Re: [monit] Monit believes web service is up even when it is not

From:

Martin Pala

Subject:

Date:

Thu, 5 Nov 2009 20:08:33 +0100

Hi,

please run monit in verbose mode (-v option) and check logs for detailed test results. The response time is reset to -1 in case that monit performs restart => it seems that monit probably tried to restart the webserver.

Regarding spawned processes - there usually is some parent process, which controls workers spawning - it can start/stop the worker processes (even if it crashes). Apache has similar model ... monit can be set to watch the parent/master process which usually creates pid file.

Martin

On Nov 5, 2009, at 7:23 PM, Balcoes wrote:

Hi

I have monit v5.0 and CentOS release 4.8 (Final).

I try monitor Lotus Domino web server with the following config file:
    check host myserver with address myserver
        start program  "/etc/init.d/domino start" with timeout 300 seconds
        stop program   "/etc/init.d/domino stop"  with timeout 150 seconds
        if failed host 192.168.1.10 port 80 protocol http 10 times within 10 cycles then restart
        if 6 restarts within 60 cycles then timeout

If the web server is up, monit status page shows:
  Remote Host 'myserver'
    status                            online with all services
    monitoring status                 monitored
    port response time                0.001s to 192.168.1.10:80 [HTTP via TCP]
    data collected                    Thu Nov  5 20:09:55 2009
as it should.

However if the server is down, monit states:
  Remote Host 'myserver'
    status                            online with all services
    monitoring status                 monitored
    port response time                -1.000s to 192.168.1.10:80 [HTTP via TCP]
    data collected                    Thu Nov  5 20:10:57 2009

I.e. the only difference is that port response time is -1.000s. But monit believes that the server is up and doesn’t try to reboot it. How come? What am I missing?

B

P.S: The config file doesn’t use PID file as Domino doesn’t have a single process. Instead it spawns typically approximately 200 processes.

--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general