[01:51] is there a recommended method to completely disable a rule, persistent-net-generator under ubuntu in this case? [02:54] man, i am so far behind the times on this stuff, dang === jodh is now known as jodh_ [12:02] litheum: that sounds like an udev question, not an upstart one ☺ [12:02] and how to override system-supplied udev rules is explained in /etc/udev/rules.d/README [16:21] JanC: yeah i wasn't sure if it was a udev question or upstart, but i do want to disable the upstart rule that is calling udev and i don't want to do anything to udev, so... [16:23] i guess i didn't understand well enough how things work to want to copy the file and edit it [16:24] if i don't want it to run at all, is that really the right approach? [16:43] I guess creating an empty udev rule with the same name in /etc/udev/rules.d/ would work [16:44] yeah this does sound more like a udev question, doesn't it :) [16:44] how about a different question that might be related to upstart! [16:45] i edited /etc/init/hostname.conf so that it sets the hostname based on the name of the VM it is running inside of [16:45] but i need /etc/init.d/vboxadd to be executed first... is there a way to make the upstart job depend on an old-school init script? [16:47] edit the SysV init script to emit an event, then depend on that event? [16:47] yeah, that's what i ended up doing [16:48] that seems really scary though if that script is just going to get overwritten if i install a new version of those tools at some point [16:48] and vbox has no hooks for this? [16:50] hooks for what, running scripts after the module is loaded? [16:50] yes? [16:50] i have no idea if it does that or not, but tearing the hostname out of upstart and tying it to some vbox hooks didn't really even enter my mind [16:50] I mean, emitting an event from such a hook [16:51] oh, right [16:51] well, i don't know about that. good suggestions [17:55] hmmm... now that there are actually people here, maybe I should reepeat my question.. [17:55] 15:48:53 < jMCg> I'm looking for a better way to handle this: http://sprunge.us/IAiV [17:56] 15:49:10 < jMCg> Because the results are just plain horrible (IMO): http://sprunge.us/WSdc [18:01] jMCg: why are they horrible? (not sure what you are trying to do) [18:02] JanC: the point is that I need the logger only during the startup, but it stays up and running for the whole runtime. I find that... wrong. [18:02] eh [18:02] JanC: the problem is that expect daemon doesn't work. [18:04] so logger works as intended, right? ☺ [18:04] JanC: what doesn't work as expected is upstarts fork tracing :-S [18:04] Can it please just work automagickally? [18:05] I mean.. how does SMF do this? [18:05] AFAIK that's a feature that's in the works [18:05] I don't remember SMF ever once caring a. Oh. Right. [18:05] Contracts. [18:07] http://www.freebsd.org/cgi/man.cgi?query=contract&sektion=4&apropos=0&manpath=SunOS+5.10 [18:08] That's a property all Solaris processes have which doesn't exist on Linux. [18:14] jMCg: how would that fix the logger issue? [18:14] also, can't you use start-stop-daemon for example? [18:16] also, you can probably do some logging in pre-/post- scripts [18:16] JanC: the logging is only needed for startup -- and only in case of failure. [18:16] what does it need to log? [18:17] JanC: the reason why it's failing to start, for instance. [18:17] That kind of stuff is *reeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeaaally* useful for debugging. [18:17] how did you do that with SysV init? [18:18] JanC: I haven't used SysV init since there's upstart, and I haven't used this setup before. [18:19] jMCg: it sounds like you're looking for https://code.launchpad.net/~jamesodhunt/upstart/job-logging [18:19] broder: I doubt that will stop logging after startup? [18:19] so? [18:20] that's what he wants, if I understand correctly? [18:20] to log startup failures, but not whatever else the application spews after that? [18:21] it's still possible to get that information out of full job logging, though [18:21] once job logging gets merged in, i could imagine extending it to "log everything until the job transitions to a running state" [18:21] pretty easily, even [18:23] broder: what I want is to be able to start httpd without -D NO_DETACH without upstart losing track of it. [18:23] Oh. [18:23] he already gets that with logger now, although upstart job-logging would make it easier [18:23] expect daemon expects two forks, my fancy pants setup has 3. [18:27] jMCg: there is some work being done about that, but in the mean time your job should work just fine? [18:28] JanC: yup. [18:28] I mean, the thing you complained about has nothing to do with that ;) [18:28] complained about first [18:29] JanC: I'm not using expect daemon. I tried doing that, but it wouldn't work. Only now did I grasp why. [18:31] ah, yes, THAT is a known problem [18:31] IIRC there are several solutions for that planned ☺ [18:37] * jMCg strongly suggests implementing contracts in Linux. [18:44] jMCg: have any good link about that? ☺ [18:45] JanC: aside from the docs I linked above, there's the OpenSolaris.. well.. OpenIndianna source code.