[11:13] <tcr> If you have a post-start script which is blocking (e.g. polling for a non-existing resource), and you stop the job when it's in "start/post-start" (with two processes associated to it, the main process, and the post-start process), the job will change into "stop/post-start" with *both* processes still running.
[11:13] <tcr> Similar with a pre-start script.
[11:14] <tcr> I find this behaviour surprising. I'd have expected the post-start process along the main process to get killed in this situation.
[11:14] <tcr> (Upstart 0.6.5)
[11:34] <tcr> I wonder if there's a detailed state diagram somewhere, along the lines of the one http://upstart.ubuntu.com/wiki/JobStates but that has its edges labelled with the conditions to trigger that particular state transition