[10:29] Hey, would you look at that. [10:30] Ok, got a quick question for you; I can't find any built in way of doing it, so, are you supposed to change directory and user in the pre-start script, or in the actual start-script? [10:33] plundra: what do you want to do? [10:33] Launch an application from a certain directory and with another user. [10:34] So (cd /path ; sudo -u foo ./crap) or whatever :) [10:34] Maybe using su instead. [10:39] do it within a script .. end script [10:40] and run su via exec [10:40] Ok, so pre-start script, right? [10:40] no script ... end script [10:40] Ok, but why not start it within the script too? [10:41] http://pastebin.ca/1276763 [10:42] Ah [10:44] Cool, seem to work just fine, respawn and everything :) [10:44] do you want to monitor the application? [10:45] ah, you seem to have figured that out already [10:45] Yeah. [10:45] (But I do think a man-page would be a good thing... :-P Not just for initctl/start/stop/whatnot) [10:46] plundra: there is a wiki [10:46] That really bugs me, more often then not, random linux-projects are awesome, but the lack of documentation makes it unusable. [10:46] so if you want to contribute [10:46] that is a great start [10:47] Wikis are for faqs and examples, you shouldn't have to use them for basic configuration-syntax etc. :) [10:47] But that's just me, and I'm to lazy to contribute right now so I'll stop complaining :-D Just wanted to say it. [10:48] Also, lunch! And thanks for your help. [10:48] plundra: yeah, the man pages could definitely be improved [10:48] but take a look at the home page http://upstart.ubuntu.com/ [10:48] especially scotts blog posts (linked on the front page) and the wiki [11:52] After respawning too fast, will it try again later or do I manually have to start it again? [11:55] Doesn't seem like it :-) So, is it possible to have a script runt WHEN it's stopped completly due to respawning too fast? [11:56] For informing/whatever that it's the case. [11:56] I know a respawn will emit a start/stop, and nothing special, but is there something that is emitted on respawning too fast? [11:57] plundra: not sure, you can check with "initctl events" what events get emitted by upstart [11:59] Ah, great :-) "stopped xxxx failed respawn" [12:09] Hmm, can't get my "on xxxx/failed" (with and without respawn afterwards) to work. [12:46] Ah! [12:47] "on stopped xxxx failed" did the trick :) [12:48] All though, I'd rather have that kind of stuff inside the xxxx-file it self. [12:49] But on the other hand, now I can create a generic script I guess. That's probably better \o/ I bet you can get 'xxxx' as a parameter or variable somewhere 8-) *searches the wiki* [12:57] Maybe not. I think I'm getting a UPSTART_EVENT=stopped, and nothing else. [17:57] Hey all [17:57] How's upstart coming along? [17:58] on your webpage, I've noticed that dbus communication is now working? [17:58] Does this mean that I can find out, as a normal user, what services are running? [18:25] johnflux: yes [18:26] needs some tweaking, though [20:28] So... I'd like to migrate from runit to upstart, but I'd like to retain the benefits of easy logging of stdout/stderr.. is there anything nice in upstart for that? [20:29] i mean, i can make my logging library use syslog, but 1) syslog sucks 2) i'd like to separate the logs from different daemons better 3) i want to capture stdout/stderr to see bugs etc [20:31] Tv: do the syslog migration and use a better system logger :) [20:31] Tv: rsyslog seems pretty powerful [20:31] sadmac: doesn't fix the "something writes to stderr" part of the equation [20:32] Tv: the problem we've had with that is that if whatever application which is reading/marshalling the stdout/stderr data gets killed, it takes all the applications it was watching with it (massive SIGPIPE) [20:33] sadmac: yeah that's why daemontools/runit puts a more reliable process in between [20:34] Tv: its not reliability per se. It could be a deliberate kill, such as during shutdown. [20:34] restarting the logger part is perfectly normal for the runsv/svlogd combo [20:35] yes, but that's a different kind of connection [20:36] well yes and it's the thing keeping me using runit [20:36] about twice a year i try to migrate away, and there's just nothing better.. [20:37] Tv: then keep using runit :) [20:37] * sadmac has no desire to dominate the planet [20:37] i wanted to have instance jobs [20:37] oh well [20:38] its a problem we've looked at a lot, and for what upstart does, none of the solutions we've found are quite right. But we'll get there