19:02:31 <jeblair> #startmeeting infra
19:02:32 <openstack> Meeting started Tue Jul 21 19:02:31 2015 UTC and is due to finish in 60 minutes.  The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:02:33 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:02:35 <openstack> The meeting name has been set to 'infra'
19:02:36 <jeblair> #link agenda https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Agenda_for_next_meeting
19:02:37 <jeblair> #link previous meeting http://eavesdrop.openstack.org/meetings/infra/2015/infra.2015-07-14-19.02.html
19:02:51 <jeblair> #topic Specs approval
19:02:55 <jeblair> #topic Specs approval: Host trystack.o.o site
19:02:59 <jeblair> #link host trystack.o.o site spec https://review.openstack.org/195098
19:02:59 <jeblair> #info host trystack.o.o site spec was approved
19:03:11 <jeblair> er, sorry i forgot to push the aprv button on this until today
19:03:26 <jeblair> but it's merged, and specs.o.o will update when we get a node
19:03:45 <pabelanger> danke
19:04:32 <jeblair> i think this is a bad week to schedule renames with so many folks afk
19:05:04 <jeblair> are there any priority-effort related things we should talk about that didn't make the agenda?
19:05:51 <fungi> none spring to mind
19:06:04 <jeblair> #topic Puppet-httpd replacement of puppet-apache (pabelanger/nibalizer)
19:06:21 <nibalizer> so paul has made most if not all of these patches
19:06:34 <nibalizer> im shy to approve them
19:06:49 <nibalizer> since it has the potential to be disruptive
19:07:18 <pabelanger> for the most part, our puppet apply testing looks good
19:07:19 <nibalizer> im wondering how we should go about landing it
19:07:23 <fungi> pabelanger: thanks for working on those!
19:07:40 <pabelanger> but I agree, we should try and merge all in same window
19:08:13 <fungi> is there a review topic they're all under?
19:08:14 <jeblair> this is moving from the old version of puppet-apache to our fork, right?
19:08:18 <nibalizer> im also nervous about lettig these changes age too much and get conflicts etc
19:08:22 <nibalizer> jeblair: correct
19:08:37 <pabelanger> fungi: puppet-httpd
19:08:41 <fungi> pabelanger: thanks
19:08:55 <jeblair> they _should_ be noops, right?
19:09:02 <nibalizer> jeblair: yes
19:09:23 <jeblair> have we landed any of them yet?
19:09:31 <nibalizer> pabelanger: the top of the stack doesnt remove the apache module any more right?
19:09:38 <pabelanger> jeblair: I believe a some have been merged
19:09:40 <pabelanger> let me find one
19:09:48 <fungi> #link https://review.openstack.org/#/q/status:open+topic:puppet-httpd,n,z
19:09:48 <pabelanger> nibalizer: right
19:10:15 <jeblair> looks like etherpad has?
19:10:41 <pabelanger> jeblair: yup
19:10:46 <mmedvede> pabelanger: tangential - I have noticed you use fully qualified 'include ::<class>' on a few patches. Is that a new recommended way to do includes? or necessity for puppet-httpd. I want to know as I was asked to use same style on one of my patches.
19:10:52 <jeblair> huh, that was random, wonder why i approved only that one.  ;)
19:11:10 <crinkle> mmedvede: yes that is preferred
19:11:19 <pabelanger> mmedvede: Ya, once we get our puppet-4 gates going, you'll need ::<class>
19:11:43 <mmedvede> crinkle: pabelanger: thank you
19:12:25 <jeblair> i think with the apply testing and at least one demonstrated success, we can probably approve them unscheduled.  does that work, or would folks prefer to batch them on friday?
19:12:42 * anteaya has no opinion
19:13:01 <clarkb> unscheduled wfm as long as the approver is able to watch that service
19:13:37 <nibalizer> ya, agree with clark
19:13:41 <pabelanger> I can be around when ever it happens, but agree our puppet-apply testing coverged everything
19:14:04 <fungi> but _do_ keep an eye on puppetboard.o.o after approving, _please_
19:14:16 <fungi> then they can at least be reverted if they cause issues
19:14:27 <pabelanger> feel free to ping me if any issues come up, I can look into the puppet side
19:14:29 <jeblair> okay cool, approve at will, and soon, then :)
19:15:27 <jeblair> #topic Open discussion
19:15:30 <nibalizer> sweet
19:16:11 <fungi> who's at oscon? pleia2 and i are at a table in the "e" area
19:16:27 <pabelanger> I'd like to talk about fedora22 dibs. For the most part, ianw has done great work getting diskimage-builder updated for it.
19:16:33 <clarkb> I am still at home but will likely wander over later today
19:16:38 <jeblair> drink a beer for me
19:16:39 <pabelanger> I have some puppet chances in our modules that could use some feedback
19:16:47 <fungi> jeblair: i already have, but i'll repeat the exercise
19:16:50 <pabelanger> https://review.openstack.org/#/c/186619/
19:16:56 <jeblair> fungi: as many as it takes :)
19:17:06 <fungi> absolutely
19:17:08 <AJaeger> fungi: Take some extra for me ;)
19:17:11 <fungi> you got it
19:17:24 <clarkb> pabelanger is there a topic to review?
19:17:36 * anteaya is at the nova mid-cycle
19:17:44 <clarkb> and is the bulk of the work making our puppet v4 compat?
19:17:46 <pabelanger> clarkb: I can use fedora-22
19:18:01 <pabelanger> clarkb: well, for this we really only need changes to puppet-ntp and puppet-jenkins
19:18:07 <fungi> clarkb: https://review.openstack.org/#/q/status:open+topic:puppet-httpd,n,z
19:18:34 <fungi> oh, i guess he meant aside from those
19:18:41 <jeblair> yeah, and hopefully we can further reduce/eliminate puppet use on test nodes, but that's later
19:18:43 <pabelanger> but ya, I also wanted to talk about the puppet-4 effort that crinkle is doing, and syncing Gemfile / Rakefiles across our modules.
19:19:02 <pabelanger> If people don't mind the automatation, I want to do something like https://review.openstack.org/#/c/204172/ for all our puppet infra modules
19:19:03 <nibalizer> fungi: pleia2 I'm around pdx but won't be osconning until tomorrow
19:19:15 <pabelanger> should help keep them in sync for our gate checks
19:19:40 <clarkb> syncing gemfiles?
19:19:44 <fungi> pabelanger: speaking of dib and rpm-based platforms, i would love some way to turn on retaining the package cache in the yum element, if you feel like hacking on that. if not, i'll likely get to it in the next week, but right now we're not actually pre-caching any packages on those platforms
19:19:55 <nibalizer> ya something simple to keep the extra bits in the modules synced is ++
19:20:00 <jeblair> pabelanger: yeah, i'm hoping that after the initial effort, we don't see a lot of churn here
19:20:09 <clarkb> dont these files never change?
19:20:17 <fungi> i _think_ we're caching debs now that my change merged over the weekend, but haven't checked back in on an image to confirm
19:20:24 <pabelanger> clarkb: they are changing now, for our puppet-4 support
19:20:28 <jeblair> clarkb: i think they are changing a bit right now as we enable puppet4 stuff
19:20:30 <nibalizer> clarkb: if we add a new lint-plugin thats a line in our gemfile
19:20:36 <crinkle> they do change sometimes, especially if we need to emergency-pin a gem in the gemfile
19:20:37 <jeblair> but after that's done, i think it would be great if they stopped changing :)
19:20:55 <clarkb> I see
19:20:55 <pabelanger> fungi: sure, I can look into it.  Any info about how you want it to look like works for me
19:20:59 <jeblair> if they keep changing, we may want to see if we could do something like 'hacking'
19:21:15 <fungi> pabelanger: something like the envvar toggle that's implemented in the dpkg element
19:21:15 <clarkb> I thibk that is what the commit message should say then
19:21:23 <clarkb> rather than "syncing"
19:21:24 <jeblair> where there is something where we control the release, and that something depends on the rest
19:21:42 <fungi> pabelanger: a lot of it will be up to what the dib reviewers are willing to accept though
19:21:43 <crinkle> i think my preference would be not to automate until it becomes clear that these files are changing at an inconvenient pace
19:22:00 <pabelanger> fungi: okay, I can start looking into it.  Expect some more questions from me about it
19:22:02 <clarkb> crinkle +1
19:22:33 <fungi> pabelanger: sure, and if it gets too black magic don't worry about it for now. i'll likely take a stab at it after i'm home from oscon
19:22:42 <pabelanger> okay, so in general people are okay with the sync of the files. If ran locally, then decide the future of the process
19:23:06 <nibalizer> I would like some assurance that the files are identical in all the modules
19:23:12 <nibalizer> but I don't care about the mechanism
19:23:25 <pabelanger> Ya, that is my main reasoning too
19:23:33 * crinkle afk but preference against automation until later is my stance
19:23:36 <pabelanger> to insure we gate the same way for all our puppet modules
19:23:44 <pabelanger> ensure*
19:23:55 <nibalizer> well its easy to write a test that just makes sure all the Gemfile, Rakefile md5s are the same
19:24:12 <nibalizer> i guess that would prevent you from landing a change to one though
19:24:21 * nibalizer shrug, not that big of a deal
19:24:25 <clarkb> it seems like massive overkill for a problem we havent had
19:24:50 <pabelanger> to be fair, we've never needed Gemfiles before
19:24:51 <nibalizer> well it is really annoying when two modules have different lint flags turned on
19:25:03 <nibalizer> thats a Rakefile thing, but its the same principle
19:25:06 <clarkb> we make them match now. then dont change them
19:25:14 <nibalizer> but yea we don't have that problem juts yet
19:25:17 <clarkb> that should be sufficiemr
19:25:49 <pabelanger> well, if we need to add new lint checks, then it has to change. Unless we centralize our lint checks outside of the module
19:26:24 <jeblair> good motivation to avoid frequently adding new lint checks :)
19:27:17 <pabelanger> Ya, only if puppet didn't introduce breaking changes between releases :D
19:27:38 <asselin> o/
19:27:51 <jesusaurus> i think it would be good to enforce some sort of consistency in the tests across the puppet modules
19:28:20 <jeblair> yeah, i don't think anyone is arguing that we should be inconsistent
19:28:22 <pabelanger> this issue is really driving the sync: https://review.openstack.org/#/c/202884/
19:28:29 <jeblair> just that a lot of automation in the area is not necessary at the moment
19:28:30 <pabelanger> since puppet 4 breaks on version checks now
19:28:59 <jeblair> so, let's stick with pabelanger's manual sync as we work through the puppet4 thing, and if it's a problem after that is complete, look into it further
19:30:28 <nibalizer> ok
19:30:51 <jeblair> anything else?
19:31:13 * anteaya has nothing else
19:31:27 <fungi> all set here
19:31:33 <pleia2> zanata testing is continuing at a good pace, good support from carlos at zanata
19:31:42 <anteaya> pleia2: yay
19:31:45 <pleia2> that's all really, happy over here
19:31:52 <anteaya> pleia2: wooooo
19:31:53 <jeblair> nifty!
19:32:11 <fungi> excellent
19:32:17 <AJaeger> And StevenK is working nicely on scripts!
19:32:36 <anteaya> yay StevenK
19:33:48 <asselin> for common-ci (downstream-puppet) we just have nodepool & a sample 3rd party script to tie them all together. Almost there! https://etherpad.openstack.org/p/common-ci-sprint
19:34:00 <anteaya> yay asselin
19:34:34 <anteaya> #link https://etherpad.openstack.org/p/common-ci-sprint
19:35:23 <jeblair> cool, i'm going to wrap it up for this week and let oscon folks get back to their beer
19:35:25 <jeblair> thanks everyone!
19:35:30 <jeblair> #endmeeting