=== sadmac_ is now known as sadmac2 [09:05] i created the following upstart conf file http://pastebin.com/2KCd7tNq , run.sh actually runs a java application, while the job starts fine, stop freezes, any idea how to fix it ? [18:35] Keybuk: https://bugzilla.redhat.com/show_bug.cgi?id=577709#c5 <- opinions? [18:36] notting: that sounds like the old init it still running? [18:36] Keybuk: right, it was the equivalent of apt-get dist-upgrade (or whatever) from a 0.3.x system to one that contains 0.6.x [18:37] in that dist-upgrade, we restart init [18:37] Keybuk: which promptly loses track of all its jobs? [18:37] when we were using 0.3, that was just gettys and stuff, which wasn't an issue [18:40] hm, would have to run some tests to see how practical that is [19:02] notting: it could work. we might not kill some things like X on shutdown but nothing major. [19:14] gdm is what handles killing X [19:15] of course upstart is what handles killing gdm [19:15] halfline: precisely [19:15] but there's that kill -TERM -1 or whatever anyway [19:17] I'm trying out ubuntu lucid beta 1, and upstart doesn't seem to be firing scripts with "start on startup" ...has anyone had success with this? [19:18] caolanm: it probably is, but whatever those scripts need isn't ready yet [19:18] Keybuk: how are those dependencies defined? all the script says is start on startup [19:19] (I'm working from this guide: http://howtonode.org/deploying-node-upstart-monit) [19:19] I tried changing it to "start on (local-filesystems and net-device-up IFACE=eth0)" [19:20] which seems to fire, but I get strange errors regarding missing files [19:31] caolanm: you should talk to the maintainers of the software you're trying to update === sadmac2_ is now known as sadmac [19:46] Keybuk: RE: Is this a mountpoint?, why not iterate mtab/proc/wherever and get a list of em? [19:46] do you realise how many corner cases that has? :) [19:47] Keybuk: that + realpath should be == profit [19:47] what about if you've got a full filesystem [19:47] then you recursively bind mount the /real directory onto / [19:49] not sure, but that's enough to make me want to not think about it anymore. Your way is fine. :) [19:49] Keybuk: also did you see my last upstart blogging? [19:49] no? [19:49] Keybuk: http://screwyouenterpriseedition.blogspot.com/2010/04/upstart-service-mangement-modeled-as-bi.html [19:53] sadmac: the funny thing is, that blog post makes about as much sense to me as most of your first draft proposals :p [19:53] Keybuk: I impersonate myself well. [19:55] Keybuk: also, I would assert that that blog post contains a lifetime supply of release codenames. [20:37] sadmac: 577709 updated with brief testing [21:06] question: can i make a upstart script dependent on the start of an oldschool init.d script? say, with "start on SYSVINITNAME"? [21:14] dvrcoder: you can put an "initctl emit somefunkything" at the end of your sysvinit script, then do start on somefunkything. You could also instrument /etc/rc in a similar way to get such events for everything (which I believe fedora is doing now) [21:17] sadmac: thanks. i'll try and pray it gets my job done :D [21:38] notting: comitted and tagged. seem to be having some trouble with koji and/or ssl. wanna kick em off? [21:38] sadmac: well, i was trolling for comments on the idea [21:39] notting: among the horrible solutions thats the one that will cause the least screaming. [21:40] notting: and the good solution might be ready for F-14 [21:40] actually it won't be applicable for F-14 [22:21] to create my own upstart script, is it enough to write a blabla.conf and put it in /etc/init, or do i have to register it somewhere additionally? [23:16] dvrcoder: ys