[01:37] <btm> 'status JOB' exists with 0 when the job is running and when it isn't. is there a way to tell from an exit code if a job is running with upstart (since starting now emits 1 if it is already running)?
[19:42] <dchen_> is there a way to reset the upstart respawn counter?
[19:42] <dchen_> with inittab, the init q command does this
[19:52] <Keybuk> dchen_: same
[21:13] <dchen_> Keybuk: init q didn't seem to do it
[21:13] <dchen_> I just waited for the timeout
[21:18] <Keybuk> oh, no, it's running
[21:20] <dchen_> pardon?
[21:27] <Keybuk> you need to stop and start the job
[21:27] <Keybuk> (with the telinit in between)
[21:29] <dchen_> hrm
[21:29] <dchen_> so I tried
[21:29] <dchen_> initctl stop blah
[21:29] <dchen_> init q
[21:30] <dchen_> initctl start blah
[21:30] <dchen_> and it had an error message that it was restarting too fast or something
[21:31] <sadmac> Keybuk: you said ubuntu doesn't re-exec init on every glibc change, but instead does it just before unmounting the FS?
[21:33] <asalkeld> hi, has anyone tried to cross compile upstart 0.63 yet?
[21:34] <sadmac> asalkeld: yes. there's been some complaints about the dependency on libnih creating issues
[21:34] <asalkeld> It has problems as nil-dbus-tool is build with the cross CC
[21:35] <sadmac> asalkeld: yep
[21:35] <sadmac> asalkeld: don't know that there's been a resolution sadly
[21:35] <asalkeld> does nil-dbus-tool generate c files (only - no other function?)
[21:36] <asalkeld> if so , it would be good if these c files could be generated as part of the shipped tarball
[21:37] <asalkeld> so we would not have to compile nil-dbus-tool at all.
[21:37] <sadmac> asalkeld: interesting idea.
[21:37] <asalkeld> ok, well thanks - I'll try find a solution myself
[21:38] <sadmac> asalkeld: technically the c files are not "source," but rather the xml they are generated from is
[21:38] <asalkeld> like a configure file ...
[21:38] <sadmac> true.
[21:39] <asalkeld> often there is some prep that is done before distributing a release (changelog, autotools, etc...)
[21:40] <asalkeld> this could be a part of this
[21:40] <asalkeld> unless there is some dbus version dependant code.
[21:41] <Keybuk> sadmac: correct
[21:41] <Keybuk> wing-commander scott% grep telinit /etc/init.d/umountroot
[21:41] <Keybuk> 	[ -f /var/run/init.upgraded ] && telinit u || :
[21:41] <sadmac> Keybuk: why re-exec at all? From a vm perspective I don't see what problems it causes.
[21:42] <Keybuk> sadmac: if you've upgraded libc or init, you have a file open on the root filesystem that is "dirty"
[21:42] <Keybuk> so you cannot remount readonly
[21:42] <Keybuk> the remount will fail
[21:42] <Keybuk> and you'll power off a dirty root
[21:42] <Keybuk> incur a mandatory fsck on reboot
[21:42] <Keybuk> etc.
[21:42] <sadmac> dunno why its dirty
[21:42] <sadmac> dirty implies there's an alternate "clean" state
[21:43] <Keybuk> yes
[21:43] <sadmac> there's no laundering procedure for the pages in question
[21:43] <sadmac> they should just be freed
[21:43] <sadmac> or rather become anonymous
[21:44] <Keybuk> they don't
[21:44] <sadmac> kernel mis-step methinks (or posix mandating stupidity as is often the case)
[21:48] <Keybuk> like most stupidity, posix
[21:48] <sadmac> fantastic
[21:57] <wasabi> My vbox init scripts work. Yay.