in_monitor_agent: Start one HTTP server per worker on sequential port numbers

[mururu]

MonitorAgentInput doesn't work on multi worker environment because workers
try to start servers with the same port.

[repeatedly]

Good catch.
BTW, how about adding launch_server like option to enable/disable server launching?
It allows users can use in_monitor_agent to emit buffer parameters in multi-process environment without this error.

[mururu]

@repeatedly Sounds good, but there is an another idea (@tagomotis sugestted) that each worker launch a server on different port(port + worker_id) on multi worker environment.

<system>
  worker 4
<system>
<source>
  @type monitor_agent
  port 24220
</source>

Using this conf, workers launche servers with port number 24220, 24221, 24222 and 24223.

I think that it is better for us to be able to use same feature on single worker environment and multi worker environment if possible.
How about that?

[repeatedly]

It is also good.

[mururu]

Thanks, I'll go with it.

[mururu]

Updated.

[repeatedly]

Thanks!