[10:07]  * xnox cannot see upstart source code side by side on my laptop & I'm away from my dual screens =(
[10:07] <xnox> 80 char limit is _good_ sometimes
[10:07]  * xnox lowers the font size & squints ;-)
[10:28] <dzhus> good day, can I run an upstart init script if it's symlinked to another location?
[10:28] <dzhus> like ~/.init/foo.conf -> ~/system-scripts/foo.conf
[10:30] <dzhus> oh, it's bug #665022
[10:36] <xnox> jodh: it now fails \o/
[10:38] <jodh> dzhus: that behaviour is by design, as shown on the bug.
[10:38] <jodh> xnox: great! so we can now just check the log and then unlink it.
[10:39] <xnox> bug 665022
[10:39]  * xnox ponders where is the bug bot.
[10:39] <xnox> jodh: ;-)
[10:49] <dzhus> Well, I was wrong. What 665022 (and the code) says is that upstart does not plug new services is when new symlinks are created in init directories. If the symlink is already there, it works finely.
[11:39] <xnox> ...simple test
[11:39] <xnox> /bin/bash: line 5: 10384 Segmentation fault      (core dumped) ${dir}$tst
[11:40]  * xnox ponders how to get a more useful fail....
[11:57] <xnox> it's ugly, but passes on the machine. Let me try in the autopkgtest now.
[12:12] <jodh> xnox: 'gdb --batch -ex r init/test_job_process' followed by 'bt' when it fails :)
[12:13] <xnox> jodh: =))))
[12:13] <xnox> thanks.
[15:16] <xnox> jodh: pushed merge-proposal to fix the above test. It's not the most optimal way but it works.
[15:16] <jodh> xnox: thanks!
[15:16] <xnox> jodh: all autopkgtest pass together with stgrabers fix & my branch fixing autopkgtests ;-)
[15:50] <Fecn> Hi Folks - For reasons I wont' go into, I'm trying to get upstart working in an initial ramdisk... it mostly sort-of works, but whenver I try to start services using initctl it gives me a variety of different 'Syntax Error: Bad fd number' messages in the debug output.
[15:50] <Fecn> Is there any way to get more debug output so I can see what is actually cuasing the bad fd messages?
[16:08] <xnox> we have a spec for eventdriven initramfs (that is upstart in the initramfs)
[16:09] <xnox> but it's not fully tested / ready yet.
[16:09] <xnox> search for upstart eventdriver initramfs if you are interested.
[16:09] <xnox> Can't help with fd number error, Fecn.
[16:09] <xnox> Hopefully somebody else will help you.
[16:10] <Fecn> xnox: Thanks - I'll have a search for the eventdriver stuff
[16:32] <SpamapS> man, this systemd/udev/w'ever thread on debian-devel just won't die
[16:54] <jodh> Fecn: try booting with 'console=ttyS0 --debug debug --no-log' and seeing what you get on the serial console. If the error is coming from initctl, you could add strace to the initramfs and use that.
[16:55] <Fecn> jodh: Serial console could be a challenge.. I'm not in the same physical location as the server... However, /dev/console is accessible to me OK (as is an ssh session if I fire up dhclient and sshd before I exec init
[16:56] <Fecn> I figure there will be some vital component missing from my initramfs somewhere along the way
[16:57] <Fecn> fwiw - the initramfs I'm working on is for a node-imaging system (formats hdd, runs rsync then does a kexec)
[16:57] <SpamapS> wait, so you're developing a custom initramfs .. remotely?
[16:57] <Fecn> Yep.. remotely via a KVM console thing
[16:58]  * SpamapS hands Fecn the "Bravest Craziest Mofo" award he got for upgrading ssh on a slackware box remotely back in 2002
[16:59] <Fecn> It's not really that crazy - I have power control, screen, keyboard etc... I can reboot it when it goes wrong.. I just can't get to the serial port
[16:59] <Fecn> sadly I have to do it this way as the only place where the infiniband adapters are located is in the datacentre machines
[17:01] <SpamapS> I've been through stuff like that before too
[17:01] <SpamapS> "Developer makes $85k/year, we can't possibly afford $2000 worth of extra equipment for their test setup."
[17:01] <Fecn> I did have it all working beautifully with a debian-based initrd... but we have our corporate RHEL standard.. so I'm having to rebuild it all using RHEL6 as the base
[17:04] <Fecn> For that matter, my RHEL based one is nearly working... apart from the kexec at the end, which hangs the machine if I do a forced reboot... hence trying to get enough of upstart to work so that it can actually do a 'shutdown -r' correctly