16:00:01 #startmeeting cinder 16:00:01 Meeting started Wed May 13 16:00:01 2015 UTC and is due to finish in 60 minutes. The chair is thingee. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:02 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:04 The meeting name has been set to 'cinder' 16:00:22 Hi everyone! 16:00:27 howdy 16:00:28 Hi! 16:00:28 Hi 16:00:29 Morning! 16:00:29 hi 16:00:35 hi 16:00:35 Hi 16:00:43 Hello. 16:00:47 evening:) 16:00:50 hi 16:00:59 Well my laptop decided to just not work at this convenient time. Reporting from my phone. 16:01:05 Yay technology. 16:01:12 hi 16:01:26 Hello! 16:01:30 Not another summit without a laptop. :) 16:01:31 lo 16:01:33 Usually I start with announcements but I'll do those last, hopefully once my laptop is working. 16:01:50 thingee: Summits are bad for your laptop. 16:01:54 HP do some nice tablets.... 16:02:03 I prefer Dells. 16:02:04 Let's start with the first agenda item for the liberty mid cycle meet up. 16:02:16 o/ 16:02:19 Swanson: Thinkpads all the way! 16:02:19 Ok 16:02:21 #topic liberty mid cycle meetup results 16:02:23 thingee: http://goo.gl/E4DoUy 16:02:24 DuncanT: hi 16:02:53 Survey says: Dates are nearly tied, with a slight preference for the earlier date 16:02:55 jgriffith: lol 16:02:59 hello all 16:03:28 Location is also nearly tied, FC wins by two votes 16:03:39 sweet! 16:03:39 were are the results? 16:03:47 DuncanT: How many times did you vote. ;) 16:03:55 smcginnis: zero 16:04:12 what about wifi? :) 16:04:12 yo 16:04:22 rhedlind: :-p 16:04:30 rhedlind: We know it works in FC at least. 16:04:31 asselin: I'll post them up to a spreadsheet later, I don't seem to be able to just make them public 16:04:43 where are the results ? 16:04:50 Wifi is apparently either very important, not funny or a reason for tears 16:04:52 DuncanT, how about openstack paste? 16:05:00 asselin: Working on it 16:05:20 Jennifer was glad to see you hadn't forgotten she couldn't get the wireless working. 16:05:37 where r the results ? are we planning to have hangout sessions or some remote video conf session for people who will be remote and can't make it 16:05:41 DuncanT: There was one vote that it wasn't required. 16:05:43 ;-) 16:05:54 :P 16:06:00 * thingee has his laptop back 16:06:03 vilobhmm1: I am sure we will do Hangout again. 16:06:13 thingee: Yay! 16:06:18 that helps..so that everyone can learn and contribute 16:06:19 jungleboyj: Two people say we should abuse your phone bill again 16:06:26 #action DuncanT will post results and link to it from the original OpenStack dev ML post 16:07:06 So when will a decision be made? Do we need to look at the survey first? 16:07:16 ok great so Fort Collins, and aug 3-5? 16:07:34 DuncanT: +2 16:07:40 and no wifi, no beer? ;) 16:07:44 There was a query on the etherpad for extending an extra day. 16:07:52 scottda: that was me 16:07:58 winston-d: Bite your tongue! There must be beer! 16:08:01 +1 on an extra day 16:08:05 wait, hold on...no beer? 16:08:07 wth 16:08:09 thingee: Optional extra day would be nice. 16:08:10 I thought it would be great if just like at the summit, we had a day for sprints 16:08:12 It's fine from a standpoint of hosting to extend to Aug 6th. 16:08:30 while we're all motivated after discussions 16:08:51 Don't spread rumours about running out of beer. That will not be a problem. 16:08:56 I think the only person that couldn't make the early date was jungleboyj \ 16:08:59 o/ 16:08:59 :( 16:09:16 To be clear, are we going to do like we did in the past where Monday is a day to travel and the real action is 4,5,6 and maybe 7? 16:09:35 I like that plan. 16:09:52 +1 to 4-6 plus maybe 7th 16:09:53 jungleboyj: +1 16:10:24 Ok, that helps. Hopefully we can get the girl watching my boys to do one over night so I can fly out Monday. 16:10:32 isn't Friday a fishing day in Fort Collins ? 16:10:33 jungleboyj: +1 16:10:41 Every day is a fishing day 16:10:42 Worst case scenario I will have to come in on Tuesday. 16:10:54 scottda: +1 16:11:09 ok so sounds like people are fine with these dates? 16:11:09 scottda, +1 16:11:14 and location 16:11:24 assuming we have someone who can host in the area again 16:11:30 so who is hosting in Fort Collins ? 16:11:33 :P 16:11:36 thingee: I will make it work. 16:11:36 I've planned on it 16:11:42 hosting, that is 16:11:46 New Belgium brewery? 16:11:46 lol shouldn't that have been the first question? 16:11:54 I've got a room booked already 16:11:59 scottda, sweet 16:12:02 scottda: ha 16:12:04 scottda: we can arm wrestle over it :) 16:12:04 why not hawaii then we'll find someone to host? 16:12:05 lol 16:12:22 ameade, +A 16:12:24 ameade: not sure if jgriffith can surf. 16:12:34 jgriffith: You want to host in Boulder? 16:12:41 OR in Fort Collins? 16:12:42 thingee: as a matter of fact, I used to surf quite a bit :) 16:13:00 whoa cool 16:13:02 scottda: I'd love to be able to offer something up, but HP in FTC is a way better setup 16:13:20 jgriffith: you do have great divide where you're at.. 16:13:22 sort of close 16:13:24 ;D 16:13:24 OK. I'm not trying to hog the hosting. 16:13:26 scottda: was thinking Boulder would be nice/different and still close, but we don't have the same facilities you do 16:13:57 scottda: I was kidding, I think you did a great job last time and we'd be stupid not to take advantage of you again :) 16:14:11 jgriffith: ++ 16:14:17 * scottda likes to be taken advantage of, but that's a different story.... 16:14:20 scottda: yeah I was happy with the HP's hosting last time. Good location 16:14:23 and we had network! 16:14:29 jgriffith: +10000 16:14:32 * jungleboyj sighs 16:14:40 :) 16:14:48 ok great. so this seems like we're fine. hosting details seem to be in the works 16:14:50 * jungleboyj starts looking into wifi jammers. ;-) 16:14:50 cruise ships have wifi 16:14:51 Ok, I'll post up more details on the etherpad, and I'll book blocks of rooms at HP discount at local hotels this time. 16:15:04 scottda: ++ 16:15:13 We can put jgriffith's horse barn as a backup plan. :P 16:15:15 So should we plan to stay through Friday? 16:15:30 smcginnis: oh yeah, good point 16:15:31 work through Friday ? 16:15:35 Friday is a hackathon day ? 16:15:40 * DuncanT might not be able to stay friday :-( 16:15:58 smcginnis: I was hoping I would find an airbnb for jgriffith's farm. Just like a clean lump of hay in the barm or something 16:16:08 full week might be a bit of a burden for some folks 16:16:09 thingee: Hah! 16:16:13 hemna: yes sprints 16:16:20 Ok. 16:16:21 hemna: I hate hackathon now. :P 16:16:35 sprints is the new hackathon ? 16:16:39 jgriffith: the friday is completely optional 16:16:40 beerathon 16:16:41 man I'm so behind in my vernacular 16:16:49 winston-d: ++ 16:16:50 thingee: cool... I get it 16:16:52 winston-d, fish-a-thon 16:17:01 you guys can work, I'm hitting the river.... 16:17:02 :P 16:17:07 hemna: :) 16:17:41 #agreed Fort Collins Aug 4-6. 7th is optional sprints 16:18:01 #action thingee announce to the OpenStack Dev ML 16:18:28 #topic Social Monday Night at the summit 16:18:37 #link http://lists.openstack.org/pipermail/openstack-dev/2015-May/064024.html 16:19:01 jgriffith brought this up from the ML today. Are we interested in having our own get together? 16:19:23 +1 16:19:26 at the summit yes +2 16:19:35 thingee: +1 Especially if there is nothing else going on. 16:19:42 +1 16:19:45 no OS parties Monday night ? 16:20:09 and if so, who can have their resources do planning/reservations? I'm spent for summit planning. 16:20:14 hemna: everybody wait for Tuesday's party:) 16:21:04 got quiet 16:21:11 +1 16:21:39 Who has the biggest hotel room? :) 16:21:40 Just everyone grab some 40's and meet down by the harbor. ;) 16:21:47 thingee: +1 on monday night social 16:22:06 smcginnis: That sounds like fun. 16:22:11 +1 on the idea, I can't vollenteer to org rhough 16:22:34 I remember a good microbrewery in the gaslight district, can't remember the name though 16:22:36 How many people are we talking? 16:22:51 inderesting idea, something new for me, so I +1 on it 16:23:17 Maybe an etherpad to get a head count, post location, etc? 16:23:27 scottda: Not a bad idea. 16:23:28 That would work. 16:23:46 scottda: http://attending.io/events/new someone mentioned 16:23:47 scottda : good idea! 16:23:55 maybe we can do that after count. 16:24:17 just was hoping someone would spearhead this 16:25:32 what is the estimated group size looking like? 16:25:34 spearheading does not mean you're paying. If vendors want to sponsor this event, that's cool. I guess I sort of saw this as us hanging out and catching up before the summit. Like we did at the last midcycle 16:25:50 thingee: ++ 16:25:51 * DuncanT can look for a venue on sat eve once I'm there if needed 16:26:01 I'll count up the people and try to book a room or something at a local pub 16:26:10 scottda: cool 16:26:15 Maybe I can work with Duncan 16:26:18 https://etherpad.openstack.org/p/LibertySocialMondayCinderEvent 16:26:20 scottda: Thanks! 16:26:35 sure. Beer motivates me, if you couldn't already tell. 16:26:40 I think people all have their own expenses they can tap from. Maybe some people will buy rounds depending on group size. 16:26:58 Just got this as one possible idea: http://the-roof.ca/ 16:27:21 So people post ideas of places we can reserve to the etherpad 16:27:28 Will do. 16:27:32 #link https://etherpad.openstack.org/p/LibertySocialMondayCinderEvent 16:27:33 I think there's an hp gathering going on monday night 16:27:55 asselin: meet us after? :) 16:28:02 asselin: can we crash it? 16:28:15 :) 16:28:17 #topic announcements 16:28:41 #idea New drivers for Liberty need to be in by June 19th with CI. 16:29:20 I guess that's more of a topic, but this sticks to our previous planning in previous releases 16:29:50 questions? comments? concerns? 16:29:53 thingee: 6/19 is the merge date? 16:29:58 thingee: I had someone asking about that earlier and that was about what I had estimaed. 16:30:03 *estimated 16:30:13 xyang1: yes needs to be merged. 16:30:59 anyone opposed to the date? 16:31:15 thingee: Is there a driver code submission deadline? 16:31:16 Is that a deadline for brand new drivers, or for feature implementation of existing drivers as well? 16:31:21 so just to be clear, in order to +A a driver, it has to have CI working and reporting on the driver patch itself right ? 16:31:22 thingee: could big queue in zuul be an issue for it before L-1 deadline? 16:31:33 thingee: I'm not sure TBH 16:31:34 hemna: yes 16:31:43 thingee, ok good. that's what I was hoping. 16:31:47 thingee: I'd also really like to figure out what we're doing going forward 16:32:02 thingee: IMHO having 50+ drivers in Cinder is not going to be sustainable 16:32:13 e0ne: definitely. so owners to the patches shouldn't wait 16:32:31 jgriffith, well, I think that can be a separate topic, but until we find a viable solution, we should stick to this no? 16:32:32 thingee: if you look at the stats on patches/bugs/review queue right now it's already greater than 85% drivers 16:32:32 jgriffith: agreed. I actually forgot about the talks about out of tree for the summit. 16:32:50 jgriffith: but I didn't have time to review previous spec on the idea when neutron did it. 16:32:54 hemna: my point is we have talks planned at the summit so why make a decision now? 16:33:15 hemna: thingee I guess it doesn't hurt 16:33:19 sure, I think the only decision at this point is to stick to the existing plan. 16:33:28 hemna: fair enough 16:33:35 thingee: do you have any plans to make current 3rd party CIs more effective? 16:33:53 jgriffith: I wanted something in the air now, so there is time for people to make arrangements. I agree we should have something planned at the summit to discuss this 16:33:55 hemna: +1 16:34:03 e.g. Oracle CI voted on my patch after about 4 days when it posted 16:34:04 Good topic after a couple of beers. :-) 16:34:08 thingee: that makes sense 16:34:20 e0ne, there's quite a few infras sessions to get and common openstack ci solution scheduled 16:34:20 e0ne, doh 16:34:24 e0ne: I'm more concerned with stats for those that regularly vote 16:34:44 jgriffith: +1 16:34:44 e0ne: by stats I mean false fail rate 16:35:01 e0ne: thingee I'm trying to figure out how to get radar to work and make it a req 16:35:14 jgriffith: how about I announce that we're looking at this being the date, but we'll have a final decision at the summit? 16:35:17 but node is being a PITA for me... maybe we should look at patrickeast 's solution 16:35:18 jgriffith: How about a cinder-drivers repo with a more open (larger) core? 16:35:25 thingee: I think that's great 16:35:43 DuncanT: yeah, I mean hell... if we have a repo for brick WTF not? 16:35:48 jgriffith: we need to try get a list of requirements to CI. maybe on Friday at Summit? 16:35:50 xyang1: do you mind proof reading my announcement to make sure it's clear? :) 16:35:55 err... WTH (why the heck) 16:36:06 thingee: Sure 16:36:10 I think I had more questions from you on the announcement for Kilo, so your feedback would be great. 16:36:17 jgriffith: repos are cheap... it seems better than totally out-of-tree 16:36:29 thingee: Ok 16:36:49 DuncanT: dont' disagree, need some thought/discussion though IMO which we have slated for Summit 16:36:50 DuncanT: +1 16:36:55 DuncanT: +1 16:36:57 DuncanT, I'm not clear what that buys us really. 16:37:13 hemna: let's people like me ignore that repo :) 16:37:18 #agreed Final decision for drivers to be submitted in Liberty will be decided at the summit. So far June 19th is a possible date 16:37:35 hemna: seriously though... it doesn't buy a ton 16:37:40 jgriffith, can't you ignore reviews in volume/drivers now ? 16:37:40 hemna: We can have more people with +2 on that repo than cinder core 16:37:41 #action thingee will announce June 19th as being a possible date, and the arrangements for discussion at the summit 16:37:58 DuncanT: do you have any feedback about out-of-tree drivers from nuetron? 16:37:59 hemna: Essentially encourage people to review each other's drivers 16:38:01 anyway, just thinking out loud sorry. 16:38:06 there's some disucssions going on in Nova about splitting core responsibilities...I think there's a session scheduled for it 16:38:08 So other announcements.. Cinder specs! 16:38:17 keep on reviewing. 16:38:26 e0ne: Neutron is a mess re code quality and functional consistency 16:38:27 we can save it for the summit, but I'd like to come up with a way to incentivize more "community" participation 16:38:37 patrickeast just posted a generic image caching spec https://review.openstack.org/182520 16:38:50 please to be reviewing ^^ 16:39:19 jgriffith also needs feedback before the summit on replication v2 spec https://review.openstack.org/155644 16:39:24 also to be reviewing ^^ 16:39:59 ++ 16:40:03 thingee: while you're asking :) 16:40:04 https://goo.gl/TrrLuF 16:40:38 jgriffith: ah yes, I have a vm waiting for me to test that 16:40:54 #link https://libertydesignsummit.sched.org/event/179f43bed9fb2afdb3999c13d9a39480#.VVN-jvlVhBc 16:40:55 saw your message when I was already reading in bed :) 16:41:17 Cross-project: in-team scaling link ^^ 16:41:46 we also have a fish bowl session at the summit for discussing Volume performance and Health collection from backends 16:41:48 #link https://review.openstack.org/#/c/178748/ 16:42:01 need vendors to give feedback to help adurbin_ with the session 16:42:19 don't wait until the session please. 16:42:29 ok I'll check it out 16:43:11 ok, so next up 16:43:12 removing v1 16:43:21 I have the patch up. it fails everything 16:43:42 there are a lot of little areas here and there on different projects that are doing weird things are default to v1 16:43:44 ship it! 16:44:04 I'm looking for people to help me on getting things ready in other projects 16:44:07 like heat, nova, etc 16:44:14 openstack client 16:44:20 thingee: sorry, I stepped out for a second 16:44:30 thingee: what about version descovering in cinderclient? 16:44:40 e0ne: it's shipped 16:44:45 1.2.1 16:44:54 thingee: great! 16:44:59 thingee: I have some ammendments to make to https://review.openstack.org/#/c/178748/, i'll try to get them up this week 16:45:20 so if anyone wants to help me with the v1 removal effort, please contact me 16:45:36 thingee: i'll ping you:) 16:45:40 other great news, cinder is now using os-brick! 16:45:43 yay hemna ! 16:45:44 https://review.openstack.org/155552 16:45:47 :) 16:46:00 hemna: got the nova patch handy? 16:46:05 to switch it to os-brick? 16:46:13 https://review.openstack.org/#/c/175569/ 16:46:19 #link https://review.openstack.org/155552 16:46:19 that's my WIP nova patch 16:46:29 #link https://review.openstack.org/#/c/175569/ 16:46:43 hemna: what's the plan so far? 16:46:51 there are some os-brick patches I'd like to see land 16:46:57 hemna: when do we expect this to be ready for review/testing 16:47:15 I was hoping to get it done before Vancouver 16:47:19 but I'm running out of time 16:47:21 does this mean we should have our ci systems start testing os-brick changes? 16:47:24 hemna: do you have some kind of roadmap/blueprints/etc? 16:47:29 patrickeast, +1 16:47:31 e0ne, not yet 16:47:41 patrickeast: hemna and I were going to have a discussion about os-brick testing 16:47:44 never happened 16:47:46 patrickeast, asselin and I have started thinking about CI'ing os-brick patches 16:47:59 basically we need every os-brick patch to kick off cinder 3rd party CI tests as well 16:48:05 patrickeast: oh crap, really? 16:48:07 making sure we aren't breaking cinder 16:48:15 patrickeast: that's a side effect I didn't think of 16:48:20 its easy enough to add it as a project to watch in zuul 16:48:32 jgriffith: sort of simple 16:48:42 ok, can I just say that's kind of crazy IMO 16:48:52 CI brick and Cinder? 16:49:01 making sure we don't break things ? 16:49:07 hemna: at least, it would be great to run tempest over the each brick patch 16:49:08 i guess the alternative is to wait for the cinder changes where we rev the brick version, right? 16:49:09 hemna: ineffeciency 16:49:23 patrickeast: yeah, but that's probably not a great option :) 16:49:44 All of the brick code used to be in cinder, so there shouldn't be a significant change in the number of changes... 16:49:45 hemna: what ROC do you expect to see in the lib? 16:49:57 ROC ? 16:49:59 it should be minimal right? Otherwise it shouldn't be a lib 16:49:59 ROC? 16:50:05 Rate of Change 16:50:05 rate of change 16:50:06 Rate Of Change 16:50:10 heh 16:50:14 :-) 16:50:15 it shouldn't be changing much 16:50:21 Wouldn't that entail that we (our CIs) would need to pull down Cinder master and brick patch to test it. 16:50:25 * jungleboyj Heard it come out in harmony 16:50:31 there is a bit of changes we need to add to it, due to features that made it into Nova that didn't get into Cinder 16:50:32 That's not a trivial change from how it works right now. 16:50:35 DuncanT raises a good point. The change would've been triggered in the Cinder project. The change is just in a different project, so it's not really adding more to CI's 16:50:44 hemna: thingee so actually this is a non-issue 16:50:45 it's just you have another project to watch 16:50:50 There are a few bugs I filed about those, but after that, the ROC shouldn't be much at all IMHO 16:50:53 smcginnis: need to take a look how oslo libs are tested 16:50:59 hemna: thingee rules for upgrading brick in Cinder are just like any other req lib 16:51:07 hemna: thingee needs a patch to reqs file 16:51:07 e0ne: True. I would see this as the same. 16:51:16 ok 16:51:19 you try and update Cinder, shit fails, you fail the merge/patch 16:51:21 smcginnis: because cinder and nova will use brick not from master 16:51:34 e0ne: exactly 16:51:59 sorrison, we need to pin the requirements.txt to a specific version then ? 16:52:02 gah 16:52:03 so 16:52:07 jgriffith: finding what broke when updating requirements is cinder will be fun 16:52:09 git bisect 16:52:10 hemna: well, yeah. 16:52:11 :) 16:52:11 right now it's open to any version. 16:52:12 and that's why devstack installs it from pypi, not from git master 16:52:22 global-requirements.txt just says 'os-brick' 16:52:26 no versioning 16:52:31 hemna: we can change that 16:52:35 thingee: no different than oslo, keystone and the giant list of libs we have currently 16:52:40 jgriffith: true 16:52:40 thingee, yah I think we should 16:52:45 hemna: well that's the first patch brick needs then 16:52:47 hemna: it doesn't metter while we have only one release 16:52:56 hemna: you can't have an externallib with no versioning 16:53:00 jgriffith, yup. 16:53:07 jgriffith, should add that today. 16:53:26 ok last announcement... 16:53:28 hemna: cool 16:53:40 LVM thin type 16:53:42 https://review.openstack.org/#/c/182498/ 16:53:42 hemna: I asked that earlier in your patch adding os-brick to requirements.txt 16:53:50 https://review.openstack.org/#/c/182499/ 16:53:58 https://review.openstack.org/#/c/182528/ 16:54:34 winston-d, we are ok right now. we only have 1 release of os-brick in pypi, with nothing planned in the short term. we can update global-requirements to pin it to 0.1.0 for now and then update cinder. 16:54:43 jgriffith is adding this option to devstack, add an experimental gate job 16:54:46 hemna: cool 16:55:22 jgriffith: what improvements do you see this bringing to gate? 16:55:26 hemna: we'll have to figure out that stuff later I think 16:55:34 thingee: LVMThin? 16:55:36 jgriffith, yah 16:55:41 thingee: it's better 16:55:50 thingee: faster 16:55:56 better snaps etc 16:56:07 and it works! :) 16:56:13 and ideally I'd like to see it made the default for Cinder in the future 16:56:22 #idea LVM Thin experimental job in gate. Better, Faster, Stronger. 16:56:25 thingee: main thing is we keep piling in options and we don't test them 16:56:30 #link https://review.openstack.org/#/c/182498/ 16:56:38 we either need to remove options, or test them propertly IMO 16:56:40 #link https://review.openstack.org/#/c/182499/ 16:56:45 and probably a little of both 16:56:49 #link https://review.openstack.org/#/c/182528/ 16:56:57 thingee: LOL 16:57:00 BFS 16:57:40 those are the reviews. interested folks, please show support ^ 16:57:49 3 minutes reminder 16:57:58 thingee, jgriffith e0ne https://review.openstack.org/182756 16:57:58 #topic open discussion 16:58:23 Have the fishbowl and working sessions been given timeslots in Vancouver? 16:58:31 cebruns: yes 16:59:02 Ok - I looked for them but didn't see 'em in the schedule. Is there a link? 16:59:28 cebruns: http://libertydesignsummit.sched.org/type/design+summit/Cinder 16:59:49 thingee: Thanks! 17:00:01 fwiw, work session topic in most of those names is on purpose. 17:00:05 #endmeeting