19:02:19 <jeblair> #startmeeting infra
19:02:20 <openstack> Meeting started Tue Apr  2 19:02:19 2013 UTC.  The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:02:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:02:23 <openstack> The meeting name has been set to 'infra'
19:02:36 <jeblair> (who wrote that anyway)
19:02:42 <fungi> crazy people
19:02:53 <jeblair> #topic actions from last meeting
19:03:10 <jeblair> clarkb: how's that hpcs account thing coming?
19:03:34 <clarkb> the account has been created. mordred has sent email requesting that it not be billed for our use.
19:03:58 <clarkb> I recently realized that I need to check if he also requested that our quotas be increased and will do that if he hasn't done so
19:04:19 <clarkb> once we have these two things I will add az3 to devstack gate using these credentials
19:04:32 <jeblair> clarkb: cool, thanks... also, do we need to ask that things be enabled, like block and blob storage?
19:04:55 <clarkb> jeblair: no, we can simply do that through the web gui which was oen of the motivations for creating a second less complicated account
19:05:03 <jeblair> clarkb: gotcha
19:05:19 <jeblair> fungi: i saw you say something about gerrit groups this weekend...
19:05:32 <fungi> yep, we changed some
19:05:36 <fungi> #link http://lists.openstack.org/pipermail/openstack-dev/2013-March/007087.html
19:05:55 <fungi> so now the old .*-drivers groups in gerrit are renamed .*-milestone
19:06:23 <fungi> also i have a review in to make use of the .*-ptl groups i created...
19:06:31 <fungi> #link https://review.openstack.org/25806
19:06:41 <fungi> eyeballs appreciated
19:07:32 <jeblair> fungi: cool, thanks
19:07:42 <fungi> also if anyone wants to check behind me that i got the ptls right (went by the election results) that would also be appreciated
19:08:08 <jeblair> #topic grenade
19:08:34 <jeblair> sdague has been doing a lot of work getting grenade in shape for running in ci
19:08:48 <jeblair> i've been pitching in a bit to fix up devstack-gate
19:09:09 <sdague> I only have minor dents in my desk from my forehead hitting it
19:09:19 <clarkb> no shrapnel?
19:09:41 <fungi> merely flash grenades
19:09:46 <jeblair> sdague: anything you need from me/us?
19:09:46 <sdague> nah, all blunt trauma at this point
19:09:58 <sdague> no, I think the gate scripts are good now
19:10:09 <sdague> now it's just a matter of getting the projects to roll across
19:10:50 <jeblair> okay, let me know if you want to enable non-voting jobs on the projects themselves
19:11:33 <sdague> it's going to probably be a bit still
19:11:42 <jeblair> ok
19:11:45 <jeblair> #topic gearman
19:12:01 <jeblair> we've started testing zaro's plugin on jenkins-dev
19:12:12 <zaro> fixing your wonderful bugs.
19:12:22 <jeblair> yay!
19:12:57 <jeblair> i've also found that python-gearman doesn't seem well-suited for doing async foreground jobs with gearman
19:13:42 <jeblair> python-libgearman is an alternative, it's a swig wrapper around libgearman
19:13:59 <jeblair> but it is incomplete, and doesn't currently wrap the functions that i think we need
19:14:07 <jeblair> so i talked with the maintainer of python-libgearman....
19:14:11 <jeblair> mordred: hi there!
19:14:35 <jeblair> and we think that doing a new python-libgearman using cffi would be a good idea
19:14:35 <fungi> there's also a change pending to open firewall rules it needs on zuul
19:14:35 <fungi> #link https://review.openstack.org/25583
19:15:29 <jeblair> i should be able to hack together enough to do hello-world pretty quickly and guage how well that's going to work out
19:15:35 <jeblair> (and how long it would take)
19:15:46 <jeblair> but libgearman isn't a big library so it shouldn't take long
19:16:16 <jeblair> and i think by doing that, we'll have a solid basis to build on that's much more compatible with the way we want to use gearman
19:16:32 <jeblair> anyway, i'll propose a new repo for it as soon as i have something
19:17:11 <jeblair> and once that's in place, i think the zuul integration shouldn't be very difficult
19:17:31 <jeblair> any thoughts or anything else about gearman?
19:17:38 <jeblair> #link https://cffi.readthedocs.org/en/release-0.5/
19:17:43 <zaro> i don't have anything
19:17:52 <jeblair> that's cffi -- it seems to be a very nice new way to wrap c libs
19:18:32 <clarkb> that doesn't look bad
19:19:00 <jeblair> clarkb: mordred was pointing me at the callbacks section which looks particularly clean
19:19:19 <jeblair> he mentioned callbacks are usually the messiest part of these things
19:19:56 <jeblair> #topic pypi mirror / requirements
19:20:50 <jeblair> so that's fully in place now, i think next up is better gating jobs (clarkb has a patch), and then things we need to figure out at the summit (like stable branches)
19:20:58 <fungi> any feel for when we should delete the old pypi.o.o vm?
19:21:21 <jeblair> fungi: i think we can do that now.
19:21:35 <fungi> i'll add that to my list for the evening
19:21:57 <jeblair> if something really bad happens, it's easy to just stick a file in the mirror directory
19:22:00 <mordred> o/
19:22:07 <clarkb> my change should be good to go if I can get some reviews
19:22:13 <jeblair> so i don't think there's anything on there we need
19:22:13 <clarkb> (for betting gating)
19:22:33 <fungi> we'll need a designated bookie
19:23:03 <jeblair> #topic baremetal testing
19:23:23 <pleia2> so I've been quiet in -infra lately because I've been over in #tripleo doing testing with echohead of his baremetal stuff
19:23:28 <pleia2> (and sometimes we break mordred's laptop!)
19:23:42 <jeblair> pleia2: be careful, they don't make mordred's laptop anymore
19:24:07 <pleia2> we've got a few challenges there still, but I think most of them have to be tackled once echohead finishes his design decisions for his bootstrap VM
19:24:54 <pleia2> so that chugs along, and I could also use something else to work on now that reviewday is done, there is downtime while he fixes scripts and we go through reviews
19:25:37 * mordred also spent a few minutes looking at devstack-gate caching logic inside of diskimage-builder elements ... but don't think I hav anything super useful for us yet
19:25:38 <clarkb> rotating logs sanely
19:25:47 <clarkb> ^ pleia if you want something different to think about
19:25:49 <uvirtbot> clarkb: Error: "pleia" is not a valid command.
19:25:53 <pleia2> mordred: great, we'll have to talk about that soon
19:26:00 <mordred> pleia2: yes. it'll be fun
19:26:01 <pleia2> clarkb: sure
19:26:48 <pleia2> that's about it
19:27:10 <jeblair> #topic releasing git-review
19:27:20 <jeblair> fungi: did that happen yet?
19:28:03 <mordred> we're still derping on that one change
19:28:06 <fungi> jeblair: no, though i think we can merge 21609 and 20452 now if we get a little more reviewing on them
19:28:11 <fungi> and then release
19:28:20 <mordred> yah. they work - there is angst over how they work
19:28:29 <fungi> there's also mordred's flake8 addition which currently sits atop them
19:28:38 <fungi> which has enough reviews
19:28:41 <mordred> but I'm voting that we merge as is and take refactoring for later
19:29:17 <fungi> and there's one more mostly cosmetic change to alter the default topic creation if anyone wants to take a look
19:29:53 <jeblair> #topic open discussion
19:29:56 <mordred> ah - I like that one
19:30:04 <fungi> so given all that, i'd like to push for tagging no later than thursday, whatever's made it in by then
19:30:09 <mordred> fungi: ++
19:30:36 <fungi> and if stuff needs more edits/reviews, then it can wait for the following release
19:30:37 <ttx> o/
19:30:38 <mordred> people keep asking me for stats on how busy CI is
19:30:48 <anteaya> today was my first day of work
19:30:57 <jeblair> anteaya: congrats!
19:30:58 <anteaya> so I still have a few loose ends to tie up
19:31:01 <anteaya> jeblair: thanks
19:31:01 <mordred> we have the great zuul status graphs, but those are for a different purpose
19:31:04 <mordred> anteaya: congrats!
19:31:12 <anteaya> so I will keep working on those until they are done
19:31:16 <anteaya> mordred: thanks
19:31:18 <mordred> cool
19:31:25 <jeblair> mordred: there's more data in graphite than we currently graph too
19:31:32 <anteaya> might have to be in my spare time though, I will see how it goes
19:31:51 <mordred> jeblair: yeah. I figured there was ... it's just amazing how much people want numbers of stuff
19:32:08 <clarkb> mordred: what does "How busy CI is" mean? how busy are we the people? how busy is jenkins? how granular numbers are we talking? :P
19:32:15 <mordred> clarkb: :)
19:32:28 <mordred> clarkb: how much activity our work enables
19:32:36 <clarkb> I see
19:32:42 <mordred> and how hard of a problem space are we solving
19:33:13 <dprince> mordred: tell the bean counters to go away dude.
19:33:14 <mordred> so things like "patches uploaded per day, tests run per day, slaves launched per day" etc
19:33:15 <fungi> "lots" and "very"
19:33:19 <mordred> dprince: good luck with that
19:33:33 <dprince> mordred: or, better yet just give everyone a week off
19:33:38 <mordred> dprince: and actually, I want to give them good numbers, because it incites them to spend more money when they're proud
19:33:39 <dprince> mordred: then tell them to go away.
19:33:43 <clarkb> dprince: hahahaha
19:33:48 <mordred> dprince: :)
19:34:34 <clarkb> mordred: I think most of those numbers are in statsd
19:34:40 <mordred> I believe you are right
19:34:44 <fungi> dprince: i haven't looked lately... are the rhel6 slaves doing any better with unit tests now?
19:34:45 <clarkb> but it may take some effort to pull them out
19:35:09 <mordred> k. well, I may try to bug folks if I run in to problems
19:35:12 <anteaya> I did one blog post on my person stats patches submitted and reviewed and such, folks in china went nuts for it
19:35:18 <dprince> fungi: yep. almost done. One quantum unit test issue left I think. I'm going to try and get garyk to look into that one.
19:35:31 <pleia2> anteaya: hah, cool
19:35:37 <anteaya> it was funny
19:35:39 <zaro> jeblair: so i'm a little confused about the gearman library stuff.  are we going to work on libgearman or something else using cffi?
19:35:45 <fungi> dprince: awesome. maybe right after the release we can set it voting
19:35:59 <dprince> fungi: yep. that would be great.
19:36:02 <jeblair> zaro: i'm going to write a new python lib that wraps libgearman using cffi
19:36:39 <zaro> jeblair: so wrapper on top of a wrapper?
19:36:51 <jeblair> zaro: libgearman is not a wrapper, it's a client library
19:37:12 <zaro> ohh. i thought it was the python swig wraper of c lib.
19:37:15 <zaro> sorry.
19:37:22 <jeblair> zaro: that's the current python-libgearman
19:37:41 <zaro> ok. makes sense now.
19:37:59 <jeblair> zaro: it isn't complete, so this will be the new python-libgearman, and will ideally be more complete, and simpler, and not swig.
19:38:21 <clarkb> not swig ++
19:39:21 <jeblair> should we wrap this up?
19:39:36 <pleia2> oh, hoping to do another bug review next week
19:39:39 <clarkb> mordred: might also try asking hashar and others how many jobs they run with zuul and JJB
19:39:48 <pleia2> after release, so we'll probably want to retarget bugs too
19:39:56 <clarkb> mordred: transitive CI :)
19:40:06 <clarkb> pleia2: ++
19:41:58 <jeblair> thanks all!
19:42:01 <jeblair> #endmeeting