Re: Lame duck state monitoring

[Martin Pala]

You can check remote host using "check host" statement. If the server that can 
enter lame duck state talks HTTP, it is most probably related to specific HTTP 
response code ... you can check for this response code with the "status" option 
of http protocol test. See more details in monit manual: 
https://mmonit.com/monit/documentation/monit.html#HTTP

Best regards,
Martin



> On 18 Sep 2016, at 00:37, jul <address@hidden> wrote:
> 
> Hello monit list
> 
> I recently read Google SRE Book and there is a part on lame duck state.
> Their definition: "the backend task is listening on its port and can serve, 
> but is explicitely asking clients to stop sending requests"
> 
> 
> Basically if a server is an overload state (cpu, memory, threads or 
> whatever), it is put out of the queue from the load balancer for x time.
> Can be the same also just after a start or a reboot for service which need 
> time to be performant (some caching)
> 
> Any monit rules to do that?
> for example to get out of haproxy load balancer or similar
> Or doing reverse, do a check remote host from load balancer ?    
> 
> 
> Thanks
> 
> Jul
> 
> --
> To unsubscribe:
> https://lists.nongnu.org/mailman/listinfo/monit-general