"check program status" shows "Initializing" status for a longer time

[Ulrich Windl]

I had defined a “check program”, testing for status == 2, also depending on a “check network”.

After a monit reload the monit status shows both, “status” and “monitoring status” as “Initializing” for a longer time, while all other checks have an “OK” status.

Execution of the program outside of monit takes about 0.2s to complete, and the exit status is zero.

It seems “check program” needs an extra polling cycle at least.

[Tildeslash]

Yes, it is given by the asynchronous implementation of the ‘check program’ along with the current test scheduler design and described in the manual: https://mmonit.com/monit/documentation/monit.html#PROGRAM-STATUS-TEST

We’ll modify it in the future (monit will collect the result immediately on program exit)

[Ulrich Windl]

So it seems I stopped reading to early

As I read it, any child events (`SIGCHLD`) are postponed until the next polling cycles.

It seems to me that updating the internal status area asynchronously might make sense (while actual processing of it could be still delayed until either next polling cycle or any summary or status request). The advantage of immediate processing would be a reduction of zombies and a good estimate how long execution actually took.

The problem I had describes is more evident if the default polling interval is being replaced by some longer interval.