[18:53] <nathwill> woot. global jam time
[18:54] <nathwill> also, anybody in the hillsboro area who is open to giving blood, yahoo's hosting a blood drive for redcross tomorrow
[18:58] <bkerensa> :D
[18:58] <nathwill> 8am-11am
[20:37] <nathwill> i think of firefly every time one of you folks joins
[21:23] <nibalizer> that was the intention ;)
[21:41] <bkerensa> heh
[22:04] <blkperl> slangasek: have you ever seen dconf-service hang on login
[22:43] <slangasek> blkperl: no.  I assume this is running against an NFS homedir? ;)
[22:44] <slangasek> blkperl: don't look at who's assigned to implement https://blueprints.launchpad.net/ubuntu/+spec/foundations-q-xdg-runtime-dir
[22:44] <blkperl> slangasek: I found a ticket https://bugs.launchpad.net/ubuntu/+source/d-conf/+bug/645448
[22:44] <slangasek> heh
[22:48] <blkperl> slangasek: so if we point XDG_CACE at /tmp or something it seems to allow login
[22:48] <blkperl> XDG_CACHE*
[22:48] <blkperl> we think that the homedir doesn't mount until in late in the login process which is why it just hangs in kernel land
[22:49] <slangasek> oh?
[22:49] <slangasek> are you using autofs?
[22:50] <blkperl> yes
[22:50] <slangasek> ah
[22:50] <slangasek> so yeah, autofs was something of a blind spot while implementing upstart boot-time filesystem handling
[22:51] <slangasek> essentially because the last time I tried to use it 8 years ago, it was crap, and I assumed everyone else had reached the same conclusion and only tested /etc/fstab-based mounting ;)
[22:51] <slangasek> I know there've been a couple of escalated bugs recently about autofs support, but I don't know where those got to
[22:51] <blkperl> well we use autofs because we can't have 5000 homedirs mounted at the same time on 200 nodes :)
[22:52] <blkperl> if you can point me to those autofs bugs that would be great
[22:52] <slangasek> right, but you could have a single /home mounted everywhere ;)
[22:52] <slangasek> but yeah, I realize this doesn't scale out
[22:53] <slangasek> looking now to see if there are any autofs SRUs that have gone through recently
[23:00] <slangasek> blkperl: there've been recent autofs SRUs for lucid and natty, but I'm pretty sure it's not related to what you're seeing
[23:03] <blkperl> slangasek: do you see any pitfalls with using /tmp for XDG_CACHE? This workaround seems to work
[23:03] <slangasek> blkperl: I couldn't say
[23:04] <slangasek> based on conversations at the last UDS I think later versions of dconf may have addressed this, but I'm not sure how
[23:06] <blkperl> I made a bug 1046079 if your interested in commenting
[23:07] <slangasek> commenting is almost certain to get me nag mail from the desktop team about XDG_RUNTIME_DIR support not being done yet ;P
[23:07] <slangasek> how about if I just work on that
[23:07] <blkperl> yeah that sounds great :)
[23:37] <nibalizer> :)
[23:40] <nathwill> blkperl: i might suggest /var/tmp
[23:40] <bkerensa> blkperl: you going to hack day on saturday at puppetlabs?
[23:40] <nathwill> instead of /tmp, as it's guaranteed to remain present until reboot
[23:40] <nathwill> whereas /tmp may be purged...
[23:40] <bkerensa> nathwill: Kitsune is alive http://15.185.226.243:8000/en-US/home
[23:40] <bkerensa> :D
[23:41] <blkperl> nathwill: that doesn't solve the problem of users on multiple hosts at the same time
[23:41] <blkperl> they changes hosts a lot and because nfs is broken....
[23:41] <blkperl> bkerensa: what hack day?
[23:41] <blkperl> bkerensa: calagator link?
[23:41] <bkerensa> blkperl: MFNW+PDX
[23:42] <bkerensa> http://calagator.org/events/1250462748
[23:42] <nathwill> blkperl, i was suggesting a better location for XDG_CACHE than /tmp
[23:42] <nathwill> but i see what you're saying... i seem to remember a bug about dconf and nfs homes, and the solution being to change the boot order...