Hi,
I'm fairly new with monit and docker/coreos but I'm doing some experiments trying to integrate the 2 of them as I believe would be the best workflow. Forgive my long explanation, feel free to jump to the last paragraph for the specific issue/question.
My idea is simple, have a container running monit that monitors "stuff" declared by other containers. Essentially I would like to "docker run someimage" that would have meta data declaring what monit should monitor, and how. I've done some tests and so far I managed to get a basic container that is able to listed to docker events on the host and start/stop containers at will.
My next step would be to orchestrate these events and read the meta data to generate proper monit configuration files. The problem I'm facing is that in my rudimentary tests, each time I invoke "service monit reload" the cycles aren't properly managed by monit. For instance, if I configure a basic monitor for an apache process and then reload monit a few times in a row it will trigger the alerts and unmonitor the process as if it had failed.
So, is it possible to have monit persist the correct cycle count/state after a reload? It would be nice if monit could remember the existing conf files, and just reload the modified ones without affecting the cycle count on the unmodified ones. As this is for a docker integration this problem would be relevant because any number of containers could be starting/stopping all the time.
--
Alexandre Gomes
http://www.alexandre-gomes.com