17:00:16 #startmeeting ironic 17:00:18 Meeting started Mon Oct 12 17:00:16 2015 UTC and is due to finish in 60 minutes. The chair is jroll. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:19 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:21 The meeting name has been set to 'ironic' 17:00:23 hi all, who's here for the ironic meeting? 17:00:30 #chair devananda NobodyCam 17:00:30 :) hi all 17:00:31 Current chairs: NobodyCam devananda jroll 17:00:32 o/ 17:00:37 \o 17:00:39 o/ 17:00:41 o/ 17:00:44 o/ 17:00:50 o/ 17:00:56 o/ 17:00:58 \o/ 17:01:00 o/ 17:01:04 as always, the agenda is here: https://wiki.openstack.org/wiki/Meetings/Ironic 17:01:05 o/ 17:01:08 #link https://wiki.openstack.org/wiki/Meetings/Ironic 17:01:18 o/ 17:01:43 #topic Announcements 17:01:50 o/ 17:02:03 first and foremost, I'd like to officially welcome vdrok and jlvillal to the ironic core reviewer team \o/ 17:02:29 \o/ 17:02:36 thanks :) 17:02:41 I just want to say thank you :) 17:02:51 congratulations, vdrok and jlvillal ! 17:02:56 o/ 17:03:07 yes, congrats! 17:03:10 \o/ 17:03:17 secondly, an update on Liberty release status. we're planning on releasing 4.2.1 this week, which will become the Liberty integrated release 17:03:27 welcome to the team, vdrok & jlvillal :) 17:03:32 trying to land and backport this patch first: 17:03:34 #link https://review.openstack.org/#/c/231215/ 17:03:35 congrats jlvillal vdrok ! 17:04:01 Thanks everyone 17:04:16 does anyone have any other announcements? :) 17:04:27 jroll, +1, I think it may need an email as well, cause that's also a breaking change 17:04:45 lucasagomes: yep, it'll be in email and release notes, I'll hack on those today 17:04:51 cool 17:06:03 ok, moving on... 17:06:03 We from the OneView driver team are glad to announce the approval of our patch :) 17:06:11 ah, yes! 17:06:20 Thank you all for all the reviews 17:06:24 the OneView driver is on it's way to merge \o/ 17:06:27 https://review.openstack.org/#/c/191822/ 17:06:40 \o/ 17:06:42 #topic sub-team status reports 17:06:43 :D 17:06:43 nice! 17:06:47 as always, these are at: 17:06:50 #link https://etherpad.openstack.org/p/IronicWhiteBoard 17:07:00 * jroll gives folks a few minutes to review 17:07:25 Some ongoing work that got spun up a couple of weeks ago but isn't on any of the updates 17:07:32 is the push to make the IPA ramdisk smaller 17:07:43 curious if anyone's aware of a specific catalyst for that? 17:08:04 gabriel, w00t :-) 17:08:08 JayF: running full tempest currently takes 4-6 hours or something ridiculous, because we can't do much build concurrency 17:08:52 uggh sorry was late 17:08:54 JayF: (and that's something we'd like to get running at least on check in nova) 17:09:29 does anyone have any questions/comments on the subteam reports? 17:10:10 jroll, JayF perhaps we can use sambetts tinyIPA for the parallel tests 17:10:26 * lucasagomes tested it locally with iscsi methodology and it works 17:10:34 lucasagomes: yeah, something to look into 17:10:51 I don't object at all to looking at new/smaller ramdisks or shrinking the CoreOS one 17:10:59 just curious as to what the explicit goal was 17:11:03 yep 17:11:38 JayF, right 17:11:57 JayF: Maybe sambetts can comment on motivation? 17:13:00 jlvillal: well it sounds like from jroll that we're just trying to get more parallelism for our tempest tests, which makes sense to me 17:13:50 anything else here or can we move on? 17:13:50 also proof-of-concepts on virtual machines: not everyone has so much RAM on their laptop.. 17:13:59 I was just interested it seeing if we could make a smaller ramdisk with a hopfully less complicated build for IPA, I just like experimenting with things like this :) 17:14:39 https://github.com/Tehsmash/tinyipa 17:14:47 for reference ^ 17:14:54 #link https://github.com/Tehsmash/tinyipa 17:15:00 ty jroll :) 17:15:09 ok, moving on for now... 17:15:10 sambetts: if you proposed that to imagebuild/ (assuming you are Tehsmash) I'd have no opposition to inlcuding that as well 17:15:25 ++ 17:15:36 sambetts: you'd probably be able to refactor it to package whatever git HEAD is at too 17:15:51 and we could start publishing a tinyipa tarball as well :) No reason to not do it 17:16:02 ++ 17:16:10 yeah I like that 17:16:39 #topic Summit session prioritization/scheduling 17:16:42 let's move on 17:16:53 so, a list of session proposals are here: 17:16:55 #link https://etherpad.openstack.org/p/mitaka-ironic-design-summit-ideas 17:17:03 we have four fishbowls and four workroom sessions 17:17:16 I've reserved a fishbowl for a joint session with nova at the bottom 17:17:34 I'd like to get a sense of what sessions we think are most valuable; we need a final schedule by thursday 17:17:39 jroll, let's make sure the nova folks will come to the room 17:17:51 lucasagomes: yep, johnthetubaguy is wrangling them to do so 17:17:55 cool 17:18:12 does anyone have any strong opinions for or against any of these sessions? 17:18:29 I think the notification one is really valuable, I would love to see it implemented 17:18:35 tho I don't know if it's controversial 17:18:44 * lucasagomes feels everyone agrees notification is pretty cool 17:18:47 I see a few suggestions on the pad to cover topics on the mailing list instead of the session, which is good 17:19:01 +1 on the driver compistion one 17:19:02 lucasagomes: yeah +1, though what/how/when we send notifications would need some work 17:19:10 lucasagomes: do you mean the #2 on that list? 17:19:13 yeah perhaps the format etc 17:19:22 gabriel, yes 17:20:05 * jlvillal tempted to remove all colors, but won't ... 17:20:08 I'd love to have that discussed/decided/implemented too :) 17:20:13 the non-tech one that I would like to see as well is the "How can we improve rate of reviews" 17:20:16 jlvillal: you can turn them off on your end in settings :) 17:20:21 n 11 17:20:33 jroll: Ah, thanks 17:20:37 I think this is the core part of the project, if we can improve it the better we are 17:20:59 lucasagomes: + 17:21:15 I think "How can we improve rate of reviews" should be solved in a bar on Monday :) 17:21:26 dtantsur, heh that also works 17:21:31 needs to be discussed 17:21:33 I'd personally like to discuss third-party drivers and how we eventually get to the point where we have third party CI for them 17:21:50 dtantsur: ++ for starting that over drinks and escalating to the ML 17:22:25 I'd like to get folks talking about driver composition again, as it's a major and controversial change 17:22:28 devananda: dtantsur anyone else, are there API versioning things we want to talk about or are we mostly good? 17:22:35 Big +1 for Ops Session 17:22:36 jroll, ++ useful as well 17:22:43 * jroll thanks whoever is bolding things 17:22:48 jroll, we're not good until we land a spec on it IMO 17:22:50 I would like to think that we can iron out the API versioning in a session as well 17:22:58 jroll: we should be cleaning up the docs around them -- but I'm not sure that a discussoin session is needed 17:23:00 things like when we need to bump it is very important 17:23:09 but I would not waste time on it, to be honest, as we have a de facto agreement 17:23:10 lucasagomes: dtantsur: can you add some notes on what we may want to discuss? 17:23:11 and I don't think we have an agreement about it yet 17:23:12 jroll: You're welcome 17:23:22 lucasagomes: oh hm. you think there's stuff not covered in the docs that we need to discuss? 17:23:23 jroll, sure... /me adds 17:23:29 thanks 17:23:39 jlvillal: what would that Ops Session cover? 17:24:01 I would like to hear feedback from operators on ironic. 17:24:04 jlvillal: in the past, I've tried to hold those sessions and found them either unattended or the feedback was not anything we didn't already have 17:24:08 devananda, I think things like "bump the version for backward compat changes" still in the air no? 17:24:11 What are we doing right for them. What are we doing wrong. 17:24:18 devananda: Oh :( 17:24:23 let us discuss versioning on the core party, and get seriously drunk before 17:24:30 every time it was brought up I saw many discussion but no affirmation of it 17:24:36 dtantsur: ++ 17:24:45 jlvillal: yea. I tried again at the ops midcycle and got no feedback from ops 17:24:47 TheJulia: dtantsur ++ 17:25:03 devananda: Me personally I have very little insight into how the operators are using Ironic. So I would like to now more. 17:25:10 devananda: That is unfortunate... 17:25:10 lucasagomes: with api versioning, I almost tend to think we should document the current rules / status quo, and debate from there 17:25:36 jroll, yeah 17:25:38 jlvillal: afaict, openstack operators are concerned with different things -- like how to manage database replication or their nagios alerts -- and aren't using ironic to deploy openstack yet 17:25:43 jroll, lucasagomes, it's not that we don't have anything documented: http://docs.openstack.org/developer/ironic/webapi/v1.html#api-versioning 17:25:47 jroll, ok, maybe an action note to do that? 17:25:53 devananda: Okay. 17:26:10 lucasagomes: what I mean is, let's not use a session for that :P 17:26:11 dtantsur, ok, are we good with it? 17:26:30 well.... we've lived with it for 2 cycles, I guess we can live on 17:26:32 jroll: ++ to fixing our documentation first 17:26:58 heh ok fair 17:27:11 dtantsur, ok :-) you were my biggest concern about it 17:27:21 ok, I'm going to remove the api versioning session 17:27:36 I still hate it, but c'mon... we have to move on one day, it looks like it's here to stay :) 17:27:51 dtantsur, heh ++ 17:28:00 "deploy driver using ansible" <- I think I want to see a spec before we start discussing this, is anyone opposed to me striking this one out 17:28:02 ? 17:28:10 major ++ to spec first 17:28:19 I like the idea, but indeed needs a spec 17:28:30 actually, we agreed that all technical discussions need a spec in advance, no? 17:28:37 jroll: pls to not using a session on that 17:28:53 devananda: yeah heh 17:28:59 dtantsur: roughly agreed, yes 17:29:13 dtantsur: yea I thought we did 17:29:20 it's an idea that I discussed in passing with folks maybe a year ago, but I don't see what problems it actually solves 17:29:21 I wonder if the next summit, when we have more slots, we could have a slot for subprojects 17:29:50 afaict, it's just "make a more flexible deploy driver" 17:30:57 "Features to increase reliabity and availability" IMO this is very broad, and some parts of it are way in the future. anyone +1/-1 on this? 17:31:53 jroll, well it has some hard dependencies that may take the whole cycle to be completed 17:31:54 cinder 17:32:00 yeah 17:32:01 so, not sure if it worth talk about it right now ? 17:32:12 at least for mitaka, I think it's a good session for the future 17:32:13 maybe next summit... 17:32:17 other than that one, it's just a list of things that aren't contentious and we have a good path forward on 17:32:26 that one being the N+1 thing 17:32:32 yeah, I'm going to boot that one 17:32:32 we had a session on cinder last summit -- it was great to share knowledge with everyone 17:32:45 devananda, ++ 17:32:51 but there wasn't anything contentious that I recall. so I htink that's just a matter of "lets get the things written and reviewed" 17:33:02 ya we did learn some new things in that session .. so ++ 17:33:08 that went very one, tho we didn't merge/implemented that work yet 17:33:22 more scalable / faster / better lock manager? 17:34:03 I like lock manager idea 17:34:43 are there plans to carry cinder integration on in this cycle? 17:34:56 devananda: so what's there to talk about there? upgrade is going to be weird/hard, other than that? 17:35:19 I'm going to move some stuff around since I've crossed a bunch of things out here 17:35:21 gabriel: yes 17:35:29 gariel, yes. iLo driver team is working on Cinder integration. 17:35:31 jroll: tooz or !tooz. wait for cross-project agreement or do-it-now. if we want to pick one, which one? 17:35:34 gabriel, ++, that have been requested many times 17:35:44 jroll: these are all things that several of us have different opinions on, I beliee 17:36:28 and since we have different opinions, I thnk that means it's worth hashing it out in person rather than in a spec 17:37:02 devananda: yeah, fair enough 17:37:08 I'd also suggest group management get a slot 17:37:21 we've punted on the chassis vs. tags vs. something else discussion for a while 17:37:28 oh yeah... 17:37:28 ++ for group management 17:37:33 yeah, I'm okay with that 17:37:48 devananda: oh that would be a good one 17:38:02 devananda and jroll: ilo driver team has a group managemt spec, it's udner internal review. We plan to submit a spec soon. 17:38:07 would love to either use or remove chassis obj 17:38:28 So shouldn't we have a session on cinder integration, then? Or is it already well discussed? 17:38:31 wanyen: can you get a draft up before the summit? this week would even be great ... 17:38:32 wanyen: it would be very very beneficial to have that up before summit 17:38:46 gabriel: it's pretty well discussed, people have work in progress on it 17:38:48 gabriel: I feel like it's well discussed 17:38:48 gabriel, it's been discussed already 17:38:52 :) 17:38:57 we know what needs to be done, we just need to do it 17:38:59 jroll: yes. Will sumbit it before summit 17:39:05 for both boot from volume and also mounting volumes 17:39:16 I see... thanks. 17:39:26 * lucasagomes will grab the etherpad for ya 17:39:39 wanyen: thanks 17:39:45 wanyen, cool! please make sure it's more or less in line with the ongoing work on adding tags (or that it explicitly supersedes it) 17:39:55 devananda: okay. we will try to submit it by end of this week 17:40:06 * dtantsur is really not sure how these two features are related 17:40:30 ok, so we're down to 9 sessions for 8 slots 17:40:35 progress! 17:40:46 just need to remove one more - rate of reviews thing? 17:40:46 :-) 17:40:59 I'm good with removing that and talking about it informally 17:41:05 dtantsur: it will be group management APi & framework. The implementaiton can be done by enhancing tag or use native hardware group mgmt capabilities. 17:41:13 jroll: 1 and 9 are very similar 17:41:28 devananda: ....kinda 17:41:31 wanyen, good! I'm just making sure you're aware of it 17:41:33 driver composition && driver API 17:41:38 devananda: I think there's 40 minutes worth of stuff in each one 17:41:40 they're different, but related 17:41:52 mmmmm... remotely related, I would say 17:42:03 oh, totally -- we could talk for hours on each of these topics, I suspect :) 17:42:05 devananda: and I don't think "how to improve rate of reviews" is a great summit session 17:42:13 jroll: *nod* 17:42:32 anyone opposed to removing that one? (#7) 17:42:39 jroll, we can talk about it in the ironic dinner (are we having one? ) 17:42:43 I mean the review thing 17:42:47 jroll: would "how to better coordinate reviews across the team, so we can land code faster" be better? 17:42:53 and if we don't have a dinner planned, we should 17:42:54 ++ for removing, ++ for handling it during the dinner 17:42:58 lucasagomes: nobody has set up a dinner yet afaik 17:43:05 devananda: see also "get organized" :D 17:43:16 jroll, we can talk about it in the open topics 17:43:26 lucasagomes: yeah, even friday 17:43:31 jroll: right. I don't see that in a slot 17:43:35 ok, nuking it 17:43:59 jroll: I actually like to discuss review process for 3rd-party driver spec & code 17:44:39 ok, we're 8/8 there, is anyone opposed to the list? I'll work with some folks this week as far as scheduling 17:44:46 jroll: and 3rd-party doc 17:45:12 wanyen: yes, I think there's a few things to talk about there, could you add those to the 3rd party session? 17:45:24 ok 17:45:35 wanyen: oh, never mind, that's 3rd party CI 17:45:51 jroll, as to the FRIDAY section, do you object if I put an ironic-inspector team meet up there as well? 17:45:55 wanyen: let's talk later about when/how to discuss that :) 17:46:04 we do need a chance to chat over our stuff 17:46:10 dtantsur: not necessarily, however that's a 30 person room we're sharing with infra already :/ 17:46:31 dtantsur: but inspector is part of the ironic umbrella, so mi casa es su casa 17:46:41 jroll, ironic-inspector team is a subset of the ironic team anyway (though not of the core team) 17:46:49 (my house is your house, meaning we can share) 17:46:52 yeah 17:46:58 that seems fine to me 17:47:31 I'm going to open up open discussion in case folks have other topics, we can of course continue this one too 17:47:36 thanks for everyone's help sorting these out 17:47:39 :D 17:47:43 #topic Open Discussion 17:47:59 jroll: I added under 3rd-party CI. You canre-org it 17:48:02 who wants to have an ironic dinner? 17:48:09 \o 17:48:20 \o 17:48:23 Should we add support for 'Openstack-*' headers? In addition to 'X-Openstack-*' headers? 17:48:25 o/ 17:48:27 o/ 17:48:28 \o for dinner :) 17:48:36 o/ 17:48:41 jlvillal, not before they finally land the cross-project spec 17:48:42 wanyen: thanks. I'm not sure we'll have time for it at a summit session. I don't know what topics you have for that, maybe best to start on mailing list? 17:48:49 jlvillal, if the rest of OS decides on Openstack-* I think so 17:48:55 jlvillal: yeah, we should follow API working group guidelines, generally 17:48:58 lucasagomes: ++ for dinner 17:49:07 Okay, sounds like we wait on header changes. 17:49:21 ok, and which day would suit most people better? 17:49:39 perhaps thursday ? (not sure if there's anything planned on thursday evening) 17:49:41 Tue and Wed are various parties 17:49:46 lucasagomes: tuesday and wed are bad for me 17:49:56 Thu lgtm (at least for now) 17:49:58 monday or thursday or friday should work afaik for me 17:50:05 i vote thursday, since people may have flights friday night or early saturday morning 17:50:07 Mon or Thurs 17:50:24 iirc there was a service to vote on day, no? 17:50:28 * jlvillal leaves Friday afternoon 17:50:42 dtantsur: doodle 17:50:46 dtantsur, doodle or somehting like that 17:50:47 yeah 17:50:53 jroll: mainly to see if upstream can have a lighter-weight review process for 3rd-party spec, code and doc. An perhaps allowing vendor to approve their own spec. 17:50:56 * lucasagomes creates a doodle 17:51:08 thanks! 17:51:09 lucasagomes: do you want to organize the dinner? or we could see if BadCub would mind? 17:51:11 nice, ty 17:51:16 o/ for dinner! sorry my IRC chose to fail sending that message 17:51:25 wanyen: that seems like a good topic for the mailing list :) 17:52:00 jroll: ok if we don't have time to discuss that at summit. 17:52:18 wanyen: we may have time friday, but let's start the conversation on the ML 17:52:20 jroll, if BadCup can do it, it would be better 17:52:26 * lucasagomes is not good organizing anything 17:52:42 but if not I can try to see something 17:52:43 NobodyCam: ^ mind checking with BadCub 17:52:44 badcup? lol 17:52:45 WELP 17:52:47 LOL 17:52:50 LOL 17:52:53 BadCub, you have been invoked 17:52:56 what did I do now? 17:52:59 lol 17:52:59 LOL 17:53:06 volunteered, don't worry ;) 17:53:20 why hast thou summoned me from my slumber? lol 17:53:24 BadCub: lucasagomes is going to send out a doodle about ironic dinner dates; would you mind organizing that beyond the doodle? 17:53:27 lol 17:53:35 folks seems to love your ablity to plan dinners ... they are asking for your help again 17:53:40 hehe 17:53:46 BadCub: The curse of being good at organizing things... ;) 17:53:47 it's his people skills :) 17:54:05 Sure. Just shoot me some ideas and I shall grumble my way around them 17:54:11 lol 17:54:26 hehe, thanks man 17:54:38 anytime :) 17:54:55 5 minutes left if anyone has something else? 17:55:02 * NobodyCam notes he is petrified of getting lost in toyko 17:55:18 lucasagomes: maybe worth putting a "things I refuse to eat" section on that doodle. e.g. dansmith won't eat tentacles. 17:55:19 just know.. BadCub does not do raw food or fish heads. So we gotta find multi-use cuisine 17:55:24 I suspect BadCub won't... yeah that 17:55:28 * jlvillal hasn't yet figured out how to get from airport to hotel... 17:55:36 jroll: +1 17:55:41 jroll, oh ok... will redo it 17:55:43 ditto jlvillal 17:55:45 jl 17:55:47 ++ 17:55:48 jlvillal: I hear "airport limousine bus" is recommended 17:55:50 jlv 17:55:52 ugh 17:55:54 i have heard taxis are ludicrously expensive there 17:55:55 or people can put in the comments? 17:55:57 http://doodle.com/poll/2nqbemmrdd9a5ypd 17:55:57 jlvillal: there is a train 17:55:58 BadCub: tab, not enter 17:56:00 why wont it work 17:56:05 jlvillal: http://www.rome2rio.com/s/Tokyo-Narita-Airport-NRT/The-Westin-Tokyo-Station << wont be exactly what you need, but that site is great 17:56:05 jlvillal: costs much less than the limo bus 17:56:08 yes, train.. that is what I was trying to say 17:56:12 roflao 17:56:17 lucasagomes: yeah that seems fine 17:56:24 ++ train 17:56:25 lucasagomes: put that on the ML maybe? 17:56:26 dansmith: devananda and others. Thanks for the tips on transport 17:56:28 * BadCub frowns at empty coffee cup 17:56:31 jroll, will do 17:56:31 I think there's even a direct train there 17:56:49 thanks 17:57:12 the train stops like acorss the park from the hotel 17:57:28 *host hotel 17:58:05 thanks for the tips :) 17:59:57 alright, I'm gonna wrap this up 18:00:02 great meeting, all, thank you! 18:00:04 cool thanks all 18:00:05 ok thanks all :) 18:00:05 thank you all o/ 18:00:07 #endmeeting