[18:34] Any Ubuntu user here? Fedora developer here wondering about how runlevels on Ubuntu behave. [18:34] On Fedora we have runlevels like: [18:34] 1 - single user mode [18:35] 3 - network with text login [18:35] 5 - network with X [18:35] 6 - reboot [18:35] We can use "init X" where X is a number to switch between them [18:35] Does Ubuntu behave anything like this? [18:48] warren: Ubuntu's runlevels are the same as Debian's [18:48] we have S and 0 thru 6 [18:48] where 0 and 6 are special (halt and reboot) [18:48] 1 is used for switching to S (single user) [18:48] 2 thru 5 are up to the user to define [18:48] with 2 as the default [18:58] Keybuk: do you understand the /etc/event.d/ file format? [18:58] sometimes [18:59] for example: [18:59] /etc/event.d/rc3: [18:59] start on runlevel 3 [18:59] stop on runlevel [18:59] right...? [19:00] what does it mean if you have no number on "stop on runlevel"? [19:00] runlevel is an event [19:00] 3 is an argument to that event [19:00] the runlevel event means that the runlevel has changed [19:00] it gets the new runlevel as an argument [19:00] so "on runlevel 3" means "when the runlevel is changed to 3" [19:00] "on runlevel" just means "when the runlevel is changed" [19:01] (ie. to any runlevel) [19:01] hm [19:01] stuff in /etc/event.d are events as well? [19:01] no [19:02] they are jobs [19:02] another event.d file has: [19:02] start on stopped rc5 [19:02] where are the "start on foo" things documented? [19:03] right [19:03] stopped is the event, it means that a job has stopped [19:03] it gets the job name as an argument [19:03] so that means start this job when the rc5 job is stopped [19:03] the rc5 job probably runs the /etc/rc5.d (or /etc/init.d/rc5.d on "strange distros" :p) directory [19:04] the events are limitless [19:04] start on bangbangwibblewheresmypopsicle [19:04] is valid [19:04] something just needs to call initctl emit bangbangwibblewheresmypopsicle somewhere [19:05] oh crap. I have to work on a bigger bug. brb. [19:31] ok, back now. [21:50] Keybuk: is there a way to emit runlevel 2 apart from using the sysv compat runlevel command? [21:50] like "initctl emit runlevel 2"? [21:50] are events allowed to have spaces like that? [22:17] brendan_: exactly like that [22:18] Keybuk: great! thanks [22:18] there's no space [22:18] you're separating an event name and its argument there [22:19] Keybuk: our /etc/event.d/rc5 http://fpaste.org/paste/1656 [22:20] Keybuk: it seems that we're setting runlevel 5 in this job [22:20] warren: looks the same as the demo one? [22:20] ok... [22:20] i'm just trying to better understand what's going on [22:20] "runlevel --set" stores the runlevel in utmp [22:20] Keybuk: we're trying to deal with a few upstart bugs [22:20] Keybuk: like ... during shutdown you don't see anything [22:21] do you have "console output" in the rc6 job? [22:22] yes [22:22] are you running anything that might change /dev/console ? [22:22] change in what way? [22:22] e.g. usplash, X, rhgb, bootlogd, etc. ? [22:22] redirect [22:22] a common issue is if you run shutdown -h now while under X, the output will go to VT 7 [22:22] not that I'm aware of [22:23] so if I switch to VT7 during shutdown I should be able to chdeck [22:23] check [22:23] right [22:23] ah, you're right! [22:23] redirected to VT7 [22:25] Keybuk: here are all of our reported upstart bugs [22:25] Keybuk: http://tinyurl.com/4w8cdr [22:26] Any ideas about these two? https://bugzilla.redhat.com/show_bug.cgi?id=441339 and https://bugzilla.redhat.com/show_bug.cgi?id=438444 [23:01] back [23:01] sorry, was taking an exam there ;) [23:02] warren: 441339: is that not the case with sysvinit? [23:03] 438444: that may be a difference in the way that runlevels work in Fedora and Ubuntu? [23:03] in Ubuntu, rc 1 will run /etc/rc1.d which contains a K script for everything in rc2-5 [23:03] and then finally will run telinit S to switch to S (single-user-mode) [23:03] maybe Fedora has those two completely backwards?