Re: Connection monitoring bug?

[Weedy]

On Tue, Apr 15, 2014 at 6:13 AM, hzwaal <address@hidden> wrote:

I'm monitoring the liveness of one of my processes through its HTTP
connection. I have the following relevant settings:

set daemon 2 # check every 2 seconds (a short period is required in my case)

check process soandso with pidfile /such/and/such
  start program = ...
  stop program = ...
  if failed url http://localhost/ping
    with timeout 5 seconds for 5 cycles
    then restart

Now, when my process gets stuck (it shouldn't, but I'm simulating this, just
in case) it will be restarted. This occurs 5 second after the 5th poll,
which is when it is not responding for 13 seconds (4 * 2 + 5). So far so
good.

However, when my process is not back online within 2 seconds, it will be
restarted again, and again, ad infinitum. This seems like a bug to me.

Ignoring the fact that 2 second poll seems silly use the timeout parameter. Find out how long it takes the stop init script to run and then double it.

check process soandso with pidfile /such/and/such
  start program = ...
  stop program = ... with timeout 10 seconds