| tjaalton | when initctl list foo shows "foo stop/killed, process 12345" starting the daemon again just hangs | 14:19 |
|---|---|---|
| tjaalton | any idea to fix that? | 14:19 |
| tjaalton | without a reboot | 14:19 |
| tjaalton | stop just says it's stopped already | 14:20 |
| tjaalton | using saucy btw | 14:22 |
| tjaalton | i've had this happen a few times with sssd being the daemon, and when it happens shutdown hangs as well, need to force poweroff the machine | 14:23 |
| tjaalton | so sounds like a bug in upstart | 14:23 |
| tjaalton | anyone? | 14:28 |
| stgraber | tjaalton: hi | 14:32 |
| stgraber | tjaalton: so I'm assuming you're talking about sssd right? :) | 14:32 |
| stgraber | tjaalton: it gets into that state if the daemon fails (crashes) before it's done forking | 14:33 |
| stgraber | tjaalton: in such a case, upstart hasn't figure the final PID for the process and so is tracking the wrong PID | 14:33 |
| tjaalton | stgraber: yep | 14:33 |
| tjaalton | but it can't clean up it's state? | 14:33 |
| stgraber | tjaalton: unfortunately in such case, only a reboot can fix upstart's state (unless xnox or jodh know of a way around) | 14:34 |
| tjaalton | okay | 14:35 |
| tjaalton | fairly easy to get in that state | 15:03 |
| tjaalton | like if there's a valid config that won't load because some other bits are missing | 15:04 |
| stgraber | tjaalton: it's probably best to run with -i then and drop the expect fork | 15:46 |
| stgraber | that way upstart won't use ptrace, so it should be way more reliable and also slightly faster (saves an extra fork and saves the use of ptrace) | 15:47 |
| tjaalton | stgraber: interesting, i'll give it a go | 15:48 |
| tjaalton | stgraber: i didn't get it to work | 17:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!