19:01:12 <jeblair> #startmeeting infra 19:01:13 <openstack> Meeting started Tue Jun 2 19:01:12 2015 UTC and is due to finish in 60 minutes. The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:17 <openstack> The meeting name has been set to 'infra' 19:01:22 <jeblair> #link agenda https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Agenda_for_next_meeting 19:01:23 <jeblair> #link previous meeting http://eavesdrop.openstack.org/meetings/infra/2015/infra.2015-05-26-19.02.html 19:01:29 <pabelanger> o/ 19:01:32 <crinkle> o/ 19:01:34 <nibalizer> o/ 19:01:36 <krtaylor> o/ 19:01:39 <ianw> o/ 19:01:40 <jhesketh> o/ 19:01:40 <docaedo> o/ 19:01:42 <AJaeger> +1 19:01:45 <rbradfor> o/ 19:01:49 <zaro> o/ 19:02:02 <jeblair> everybody gets a 'line said' in the meeting today! :) 19:02:05 <anteaya> I find myself going to the meetings wikipage to get the history to find links to meeting agendas 19:02:16 <yolanda> o/ 19:02:17 <anteaya> just as a data point to file for later 19:02:24 <anteaya> jeblair: one for everybody 19:02:27 <pleia2> o/ 19:02:30 <asselin_> o/ 19:02:32 <jeblair> #topic Announcements 19:02:33 <mrmartin> o/ 19:02:36 <jhesketh> yep, job done, back to bed ;-) 19:02:42 <anteaya> jhesketh: ha ha ha 19:02:45 <jeblair> #link http://docs.openstack.org/infra/system-config/project.html#teams 19:02:53 <jeblair> the process changes we talked about last week are in place, so this meeting structure has changed accordingly 19:03:00 <jeblair> we'll have a new section where we announce specs that are ready for the council to vote on 19:03:06 <jeblair> also, to many of you, welcome to the infrastructure council! i'm not sure anyone but gerrit knows who you are :) 19:03:32 <fungi> i have scripts that know ;) 19:03:56 * AJaeger would be suprised if fungi wouldn't have ;) 19:04:07 <jeblair> any questions or comments on this? or shall we go try it out? 19:04:23 <fungi> let's get this party started 19:04:23 <AJaeger> yes, let's give it a try... 19:04:32 <pabelanger> trial by fire! 19:04:37 <jhesketh> jeblair: might be we worth linking who is on the council if that's around somewhere? 19:05:02 <fungi> jhesketh: i'll get a link together in a bit. i think i can do it with a single gerrit api call 19:05:10 <jeblair> ooh, neat 19:05:14 <jhesketh> shiny 19:05:27 <jeblair> it's implemented as a gerrit group with about 100 member-groups (some of which have their own groups) 19:05:41 <AJaeger> jeblair: yeah https://review.openstack.org/#/admin/groups/925,members looks impressive 19:06:06 <jeblair> i had to write a script (which is in review; failing pep8 i think) to create it 19:06:23 <jeblair> #topic Actions from last meeting 19:06:29 <jeblair> mordred write a spec to move infra bugs to maniphest 19:06:46 <jeblair> mordred said he might not make the meeting today, but this is in progress 19:07:04 <jeblair> he did push up the shade spec which is great 19:07:17 <jeblair> hopefully that one will make the agenda next week 19:07:23 <jeblair> #action mordred write a spec to move infra bugs to maniphest 19:07:27 <jeblair> jeblair write infra-cloud spec with SpamapS 19:07:29 <jeblair> is done 19:07:41 <jeblair> and there's one more action item we'll talk about later in the meeting 19:07:46 <jeblair> #topic Specs approval 19:07:56 <jeblair> when specs are ready for final voting by the infra council, add them to the agenda in this section and we will announce them here 19:08:02 <jeblair> we can discuss them a little as needed, but mostly this is to serve as notice that people should vote on them 19:08:07 <jeblair> i was thinking we would start with a ~48 hour voting period 19:08:12 <jeblair> how does that sound? 19:08:15 <nibalizer> i like that 19:08:20 <pleia2> sounds good 19:08:27 <anteaya> what are you looking for in terms of votes? 19:08:34 <anteaya> how many to move ahead? 19:09:08 <jeblair> anteaya: simple majority? 19:09:12 <jeblair> and if we think it's not working out, we'll change it 19:09:19 <anteaya> how big is the council? 19:09:32 <jeblair> unknown :) 19:09:36 <anteaya> is majority 51% or better 19:09:56 <jeblair> let's go with "more positive than negative votes" 19:10:10 <anteaya> so 2 positive and 1 negative is a merge? 19:10:16 <anteaya> or mergable? 19:10:26 <jeblair> if that's all that can be bothered to vote within 48 hours, sure 19:10:52 <jhesketh> votes can be added before the 48 hours is open yeah? 19:11:02 <jeblair> jhesketh: i don't see why not 19:11:07 <anteaya> so window for voting set in the meeting, more positive than negative and the spec merges 19:11:08 <nibalizer> jeblair: if someone wants to abstain from voting on a partictulr spec, how do you want them to do that, +0? 19:11:11 <jhesketh> ie, we've reviewed it earlier or saw it was in the agenda ahead of time and added a vote 19:11:24 <anteaya> nibalizer: yes I 0 vote is abstain 19:11:33 <anteaya> a 0 vote 19:11:37 <pleia2> yeah, it's noted in the interface 19:11:43 <pleia2> webui anyway :) 19:12:08 <nibalizer> cool 19:12:17 <jeblair> you can also leave a differing code-review vote. i'm not entirely sure what that means in all circumstances, but we might find it useful. you might -1 and +0 for some really good reason. 19:13:24 <jeblair> #topic Specs approval: Infra-cloud (jeblair, SpamapS) 19:13:28 <zaro> pleia2: does it really say '0 abstain'? 19:13:31 <jeblair> #link infra cloud spec https://review.openstack.org/186960 19:13:37 <pleia2> zaro: yes 19:13:37 <jeblair> #info voting on https://review.openstack.org/186960 open until 2015-06-04 19:00 UTC 19:14:31 <jeblair> we've talked about this for a while, and it's worth noting that the technical decisions will be made over in system-config, so be sure to participate there if you have opinions on openstack deployment choices 19:14:36 <mordred> o/ 19:14:38 <jeblair> but this covers the process and overall effort 19:14:45 <asselin_> can we add this one too? https://review.openstack.org/#/c/135170/ 19:15:00 <jeblair> asselin_: let's do it next week 19:15:20 <AJaeger> asselin_: please add to the agenda for next week 19:15:36 <asselin_> will do, thanks 19:15:38 <krtaylor> thanks asselin_ 19:16:22 <jeblair> #topic Specs approval: Host OpenStack Apps Catalog Service (docaedo, fungi) 19:16:31 <jeblair> #link apps.o.o spec https://review.openstack.org/187646 19:16:32 <fungi> yep 19:16:37 <jeblair> #link apps.o.o topic https://review.openstack.org/#/q/status:open+topic:apps-site,n,z 19:16:41 <fungi> there's this apps site 19:16:44 <fungi> and we should host it 19:16:52 * fungi wins at tl;dr 19:16:55 * pleia2 nods 19:17:00 <jeblair> ayup 19:17:07 <fungi> any... questions? 19:17:19 <AJaeger> fungi: if the specs is as easy to understand, you have my +1 vote ;) 19:17:21 <jeblair> and i think this is already in progress 19:17:25 <fungi> it is 19:17:40 <fungi> the review topic already has associated changes awaiting review 19:17:41 <jhesketh> noting that the spec was uploaded today, should there be a minimum review time before it is voted on? 19:18:20 <jhesketh> since it also (unless I missed it) wasn't in the agenda for very long as a heads up 19:18:29 <fungi> perhaps, yes. i suggest a minimum review time of at least 0 seconds 19:18:34 <fungi> we're not deciding on it in this meeting 19:18:43 <fungi> just announcing that it's ready for review/council vote 19:19:01 <jeblair> yeah, i think we'll want to get to the point where things are on the agenda longer than "immediately before the meeting", but we're also sort of catching up current efforts to the new process 19:19:03 <jhesketh> isn't announcing it is ready for vote putting it in a 48 hour window 19:19:45 <jeblair> and in this case, i think with the summit discussion and key participants agreeing on the current spec, it's okay for this one 19:19:55 <fungi> also it's sort of a rush job 19:19:57 <jeblair> jhesketh: yes 19:20:07 <jhesketh> yep, sure, I don't think this is an issue, just pointing out that if somebody were away for 2 days it'd be easy to miss a spec like this 19:20:38 <jeblair> yeah, we'll get better in the future; if there isn't an objection for this particular instance, i'd say we should go ahead with the vote 19:20:42 <fungi> some decisions were made, for better or for worse, and this is a "bug" we need to correct as soon as possible, before it becomes a major pr issue 19:20:44 <jhesketh> but that's not going to be common and objections can be given after it is merged 19:21:40 <jeblair> well, ideally we'd get objections before it's merged 19:22:04 <jhesketh> I meant worst case 19:23:41 <jeblair> yeah, if something looks like it's going to go off the rails, let's bring it up in this meeting (we can choose to defer voting), or -1 council vote it with a suggestion for more revising, or infra-cores can veto it on that basis too, and we'll get the process back on track 19:24:07 <jeblair> the goal is to get all the participants on the same page, so we should work toward that :) 19:24:35 <jeblair> i'm glad i rushed something on the agenda, that's useful :) 19:24:35 <jhesketh> sounds good :-) 19:24:42 <jeblair> #info voting on https://review.openstack.org/187646 open until 2015-06-04 19:00 UTC 19:24:55 <jeblair> #topic Priority Efforts 19:25:25 <jeblair> reminder that this section is by prior request on the agenda only, so i'll skip efforts that didn't have something in the agenda 19:25:30 <jeblair> #topic Priority Efforts (Downstream Puppet) 19:25:44 <asselin_> Hi, I would like to propose a virtual sprint 19:25:52 <jeblair> #link https://wiki.openstack.org/wiki/VirtualSprints#Schedule_of_Upcoming_OpenStack_Virtual_Sprints 19:26:17 <asselin_> that was very effective last time, and I think the initiative would benefit from doing this 19:26:52 <asselin_> I'd like to schedule this after L1 is done...so early L2 if there's enough people avaiable / not on vacation 19:27:00 <fungi> asselin_: what goal do you have in mind for the common ci solution sprint? 19:27:24 <fungi> just work toward implementation of teh spec and any oassociated work items? 19:27:32 <asselin_> fungi, yes 19:27:45 <yolanda> asselin, ++ 19:27:46 <jeblair> maybe we could identify one or more modules to complete? possibly identify one that we might be able to move openstack-infra to using? 19:28:17 <jeblair> s/modules/classes/ you know what i mean... maybe 'compenents'. 19:28:22 <asselin_> I'd like to get nodepool, zuul, and jenkins done 19:28:36 <nibalizer> ya moving -infra to consuming openstackci would be huge 19:28:57 <anteaya> asselin_: setting priorties would help 19:28:59 <fungi> i like having achievable or measurable goals for a sprint. if completion of the entire spec is doable within the sprint, then seems fine as a goal. otherwise would be great to see it scoped more clearly 19:29:01 <asselin_> the alternative is to keep doing as we are 19:29:11 <nibalizer> I'm in for this, though i think the parallelism here is much lower than the modules split sprint 19:29:20 <anteaya> asselin_: so folks can focus on one and then move to the next 19:29:59 <asselin_> anteaya, that's another good approach. We can target e.g. 1 module per week and focus on that 19:30:11 <anteaya> asselin_: sorry no I meant for the sprint 19:30:17 <anteaya> I support the sprint 19:30:27 <nibalizer> for i in ls system-config/modules/openstack_project/*; do (move stuff into either the correct module or to openstackci; move infra to consume that); done 19:30:34 <anteaya> but people need you to make a decision, however arbitrary it may feel 19:31:01 <anteaya> asselin_: so select a time and select 3 modules (you did already) and prioritize them 19:31:07 <anteaya> asselin_: then people can take action 19:31:33 <anteaya> asselin_: until you make a decision, people will just keep giving you suggestions 19:32:01 <asselin_> anteaya, I am asking for peoples opinions 19:32:09 <nibalizer> sprinting is good i think 19:32:18 <yolanda> asselin, i think that these modules are the ones that need attention 19:32:20 <anteaya> asselin_: so far I'm hearing when you decide on dates for a sprint, people will show up 19:32:26 <yolanda> i'm ok to have an sprint for that 19:32:37 <nibalizer> a sprint like atmosphere is the best way to accomplish the 'pivot' needed to get infra using openstackci I think 19:32:42 <pabelanger> I would think nodepool would be the first to break out of the list, limited dependencies 19:32:48 <nibalizer> the slow way of doing that will take eons and introduce risk 19:32:52 <asselin_> ok, so yolanda and nibalizer are in for sprint. What about infra cores? 19:32:58 <yolanda> pabelanger, i started work on nodepool 19:32:59 * zaro has no opinion just willing to help 19:33:20 <yolanda> i'm trying to find gaps during the week to continue with that 19:33:35 <anteaya> asselin_: I support the sprint, my attendance depends on dates selected, and don't optimize for me 19:33:44 <fungi> we have it designated as a priority effort, so i'm in favor of getting it done however we can to free up bandwidth for other priorities 19:33:44 <jeblair> i'm happy to help 19:33:47 <pleia2> I can participate 19:33:48 <asselin_> fungi I think is in as long as there is clear measurable scope. We can do that. 19:34:04 <pabelanger> yolanda, Ya, seen your review. Was going to comment on it about allow a template or content to just be passed, over individual elements. But, I'm not against what you started 19:34:08 <nibalizer> our testing is still 'in progress' so we should make sure that we leave ample time between now and the date of the sprint to get our framework for pupept testing locked in 19:34:16 <nibalizer> locked down? set up? established? 19:34:20 <jeblair> nibalizer: that will help tremendously 19:34:24 <yolanda> pabelanger, i got some feedback from jeblair, asselin, so i'm going to refactor that a bit 19:34:50 <asselin_> nibalizer, you think you can get that done by L1? 19:35:24 <asselin_> Libery schedule is her: https://wiki.openstack.org/wiki/Liberty_Release_Schedule 19:35:25 <nibalizer> asselin_: i dont know when L1 is 19:35:27 <asselin_> #link https://wiki.openstack.org/wiki/Liberty_Release_Schedule 19:35:33 <jeblair> (i don't normally thing much about milestone dates, they don't really affect us too much) 19:35:35 <asselin_> June 25th 19:36:16 <yolanda> i'm on holiday since 29th june, i hope it's scheduled before that... 19:36:16 <asselin_> jeblair, I'm avoiding the sprint bfore L1 b/c I assume many people are busy then 19:36:20 <crinkle> is puppet functional testing included as part of downstream-puppet? clarkb's spec doesn't have a decision yet so I don't know if it's appropriate to try to lock down testing https://review.openstack.org/#/c/178887 19:36:40 <nibalizer> crinkle: thats fair... 19:36:49 <fungi> milestone dates are only a concern insofar as if they indicate a huge uptick in dev activity for openstack projects on the whole then we want to strive to keep things more stable during those windowsa 19:37:09 <asselin_> fungi, +1 19:37:11 <nibalizer> clarks spec was sortof subsumed by the discussion at the design summit, as i understand it 19:37:35 <nibalizer> so we didn't resolve or land clarks spec, but we did make a decision on a path forward 19:37:43 <asselin_> ihmo we can do this in parallel 19:37:51 <jeblair> i think we decided to head towards beaker-rspec, we should update clark's spec to reflect that 19:37:53 <nibalizer> then when we sprinted at the summit we (or I and some others) focused on testing using teh decision we made 19:38:03 <crinkle> jeblair: ++ 19:38:13 <nibalizer> cool 19:38:41 <asselin_> so how about June 29 and 30? 19:38:43 <nibalizer> but I do think we can get beaker-rspec in place before the 'late june' sprint that asselin_ is suggesting 19:39:00 <jeblair> tuesdays are bad for me :) 19:39:01 <pabelanger> asselin, Close to Canada and USA holidays 19:39:01 <nibalizer> i also think testing is one of the ways virtual sprints prevent us from hurting ourselves 19:39:06 <pabelanger> Canada day is July 1st 19:39:20 <yolanda> 29 and 30 don't work for me, but i'll leave my work prepared for you to take it 19:39:28 <fungi> i'm closing on a house on june 30 so will likely not be around for that timeframe, but don't plan around me 19:39:29 <asselin_> ok July 8 & 9? 19:39:36 <yolanda> ++ 19:39:38 <jeblair> wfm 19:39:41 <nibalizer> wfm 19:39:43 <pabelanger> ya, better here 19:39:52 <asselin_> I will create a vote in the etherpad and send to the e-mail list. 19:39:53 <fungi> i can probably do july 8-9, yes 19:39:55 <pleia2> I might be out of town then, but I have to confirm 19:40:05 <anteaya> asselin_: I also may be out of town 19:40:19 <anteaya> but july 8 and 9 seem to be popular with others 19:40:26 <zaro> wfm 19:40:37 <jeblair> #action asselin_ propose july 8,9 for openstackci sprint to mailing list 19:40:49 <zaro> reminder would be excellent 19:40:51 <asselin_> ok thanks 19:40:55 <jeblair> asselin_: cool, thanks 19:40:59 <asselin_> zaro, will do. thanks 19:41:03 <jeblair> anything else? 19:41:07 <asselin_> that's it 19:41:13 <jeblair> #topic Priority Efforts (Upgrading Gerrit) 19:41:20 <zaro> https://gerrit-review.googlesource.com/#/c/68149/ 19:41:33 <zaro> i tested the proposed fix, but it didn't fix. 19:41:48 <pleia2> :\ 19:42:13 <jeblair> zaro: uh oh 19:42:27 <anteaya> did you post your results to anywhere google can read? 19:42:44 <fungi> he commented on the review 19:42:45 <zaro> yes, on mailing list + on the review 19:43:25 <fungi> #link https://groups.google.com/forum/#!msg/repo-discuss/ZeGWPyyJlrM/mQsZIsy2740J 19:43:40 <zaro> not sure what to do besides wait for real jgit fix. other option is to increase timeout like the sony guys did 19:44:05 <jeblair> what did they set their timeout to? 19:44:27 <zaro> not sure, but can ask sven. 19:44:50 <zaro> he said he increased in the mailing list and that made error go away. i will ask him 19:45:11 <jeblair> i'm a little concerned about that, because occasionally we have some _really huge_ changes; we either set ourselves up to a dos by letting gerrit perform all that computation, or a dos by killing repositories 19:45:33 <jeblair> on the plus side, it's really cool this is possible to test now :) 19:46:06 <zaro> yeah, not ideal but i think that's what sony is doing now 19:46:22 <zaro> just an option wanted to throw out. 19:46:35 <jeblair> zaro: can you keep testing their proposed fixes 19:46:49 <jeblair> and let's see if they can resolve this soon-ish 19:46:49 <zaro> yes, i plan to. 19:46:58 <anteaya> is moving to 2.9 an option? 19:47:13 <jeblair> and if they give up, we'll work something else out :) 19:47:41 <jeblair> anteaya: i don't feel like we're under significant pressure to upgrade to 2.9 19:47:46 <anteaya> okay 19:47:53 <anteaya> I would like close connection 19:48:01 <anteaya> which works with 2.9 but not 2.8 19:48:07 <anteaya> but that's just me 19:48:15 <anteaya> 2.9+ 19:48:40 <jeblair> anything else? 19:48:54 <zaro> i'm not sure that's true, i think i had to add it to our 2.10 branch 19:49:03 <zaro> anteaya: ^ 19:49:18 <zaro> http://git.openstack.org/cgit/openstack-infra/gerrit/commit/?h=openstack/2.10.2&id=5b6d0cd846ed2389584a0559866aeff3c1fff20a 19:49:29 <anteaya> close connection works with 2.9+, it is a feature in 2.11 but you backported successfully to 2.10 and 2.9 19:49:33 <zaro> which means it's only in 2.11 19:49:34 <anteaya> it doesn't work with 2.8 19:49:56 <anteaya> at least that is what I recall you told me 19:49:59 <zaro> ohh, i see what you mean 19:50:16 <zaro> nothing else from me on this topic 19:50:18 <anteaya> so staying on 2.8 means no close connection feature, which I would like 19:50:20 <jeblair> okay, let's see where the gerrit folks get to on this 19:50:22 <anteaya> I'm done too 19:50:25 <jeblair> #topic puppet-stackalytics (pabelanger) 19:50:33 <pabelanger> ohai 19:50:45 <nibalizer> hello 19:50:55 <nibalizer> we cooking this up again, then? 19:51:06 <pabelanger> So, mostly helping mordred with the effort to get stackalytics under infra. 19:51:23 <pabelanger> jeblair asked to get a spec up, so I did that today: https://review.openstack.org/#/c/187715/ 19:51:37 <jeblair> we had a nice work session at the summit where a lot of folks generally thought that stackalytics was the way to go 19:51:44 <jeblair> pabelanger: oh cool, that was fast :) 19:51:54 <mrmartin> so it means we will drop activity.o.o ? 19:51:57 <pabelanger> I also have a base puppet-stackalytics modules already created and ready for governance. https://review.openstack.org/#/c/187645/ and https://review.openstack.org/#/c/187269/ 19:52:20 <fungi> the direction forward we determined from the infra analytics tools discussion at the summit was to try and get stackalytics hosted/managed within the openstack community and supplement it with anything we were originally relying on other analytics tools for 19:53:05 <fungi> so that includes probably activity.o.o reports, things ptls want from reviewstats, et cetera 19:53:13 <jeblair> mrmartin: i don't think reed has plans to do that any time soon, it has a lot of things he still needs 19:53:13 <mrmartin> ok. 19:53:14 <nibalizer> cool 19:53:21 <pabelanger> I'm not sure who at mirantis to talk to about the move, I know mordred knows some peeps. But would be good to get them involved with the spec if possible 19:53:33 <jeblair> pabelanger: SergeyLukjanov and docaedo offered to help on that front 19:53:38 <fungi> SergeyLukjanov volunteered to assist' 19:53:41 <fungi> er, that 19:53:41 <pabelanger> roger 19:54:19 <pabelanger> Ya, so once the puppet-stackalytics is created, I'll port over the work mordred did a few months ago and polish it up 19:54:34 <pabelanger> Should be a straightforward move 19:54:34 <jeblair> cool, so people should go review that spec, and we'll certainly want some mirantis folks signing off on that too 19:54:38 <jeblair> pabelanger: sounds good 19:55:01 <jeblair> reed: ^ you'll probably want to look at that 19:55:06 <jeblair> #link https://review.openstack.org/#/c/187715/ 19:55:25 <jeblair> anything else on this? 19:55:37 <pabelanger> nothing from me 19:55:48 <jeblair> #topic Schedule next project renames 19:55:55 <docaedo> please add me as a reviewer for the stackalytics stuff :) 19:55:59 <jeblair> (technically not on agenda) 19:56:07 <jeblair> but we have a boatload of these 19:56:13 <jeblair> many of which actually have changes ready now 19:56:35 <jeblair> did we decide on an interval for this? like once every N- weeks/months? 19:56:55 <fungi> not that i recall 19:56:56 <nibalizer> I also heard chatter about an ansible playbook to streamline it 19:56:57 <anteaya> I didn't know we were trying to decide on an interval 19:57:11 <jeblair> okay, we might want to at some point... 19:57:15 <jeblair> anyone seen the playbook? 19:57:17 <fungi> yeah, mordred wanted to test-drive his rename playbook this time 19:57:30 <anteaya> where does it live? 19:57:48 <jeblair> anteaya: that's the question :) 19:58:07 <fungi> #link https://review.openstack.org/105057 19:58:09 <fungi> i think 19:58:14 <pabelanger> fungi, yes 19:58:24 <jeblair> we should probably go ahead and schedule, and if the playbook shows up, cool 19:58:30 <anteaya> 3 -1's 19:58:38 <anteaya> jeblair: yeah I agree with that 19:58:45 <jeblair> this friday? next friday? 19:59:05 <anteaya> either is fine with me 19:59:14 <anteaya> we have quite a load do we want to split them up? 19:59:33 <jeblair> anteaya: they won't take much longer 19:59:39 <anteaya> okay 19:59:44 <jeblair> anteaya: it's easier on us to have bigger batches 19:59:57 <anteaya> ah okay then I'm for next friday then 20:00:07 <fungi> agreed, big batches are fine 20:00:20 <anteaya> to ensure all the patches are ready 20:00:32 <jeblair> deferred until next meeting then 20:00:34 <jeblair> #endmeeting