Zabbix_agentd never recovers from a broken connection to server

[cameronsto]

I run zabbix_server on my home connection and have agents on the local network at home and outside the local network. Whenever my home internet connection has issues and the agents never seem to recover. This happened last night around 6:30 and I had to restart the external agents in order for them to begin processing again. Here's the log file from one of the win32 agents:

[05-Jan-2006 15:59:28] Processing took more then 1000 milliseconds (2265 milliseconds)
[05-Jan-2006 16:02:30] Processing took more then 1000 milliseconds (1125 milliseconds)
[05-Jan-2006 18:34:21] Active checks [Error in connect()]
[06-Jan-2006 11:24:14] Zabbix Win32 Agent stopped

Any ideas?

-cameron

[cameronsto]

Has anybody else experienced this problem?

-cameron

[bbrendon]

Yes. The below caused the zabbix server to report the win32 server was down because it wasn't sending information to the zabbix server in a timely fashion. The server was under high load fro the time when the errors below occured. Hmmm

Solution? Still looking

[01-May-2006 11:00:52] Processing took more then 1000 milliseconds (3996 milliseconds)
[01-May-2006 11:08:37] Processing took more then 1000 milliseconds (1542 milliseconds)
[01-May-2006 11:08:58] Processing took more then 1000 milliseconds (3004 milliseconds)
[01-May-2006 11:10:16] Processing took more then 1000 milliseconds (4006 milliseconds)
[01-May-2006 11:10:22] Processing took more then 1000 milliseconds (3936 milliseconds)
[01-May-2006 11:10:49] Processing took more then 1000 milliseconds (4998 milliseconds)
[01-May-2006 11:10:59] Processing took more then 1000 milliseconds (5889 milliseconds)
[01-May-2006 11:11:09] Processing took more then 1000 milliseconds (5237 milliseconds)
[01-May-2006 11:11:22] Processing took more then 1000 milliseconds (3395 milliseconds)
[01-May-2006 11:11:28] Processing took more then 1000 milliseconds (3535 milliseconds)
[01-May-2006 11:11:37] Processing took more then 1000 milliseconds (4737 milliseconds)
[01-May-2006 11:11:44] Processing took more then 1000 milliseconds (1973 milliseconds)
[01-May-2006 11:11:52] Processing took more then 1000 milliseconds (1532 milliseconds)
[01-May-2006 11:12:03] Processing took more then 1000 milliseconds (4256 milliseconds)
[01-May-2006 11:12:13] Processing took more then 1000 milliseconds (3475 milliseconds)
[01-May-2006 11:12:17] Processing took more then 1000 milliseconds (1042 milliseconds)
[01-May-2006 11:12:26] Processing took more then 1000 milliseconds (4406 milliseconds)
[01-May-2006 11:12:32] Processing took more then 1000 milliseconds (1462 milliseconds)
[01-May-2006 11:12:38] Processing took more then 1000 milliseconds (3404 milliseconds)
[01-May-2006 11:12:46] Processing took more then 1000 milliseconds (3004 milliseconds)

[bbrendon]

It seems these problems have been resolved in the current release.

More descriptive information on the connect errors messages would still be helpful for active agents.