Service stopped - a lot

[calderwood]

Since upgrading to 8.2 (COS 4.2)from 8.1 I am getting a lot of service stopped messages, and I assume, services closing down, although whenever I check they seem to be running. Mostly IMAP/POP3 and WEB SERVER. Here is a typical log:

Events:
[Jul 27, 2007 01:51 PM] 'Plesk SpamAssassin' service started.

[Jul 27, 2007 01:46 PM] 'Plesk SpamAssassin' service stopped.

[Jul 27, 2007 05:28 AM] 'Plesk Web Server' service started.

[Jul 27, 2007 05:11 AM] 'Plesk Web Server' service stopped.

[Jul 27, 2007 04:55 AM] 'Plesk Web Server' service started.

[Jul 27, 2007 04:45 AM] The 'IMAP/POP3 Server (Courier-IMAP)' service was released from monitoring because it failed frequently.

[Jul 27, 2007 04:39 AM] 'Plesk Web Server' service stopped.

[Jul 27, 2007 04:34 AM] 'IMAP/POP3 Server (Courier-IMAP)' service stopped.

[Jul 27, 2007 04:34 AM] 'IMAP/POP3 Server (Courier-IMAP)' service stopped.

[Jul 27, 2007 04:33 AM] 'IMAP/POP3 Server (Courier-IMAP)' service stopped.

[citypro]

I am a novice that purhcased a server and have just started to get this problem. I don't now how to fix it. I stop and started the service from Plesk 8.2.0 but the problem persists.

Can somebody please advise a mere novice what I should do!

[garan]

I've found that the watchdog has been timing out for Courier-IMAP because I simply didn't have enough daemons running to service the load that was going through this particular server. Upped the number of daemons and I've not seen a watchdog message since..

[64bithost.com]

[Nov 20, 2007 03:55 AM] 'Plesk Web Server' service started.
[Nov 20, 2007 03:43 AM] 'Plesk SpamAssassin' service stopped.
[Nov 20, 2007 03:43 AM] 'Plesk Web Server' service stopped.
[Nov 20, 2007 03:37 AM] 'Web Server (Apache)' service stopped.
[Nov 20, 2007 03:31 AM] 'Web Server (Apache)' service stopped.
[Nov 19, 2007 07:03 PM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 06:57 PM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 10:49 AM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 10:44 AM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 08:58 AM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 08:53 AM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 02:54 AM] 'Plesk SpamAssassin' service started


Anybody got a clue how to fix this?

[calderwood]

Originally posted by garan
I've found that the watchdog has been timing out for Courier-IMAP because I simply didn't have enough daemons running to service the load that was going through this particular server. Upped the number of daemons and I've not seen a watchdog message since..

Did this stop all services errors or just the IMAP related errors? What did you increase the daemon number to?

[garan]

Originally posted by 64bithost.com
[Nov 20, 2007 03:55 AM] 'Plesk Web Server' service started.
[Nov 20, 2007 03:43 AM] 'Plesk SpamAssassin' service stopped.
[Nov 20, 2007 03:43 AM] 'Plesk Web Server' service stopped.
[Nov 20, 2007 03:37 AM] 'Web Server (Apache)' service stopped.
[Nov 20, 2007 03:31 AM] 'Web Server (Apache)' service stopped.
[Nov 19, 2007 07:03 PM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 06:57 PM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 10:49 AM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 10:44 AM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 08:58 AM] 'Plesk SpamAssassin' service started.
[Nov 19, 2007 08:53 AM] 'Plesk SpamAssassin' service stopped.
[Nov 19, 2007 02:54 AM] 'Plesk SpamAssassin' service started


Anybody got a clue how to fix this?

Is your spamassassin daemon still there between the time stated as stopped and started? If the daemon is still running when it is stated as stopped by the watchdog it could be because you aren't running enough daemons to service load.

[garan]

Originally posted by calderwood
Did this stop all services errors or just the IMAP related errors? What did you increase the daemon number to?

I increased MAXDAEMONS to 100 in the /etc/courier-imap/imapd file. I also increased MAXPERIP from 4 to 100 as Horde uses imap to connect locally meaning that the watchdog is competing with the Apache daemon for connections on 127.0.0.1.

Garan.

[calderwood]

Originally posted by garan
I increased MAXDAEMONS to 100 in the /etc/courier-imap/imapd file. I also increased MAXPERIP from 4 to 100 as Horde uses imap to connect locally meaning that the watchdog is competing with the Apache daemon for connections on 127.0.0.1.

Garan.

Thanks - I've reset to 100 and Maxip to 40 as I have 12 IPs on the server and don't want to overload the CPU too much - we'll see how that works. Am I correct in assuming that the setting is up to 100 daemons at once, but no more than 40 per IP?

[garan]

Originally posted by calderwood
Thanks - I've reset to 100 and Maxip to 40 as I have 12 IPs on the server and don't want to overload the CPU too much - we'll see how that works. Am I correct in assuming that the setting is up to 100 daemons at once, but no more than 40 per IP?

My understanding is that MAXDAEMONS is the number of imap daemons that can spawn on the box and that MAXIP is the maximum number of connections to accept from a particular IP. So they are unrelated to each other. The MAXIP is needed if you have either a lot of connections on localhost (many 100's of webmail users for example) or if you have a customer who uses IMAP and has a number of IMAP users sitting behind a single NAT'd IP address (e.g. on a corporate LAN).

Garan.

[calderwood]

Originally posted by garan
My understanding is that MAXDAEMONS is the number of imap daemons that can spawn on the box and that MAXIP is the maximum number of connections to accept from a particular IP. So they are unrelated to each other. The MAXIP is needed if you have either a lot of connections on localhost (many 100's of webmail users for example) or if you have a customer who uses IMAP and has a number of IMAP users sitting behind a single NAT'd IP address (e.g. on a corporate LAN).

Garan.

I have to dive into a book and see if I can get the answer - I did a quick internet search and got three different answers. Nothing in the Plesk KB.

[originalmguy]

I am experiencing this behaviour with the Plesk Web Server.
Any suggestions?

Thanks in advance,

M

:-)

[calderwood]

By increasing the MaxDaemons I have reduced the shutdowns, but I've had two Plesk Web Service stop/starts this week, but that is better than before I increased where I was getting six or more a day. Two in two weeks is good.