[11:35] <MarcWeber> Is there some logging support again or do I have to use script\n &> /var/log/foo \n end script ?
[12:20] <keesj> in scripts?
[12:22] <keesj> in 0.5.x upstart self tries to use syslog, for scripts it should be possible to lot to syslog but I have not seen that work
[12:23] <keesj> the "easy" thing to do IMHO is to use the logger command
[13:27] <MarcWeber> keesj: It's about reading error messages from startup scripts etc.
[13:28] <MarcWeber> upstart-0.3 had its own (bitrotted) daemon to log stdout and stderr of the scripts into a file
[13:29] <keesj> MarcWeber: how is this daemon called it that hte logd?
[13:30] <MarcWeber> It was called logd
[13:30] <MarcWeber> However using script &> file works fine in practise..
[13:33] <keesj> Yes, , what I sometime do is a ( echo bla ; do this; echo bla ) | while read line ;do  echo "SCRIPT_NAME:" $line ; done
[16:53] <geiseri_> hi, i have a even  that needs to wait for hal to be loaded, is there a way with the current upstart in Ubuntu intrepid to do this?
[17:18] <pkt> perhaps you mean a service that needs to wait for hal ?
[17:19] <geiseri_> well i want to start Xorg on boot for a live CD, but i guess the new Xorg wants to use hal to get input devices.
[17:19] <geiseri_> i had an upstart event script in event.d in hardy, but when i upgraded to intrepid it stopped working
[17:20] <geiseri_> diagnosis told me that it needed hal... so that is where i am stuck
[19:07] <MarcWeber> I'm trying to add upstart-0.5.1 support to nixos now. init ist started then nothing happens. Probably the job files aren't found? Is there a simpler way to debug this than patching upstart to support running as pid2 and running strace?
[19:13] <sadmac> MarcWeber: pass init=/sbin/init -v on the kernel command line
[19:14] <damjan> Anybody here? I'm installing Upstart in ArchLinux, and on startup I get a "unable to set oom adjustment" so the first job is not started
[19:15] <sadmac> damjan: your initrd needs to mount /proc
[19:15] <damjan> it seems to me that upstart tries to write to /proc/xxx/oom_adj .. but /proc is not mounted when upstart 
[19:15] <MarcWeber> sadmac :-) After starting udev before upstart I get some output as well now :-)
[19:15] <damjan> sadmac: I'm not using a initrd
[19:15] <MarcWeber> damjan: Mount proc before running upstart.
[19:16] <sadmac> damjan: then it won't work right now (we don't like it either, it'll be fixed at some point. patches welcome)
[19:16] <MarcWeber> damjan: Also make sure that your kernel supports it..
[19:16] <MarcWeber> if it doesn't you can still patch upstart.
[19:16] <damjan> MarcWeber: supports what?
[19:16] <MarcWeber> oom
[19:17] <damjan> it should, I see /proc/1/oom_adj
[19:17] <damjan> why is that needed at all? how about completelly disableing that part of the code?
[19:17] <sadmac> damjan: go for it
[19:18] <damjan> what's the "But.." ? :)
[19:19] <sadmac> you don't get OOM support, and there's no flag, so you're doing it on your own. In a text editor.
[19:19] <geiseri_> what are the files called in the /etc/events.d directory?  are those events themselves?
[19:19] <sadmac> geiseri_: those are jobs
[19:19] <sadmac> geiseri_: its a poorly named folder
[19:19] <geiseri_> okay, no problem
[19:19] <geiseri_> so is there a list of events that i can start on and stop on somewhere?
[19:20] <MarcWeber> geiseri_: initctl list will show them all
[19:20] <sadmac> MarcWeber: those are jobs
[19:20] <MarcWeber> oh. I've missed that. sry
[19:20] <sadmac> geiseri_: there's an events manpage in fedora. think its on die.net too
[19:20] <damjan> well ... ok, going to reboot to try this :)
[19:20] <damjan> brb
[19:20] <geiseri_> okay, ill check there
[19:20] <geiseri_> im trying to start a job when hal is finished starting
[19:21] <geiseri_> if that is even possible right now
[19:21] <sadmac> start on started hal
[19:21]  * geiseri_ checks what he had
[19:22] <geiseri_> yeah i have that but it doesnt seem to be working
[19:22] <geiseri_> im on ubuntu intrepid, is this not supported in that version?
[19:22] <sadmac> are you starting on something else?
[19:24] <MarcWeber> geiseri_: You can log events when jobs are started..
[19:24] <geiseri_> im sorry im not parsing your question...  my line is "start of started hal" in my job in /etc/event.d/Xorg
[19:24] <sadmac> geiseri_: on, not of
[19:25] <geiseri_> MarcWeber: that is the --debug flag you can pass to the boot args?
[19:26] <geiseri_> sadmac: im trying to launch Xorg right after hal is started successfuly 
[19:26] <sadmac> geiseri_: then start on started hal
[19:26] <MarcWeber> geiseri_: Try grep init /var/log/messages. Maybe it is all there..
[19:26] <sadmac> geiseri_: not of started, on started
[19:27] <geiseri_> sadmac: sorry i had a typo, that is what i have now... 
[19:27] <sadmac> geiseri_: is the job that starts hal in fact named hal?
[19:27] <geiseri_> that im not sure... how would i find that out for certain?
[19:28] <sadmac> ls /etc/events.d
[19:28] <geiseri_> ah, i dont have one in there, it seems ubuntu still uses the sysv init compat stuff
[19:28] <geiseri_> so im assuming ill never get an event from there
[19:29] <sadmac> probably not. Fedora has some kinda chintzy event emulation for that stuff, but I don't think ubuntu followed on
[19:29] <geiseri_> ah okay
[19:29] <geiseri_> so if i put an initctl event hal in the end of my rc.local, that might do it?
[19:30] <sadmac> initctl emint
[19:30] <sadmac> *emit
[19:30] <sadmac> yes, that'd do it
[19:30] <geiseri_> ah, right
[19:30] <geiseri_> okay, let me give that a whirl
[20:41] <geiseri_> sadmac okay that seems to have worked, thanks
[20:41] <sadmac> cool
[20:47] <MarcWeber> Can I syntax check job files somehow?
[20:47] <geiseri_> sadmac: hrm... only one problem when i do the emit the rc.local blocks... should i be doing "initctl emit xorg &" so it starts in the background?
[20:48] <sadmac> geiseri_: initctl emit --no-wait I think was the option
[20:48] <geiseri_> ah
[20:48] <geiseri_> okay
[20:48]  * geiseri_ checks the man page
[20:50] <geiseri_> that looks like it might do it
[21:01] <MarcWeber> Can the default PATH be set to a sensible default?
[21:02] <MarcWeber> Or do I have to use env PATH=... in all job files?
[21:03] <sadmac_> define sensible?
[21:03] <MarcWeber> containing corutils such as touch and upstart itself..
[21:04] <sadmac_> don't think we have global environment. Might be a good RFE
[21:04] <MarcWeber> I don't mind adding env PATH=.. everywhere either
[21:04] <MarcWeber> It does'nt matter that much because nixos does already have some kind of abstraction. adding this myself is done very fast.
[21:05] <MarcWeber> I think this differs from upstart-0.3 somehow