[05:07] <pitti> Good morning
[07:20] <Saviq> cjwatson, hey, do you have pointers on how to mk-sbuild for rtm?
[07:51] <dholbach> good morning
[08:04] <sil2100> pitti: hello! Could we ask you to re-run the i386 autopkgtest for network-manager?
[08:04] <sil2100> pitti: the amd64 one seems to have passed
[08:06] <sil2100> pitti: ah, ok, just learned it's a real regression, nvm!
[08:46] <pitti> sil2100: right, already discussed that with cyphermox_ yesterday
[09:32] <cjwatson> Saviq: no
[09:33] <cjwatson> Saviq: it's awkward because debootstrap requires non-trivial fiddling to work for ubuntu-rtm.  You're possibly best off downloading the Launchpad buildd chroots from https://api.launchpad.net/devel/ubuntu-rtm/14.09/amd64 etc.
[10:14] <doko> Laney, replied to #1368094
[10:16] <Laney> ty
[12:03] <xnox> Logan_: what's up?
[12:03] <xnox> but 1368094
[12:03] <xnox> bug 1368094
[12:07] <Laney> pad.lv/ :P
[12:45] <jibel> jodh, I've two desktop machines running utopic that hangs on shutdown. It seems to stop on unmountfs. Any idea how I can debug that?
[12:50] <jodh> jibel: have you recently added some remote filesystems to fstab? I'd remove splash and quiet from the boot options, add '--debug' (which will put both upstart+mountall in debug mode). You may want to create mountall.override containing 'console output'. Also https://code.launchpad.net/~jamesodhunt/ubuntu/trusty/sysvinit/log-open-files-on-shutdown/+merge/196141 could be useful (never made it into the
[12:50] <jodh> distro alas)
[12:53] <jibel> jodh, my fstab is rather basic http://paste.ubuntu.com/8326868/ , I already removed quiet and splash, that's how I found it stopped on unmounting filesystems. I'll add --debug and have a look at your patch.
[12:56] <zul> doko: thanks for promoting python-oslo.utils, however launchpad says its in main but the archive thinks otherwise
[12:57] <doko> zul: launchpad says "1 hour ago"
[12:58] <doko> just wait
[12:58] <zul> ok
[13:03] <doko> glance (1:2014.2~b3-0ubuntu3) utopic; urgency=medium
[13:03] <doko>   * debian/control: Fix typos.
[13:03] <doko>   * debian/pydist-overrides; Add orderdict to overides.
[13:03] <doko> zul: looking at the changelog I can't believe that you actually fixed typos ;)
[13:04] <zul> doko: gee thanks
[14:06] <dholbach> hey mlankhorst, how are you doing?
[14:06] <dholbach> mlankhorst, do you think you'd need some more information for bug 1368784?
[14:19] <doko> pitti, just in case you are bored ... libcitygml seems to have some issues with the pkgbinarymangler
[14:20] <pitti> I'm still trying to get out of IRC/meetings/backlog swamp; at this point I need to ask to send bug reports
[14:36] <mdeslaur> slangasek: could someone from foundations please take over bug 1271591, it's really annoying on trusty
[14:45] <slangasek> jodh: ^^ can you have a look at this?
[14:45] <jodh> slangasek: ack
[14:45] <mdeslaur> slangasek, jodh: thanks
[15:39] <smoser> slangasek, when you have a  minute, i'd appreciate some thoughts on http://paste.ubuntu.com/8327941/ . which is attempt block networking coming up until cloud-init-local has run (as it might write networking).
[16:10] <Laney> whoops, had utopic-proposed on
[16:10] <Laney> wonder how much crack I've got
[16:10] <ogra_> the good stuff
[16:11] <Laney> you install ogra's upload, very nice
[17:19] <slangasek> pitti: hi, so bfiller has pointed out that libthumbnailer0 1.2+14.10.20140814-0ubuntu1 in ubuntu-rtm/14.09 is missing the corresponding ddeb.  Any idea?
[20:07] <smoser> hallyn, can i ask lxc: what is the real/outside pid of pid X inside container named foo
[20:15] <hallyn> smoser: no.  the kernel doesn't tell you that yet
[20:15] <hallyn> there is a guy on lkml trying to push a patchset that would let us figure that out
[20:15] <hallyn> but right now that info is simply not available
[20:16] <hallyn> smoser: unless you can have a process inside the container send a ucred as SCM_CREDNETIAL to the outside task
[20:16] <hallyn> the kernel will translate the pid for the receipient
[20:16] <hallyn> that's how cgmanager=-proxy works
[20:16] <smoser> but if i'm outside the container ?
[20:16] <smoser> ie, i'm on outside. i want to figure out "whats the real pid of the container's pid 1"
[20:17] <hallyn> well you could still write a program that setns's into the container's pidna and sends you a ucred
[20:17] <hallyn> maybe i should write that
[20:18] <smoser> how does 'lxc-stop' do it ?
[20:18] <hallyn> it gets the pid from the lxc monitor
[20:18] <hallyn> when a container starts, it starts a 'monitor' that respones to some commands over a unix sock
[20:18] <hallyn> includidng "wha tis your pid" and "what is your cgroup"
[20:18] <hallyn> and "stop"
[20:19] <hallyn> so if you jus twant it fo ra container, you can do 'lxc-info -p -H -n $containername"
[20:19] <hallyn> sorry, if that's all you wanted, i was overcomplicating :)
[20:20] <hallyn> unheard of from me :)
[20:20] <hallyn> smoser: is that all you needed?
[20:20] <hallyn> i still may write that utility though
[20:21] <smoser> actually, yeah. i think that is.
[20:21] <smoser> so is there any difference then between:
[20:21] <smoser>  pid=$(lxc-info -p -H -n "$name")
[20:22] <smoser>  kill --signal=SIGQUIT $pid
[20:22] <smoser> and
[20:22] <smoser>  lxc-stop --name=$name
[20:27] <hallyn> yeah, lxc-stop has some default behavior;
[20:27] <hallyn> it'll send stopsignal first, or SIGKILL if that's not defined
[20:28] <hallyn> oh no, first i'tll send SIGPOWER
[20:28] <hallyn> which upstart will respond to kindly
[20:29] <hallyn> then it'll send the quit signal
[20:34] <stgraber> yeah, lxc-stop by default is SIGPWR + 30s wait + SIGKILL, that's all configurable though. So the equivalent to smoser's command would be to have the container starting with stopsignal=SIGQUIT and then running lxc-stop -n container --nokill which would send SIGQUIT but never SIGKILL
[20:35] <stgraber> any container created by lxc should have a proper stopsignal set, either responding to the default SIGPWR or have the default config for the distro specify the right signal. Which means that lxc-stop -n blah is always the right way to cleanly stop a container and lxc-stop -n blah -k is always the right way to kill it immediately.
[20:38] <smoser> thanks
[20:41] <hallyn> think i'll write that pidtranslate tool tonight, maybe that'll obviate the need for a kernel patch
[22:19] <hallyn> done, that wasn't too bad.  though making it not hang on bad info would take a bit more work