=== robbiew-afk is now known as robbiew [13:36] any way to debug why upstart complains about «start: Unknown job [foo]»? [13:39] Because there’s no /etc/init/foo.conf or /etc/init/foo.conf has a syntax error. [13:39] ion: ok, the foo.conf exists. is there any systax checker? [13:39] syntax* [13:40] See the error message in syslog. [13:40] ion: good! [13:41] bad [13:41] guess who can´t start... [13:41] ... rsyslog [14:01] hi all [14:01] I have a server that seems to get "stuck" on booting... runnig upstart with --verbose doesn't really help, sinec that allows me to see what tasks have been completed, but not what it's waiting for [14:02] with a serial init I could predict order, thus the last finished modules would implicitly tell me what next failed [14:02] this isn't true for parallel startup [14:02] can someone give me a hint as to effectively troubleshoot this issue [14:03] ion: so, if I don´t have syslog running, there are no logs? [14:05] http://upstart.ubuntu.com/wiki/Debugging [14:05] ok I'll try that :) [14:06] StucKman: add --verbose to your cmdline [14:06] wraiden: service mysql start --verbose? [14:06] no [14:06] start --verbose mysql? [14:06] no [14:06] kernel commandline [14:06] what>? [14:06] in your grub config [14:06] do I need to reboot my machine? [14:07] the args are appended to inits start [14:07] so init gets the --verbose [14:07] yougottabefuckingkiddingme [14:07] no, sorry, I can´t reboot this machine just like that [14:08] I don´t even have a monitor attached to it [14:08] I'm not even in the same state [14:08] hehe [14:09] not funny! [14:09] depends [14:09] you can try to add echos to every suspected job config [14:09] as pre-start [14:09] and let that echos write to a file [14:11] e.g. [14:11] pre-start script [14:11] echo "jobname trys to start" >> /debug.log [14:11] end script [14:11] and add a post-start with the same [14:12] wraiden: wait [14:12] oh [14:12] I'm trying to launch it by hand [14:12] poor mand debugging ... [14:13] *mans [14:13] whats your upstart version [14:13] ? [14:13] you mean that of the script has a syntax error it will get executed up to that point? [14:13] lemmesee... [14:13] 0.6.5-8 [14:14] (ubuntu lucid) [14:14] mhm [14:14] man initctl | grep check [14:14] ? [14:14] nothing [14:15] initctl has a check-config arg on 0.9 versions but i don't think that was in 0.6 also [14:15] chízus [14:15] can you post the config somewhere? [14:15] i'll take a look [14:16] ok [14:16] initctl list does not have either rsyslog or mysql in it [14:16] wraiden: it's ubuntu's provided job file [14:17] let me paste it anyways [14:17] taht means that the jobs are not loaded because of syntactical errors [14:17] huh? [14:17] the ubuntu provided ones don'T work? [14:17] wraiden: don't get me started on that [14:17] are you shure that the jobs are in /etc/init ? [14:18] ubuntu has not ported all init scripts over to native upstart jobs [14:18] it could be that they are in /etc/init.d [14:18] wraiden: it is doing it, and the wrong way [14:18] s/it is/they are/ [14:18] s/and/but/ [14:19] wraiden: http://pastebin.lugmen.org.ar/6691 [14:25] have a lucid box here [14:25] 10.04.2? check /etc/issue [14:25] 10.4.3 [14:25] ok [14:31] or, more interetingly, this other: http://pastebin.lugmen.org.ar/6692 [14:39] if I add or modify a script in /etc/init/, do I need to do something so upstart rereads it? [14:41] normaly not [14:41] and anormally? [14:42] as upstart looks with inotify for file alteration in its config folder and reloads [14:42] ack [14:42] you can poke init to reload with [14:43] initctl reload-configuration [14:43] wait [14:43] wait wait wait [14:43] I have a question [14:44] ok, here's te thing [14:44] I'm trying to start a rsyslog and/or mysql in a chroot [14:44] how does that afeect upstart? [14:45] affect* [14:48] where do you execute the initctl or links to it like start stop reload whatever ? [14:48] inside the chroot? [14:49] where are the job config files located? [14:49] inside the chroot? [14:49] upstart is pid 1 [14:50] for instance, if I do «initctl reload-configuration», it contacts PID 1, ... I see where's the problem [14:50] pid 1 is located on your real root [14:50] I know [14:50] it will only look inside of /etc/init [14:50] ok, let me (re)think about it [14:51] and yes, neither mysql or rsyslog are in the host [14:51] you ca however change the jobs to be started outside of a chroot an let the jobs start inside of the chroot [14:51] wraiden: uhm? [14:51] scripting? [14:52] yes, but what did you mean? [14:52] or even easier [14:52] man 5 init [14:52] look for the chroot stanza [14:53] wraiden: good [14:53] RTFM might help ;-P [14:53] wraiden: ups [14:53] didn't use the chroot stanza so far [14:54] so do I have to pray it works? [14:54] :? [14:54] but will soon as my inhouse cluster distro will change to native upstart over the next few weeks [14:54] I'm sorry, I'm just a little bit bitter [14:54] so i have to use it for bind [14:55] well. upstart is covered by a very strict testsuite [14:55] you can expect a stanze to do what was documented [14:56] *stanze [14:56] *stanza [14:56] ok, good [14:56] *lokking fatfingered* [14:56] i need some sleep *g* [14:56] go, go, go! [14:56] it daytime here [14:56] and i'm @ work [14:56] heeh, same here [14:57] not a good place for a powernap *g* [14:57] take a nap anyways [14:57] as network admin one would wake dragons that one should better not wake if someone finds me sleeping *g* [14:59] you can drop me a line here if the stanza got you where you wanted... [15:03] wraiden: I'm thinking if that's the best approach for my particular case [15:03] in general, it looks like yes, but... [15:11] "initctl log-priority debug" on a running system will do the same as --verbose on the kernel commandline [15:13] wraiden: ok, it was better your way [15:14] seems to work now [15:15] JanC: but where does it log them? [15:16] becuase rsyslog was not loading before [15:16] because* [15:17] right, maybe not so useful right now then (unless you start syslog manually first, or such) [15:17] mysql didn't start either? [15:18] JanC: I'm finishing fixing the rsyslog now [15:20] you don't have a circular dependency between rsyslog & mysql or such? ☺ [15:20] no [15:20] it's more complicated [15:20] so I have to do extra work to leave it fixed [15:21] we use cfengine and we switch from chroot to chrroot depending on development states [15:24] ah, using upstart in a chroot or in container-style VMs is a bit complicated sometimes [15:26] not anymore [15:26] chroot works fine [15:27] and I just have to add some stuff to our cfengine repo and modify some scripts that change the dev environ [17:15] hi.. I'm developing a daemon which is currently in early Alpha/Beta stage.. I'm using the upstart restart functionality to restart it if it crashes.. however, I'd also like to get a coredump, if possible.. [17:18] I've tried adding "limit core 100000 100000" to the conf, but I can't find the corefile.. I've also tried adding "chdir /tmp/" but no luck.. am I missing something? [23:23] any way for me to prefix an env var?