[12:29] <vladanian> I've got an upstart question -- I'd been running it on edgy for a few weeks and it was working fine, but for the past couple weeks I've gotten hung up at boot on, "upstart: Error reading control message: Invalid message received."  Anyone encounter this?
[12:30] <johnnybuoy> why is upstart good for me?
[12:32] <johnnybuoy> hmm
[12:45] <vladanian> In case anyone's interested, I've filed a bug: https://launchpad.net/products/upstart/+bug/61322
[12:46] <johnnybuoy> this is kewl! init IS a bit old
[12:47] <johnnybuoy> the problem: so is *nix :(
[12:47] <johnnybuoy> the whole net thing in *nix is a hack
[12:47] <johnnybuoy> prtty good hack, but still a hack
[12:47] <johnnybuoy> but GO upstart
[12:59] <mjg59> 23:47 < johnnybuoy> the whole net thing in *nix is a hack
[12:59] <mjg59> ?
[01:00] <johnnybuoy> hmm
[01:00] <johnnybuoy> makers of UNIX admitted this, I fell on it when exploring a novel OS from them (the authors of UNIX): plan9
[01:01] <johnnybuoy> UNIX was not designed with networks in mind, networks came later
[01:03] <mjg59> I think it's hard to say it's a hack, given how well it works
[01:04] <johnnybuoy> good hack
[01:04] <johnnybuoy> but hack
[01:04] <johnnybuoy> it *does* work well
[01:05] <johnnybuoy> but it wrks beter in plan9 :)
[01:05] <mjg59> In what ways?
[01:11] <johnnybuoy> well, for egzample it only has one protocol for everything, but I'd think #plan9 would be a better place for this discussion
[12:29] <che> the topic isnt accurate anymore... 0.2.7 is up
[02:14] <johnnybuoy> hello! how do I know if a service is started or not? with upstart I don't get the scrolling stuff when starting the pc.
[02:14] <_ion> Upstart jobs: status jobname
[02:15] <_ion> init.d scripts: /var/log/boot
[02:15] <johnnybuoy> ok, thanks! Oh and great job all! my pc starts a *lot* faster now with upstart
[02:29] <mbiebl> johnnybuoy: probably a placebo effect. atm upstart should be appox. the same speed as sysvinit.
[02:30] <johnnybuoy> :D
[02:30] <johnnybuoy> okok
[02:30] <mbiebl> None of the init scripts was converted to upstart jobs yet, the boot procedure is basically the same (maybe the missing output makes it abit faster)
[02:30] <johnnybuoy> tha problem is I do status fbsplash, and it doesn't know
[02:31] <mbiebl> If you want comparable bootup times, install bootchart+bootchart-view
[02:32] <johnnybuoy> what is that?
[02:33] <mbiebl> bootchart - Boot process performance analyser
[02:33] <johnnybuoy> ah ok.
[02:33] <johnnybuoy> but now I don't have init...
[02:33] <johnnybuoy> I can't compare :)
[02:34] <johnnybuoy> I think it must be a little faster...
[02:34] <mbiebl> If you are happy with upstart, that's all that matters ;-)
[02:34] <johnnybuoy> or maybe that upgrade last night ..
[02:34] <johnnybuoy> :)
[02:34] <johnnybuoy> yes, very.
[02:34] <mbiebl> Md: ping
[02:34] <johnnybuoy> I'd like to know what jobs are started tho
[02:34] <johnnybuoy> status don't work
[02:35] <mbiebl> initctl list
[02:35] <johnnybuoy> ahh
[02:35] <johnnybuoy> maybe
[02:36] <johnnybuoy> I don't get it
[02:36] <johnnybuoy> this is not it
[02:37] <mbiebl> You probably want to read /var/log/boot or remove "quiet" from the boot options.
[02:37] <johnnybuoy> ah ok!
[02:38] <johnnybuoy> that might be it
[02:38] <Md> mbiebl: pong
[02:38] <johnnybuoy> /var/log/boot: (Nothing has been logged yet.)
[02:38] <johnnybuoy> :(
[02:39] <mbiebl> Got a problem with one of your udev rules in /etc/udev/udev.rules
[02:39] <mbiebl> Specifically the dm rule that prevents /dev/dm-X from being created.
[02:39] <mbiebl> This breaks gnome-mount when used with luks encrypted partitions.
[02:40] <mbiebl> johnnybuoy: Install upstart-logd
[02:40] <Md> mbiebl: discuss this with the maintainers involved in dm. feel free to Cc me
[02:40] <johnnybuoy> it's doing it right now, i think
[02:40] <johnnybuoy> ok, thanks
[02:41] <mbiebl> I talked to the hal dev (davidz and kay) and the said there is no good reason not to have /dev/dm-x
[02:42] <mbiebl> I asked you, because /etc/udev/udev.rules is part of the udev package.
[02:42] <Md> mbiebl: yes, but I do not know how dm works. I am sure that I added that rules for a reason (and every other distribution did too), so I will not remove it unless somebody can explain with authority why it's not needed anymore
[02:43] <mbiebl> Redhat does not.
[02:44] <Md> ok. still, I will not change the package in a way I do not understand so please look for a rationale
[02:45] <mbiebl> The rational is, that cryptsetup itself does not need /dev/dm-*
[02:45] <mbiebl> So it was removed with this udev rules.
[02:45] <mbiebl> But 3rd party apps, like gnome-mount, rely on /dev/dm-*.
[02:46] <Md> I think there was much more than cryptsetup
[02:46] <Md> are you really positively sure that having udev create the devices will not break other dm components?
[02:47] <mbiebl> According to kay and davidz it should not hurt (otherwise redhat would be busted anyway)
[02:47] <mbiebl> I can ask the dm maintainer first if you prefer that.
[02:47] <Md> ok, I will change this in the next upload today or tomorrow
[02:47] <Md> yes, please ask him as well. just to be sure
[02:47] <mbiebl> Ok, will do.
[02:48] <Md> I added that rule may 30. I think there was a reason, but I really cannot remember it
[08:23] <_ion> keybuk: grep respawn /etc/event.d/logd
[08:25] <Keybuk> the // thing?
[08:26] <_ion> Yep. :-) Not that it really matters.
[08:28] <Keybuk> yeah, common side-effect of --exec-prefix=/
[10:02] <maro_> hi :)
[10:03] <maro_> do you know if upstart is supposed to completely replace sysvinit, or is it a long-term solution to depend on sysvinit-utils?
[10:03] <Keybuk> maro: the things in sysvutils/sysvinit-utils are things that aren't really part of init
[10:03] <Keybuk> so there's no particular reason for upstart to provide its own implementation of them
[10:04] <maro_> hm, true :)
[10:06] <maro_> are there examples of no-rc-compat setups yet?
[10:08] <Keybuk> not yet
[10:09] <Keybuk> we're doing the "release early and release often" thing, we're still at "early" :p
[10:09] <Keybuk> the idea of doing the rc-compat stuff is that it proves the daemon works, while still letting us revert to sysvinit at a moment's notice if it doesn't
[10:09] <maro_> ok, perhaps I can be the first then ;)
[10:09] <Keybuk> go for it :)
[10:10] <maro_> already did a full minit migration once, still have a graph somewhere
[10:12] <maro_> minit is one-way depedency based though, so this will probably be a bit harder... anyway: http://borkware.net/~mark/minit.png :)
[10:16] <Keybuk> heh, cute
[10:19] <maro_> doesn't look that complex on a graph, but imagine every service being a directory containing a symlink to the executable, a file for parameters, optional (empty) "respawn" and "sync" files...
[10:20] <maro_> running 'tree' in there gives me 63 directories and 211 files ;)
[10:20] <Keybuk> yeah, djb-style config sucks :p
[10:21] <maro_> djb-style does in general ;)
[10:21] <Keybuk> *shrug* I still use qmail
[10:24] <maro_> I wonder how hard it will be to make scripts for upstart ala the one I did to draw the graph
[10:25] <maro_> (100 lines of sh)
[10:25] <Keybuk> quite easy I should think
[10:25] <maro_> we'll see :)
[10:26] <maro_> anyway, need to get some sleep - sleep tight when the moon hits your timezone :)
[10:29] <maro_> ah, just saw your reply on the list - sounds good :) - off now
[10:31] <Keybuk> I tend to sleep when the sun is up :p