting me every 5 mins (this was on a brand new install), so I just removed Watchdog then and I thought it was erroneous. Now I'm re-visiting it a year on and have the same issue. So nothing's changed.

All other services are reported as "up" all the time, it's just the Plesk specific service "Plesk Web Server" that keeps alerting.

I'm currently looking at alternatives such as Monit (http://mmonit.com/monit/), but would be nice if they sorted Watchdog out.

 

[bskrakes]

Interesting... let me know what you decided to go with and how the deployment goes

Have you ever looked into http://www.atomicrocketturtle.com/?

 

[Justin Clarke]

Interesting... let me know what you decided to go with and how the deployment goes

Have you ever looked into http://www.atomicrocketturtle.com/?

I looked at them before for Anti-virus solutions but never implemented. I will take another look but not for a few weeks as I'm away and want to leave the server as is for now. I will try to remember to get back to this thread to let you know what I eventually did.

 

[Justin Clarke]

Interesting... let me know what you decided to go with and how the deployment goes

Have you ever looked into http://www.atomicrocketturtle.com/?

Right, I got back off hols and decided to play with system monitoring.

I did look at atomic rocket turtle but that's not free, so I went for Munin and Monit in the end.

Good tutorial here: http://www.howtoforge.com/server-monitoring-with-munin-and-monit-on-centos-5.2

Up and running now. If anyone follows that and needs help then let me know and I'll try and help you out.

 

[bskrakes]

thanks Justin, I am going to check that out... err just have to make sometime!