Re: restart failed

[Jan-Henrik Haukeland]

On 17. jul. 2005, at 10.03, Pavel Urban wrote:

> Hello,
>
> we have separated 'alert' and 'timeout' notifications and thess  
> lines in monit.conf
>
> check process mta-dispatcher with pidfile /usr/iplanet/mta-in1/msg- 
> mta-in1/config/pidfile.imta_dispatch
>   start program  "/usr/iplanet/mta-in1/msg-mta-in1/imsimta start  
> dispatcher"
>   stop program  "/usr/iplanet/mta-in1/msg-mta-in1/imsimta stop  
> dispatcher"
>   if 5 restarts within 5 cycles then timeout
>   if failed host 192.168.10.12 port 25 protocol smtp then restart
>
> It work ok, but today we've noticed that 'stop' method failed - and  
> isn't tried again. It is probably normal.

Yes, if it didn't work the first time it is assumed that it wont work  
(in this cycle). I think this is okay..

> BUT - no timeout occures and therefore no 'timeout' notifications  
> are sent, while the service is effectively not running. Can this be  
> accomplished somehow?

Well, timeout occurs only in the reverse situation - if monit failed  
to start the service in x out of y cycles. In this case monit failed  
to stop. However if you put this at the top of your monitrc file,  
"set alert address@hidden", or "alert address@hidden [on {exec}]" in the check  
process entry you _will_ at least get an alert message if monit  
failed to execute the stop-program or start-program.

--
Jan-Henrik Haukeland
Mobil +47 97141255

smime.p7s
Description: S/MIME cryptographic signature