[20:43] <rrva> is upstart really mature for servers?
[20:43] <rrva> I've had issues with upstart hanging when trying to stop jobs if the process died
[20:43] <rrva> or upstart thinking the process was running when its really not
[20:55] <ion> Usually that symptom is the result of lying to Upstart about the forking behavior of a main process, i.e. an inappropriate “expect” value.
[20:56] <ion> http://heh.fi/tmp/workaround-upstart-snafu to work around that.
[21:01] <rrva> how to recover upstart state if I misconfigured this
[21:01] <rrva> only reboot would help?
[21:05] <rrva> ion: upstart should auto-heal from such lies. This is all too common
[21:05] <rrva> ion: thanks for the script
[21:17] <ion> Yes, a robust implementation of fork following is in TODO and a working prototype exists. But for the moment, “expect” should only be used if you’re completely sure of the behavior of the main process wrt. forking. By doing that you should find Upstart rock solid.
[21:18] <ion> Note that Upstart never crashed, it only lost track of a misconfigured job.
[21:49] <rrva> yes.. please consider "auto-heal losing track of process" code