16:00:05 #startmeeting cinder 16:00:06 Meeting started Wed Sep 23 16:00:05 2015 UTC and is due to finish in 60 minutes. The chair is thingee. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:10 The meeting name has been set to 'cinder' 16:00:12 o/ 16:00:15 o/ 16:00:16 Hi! 16:00:17 hey all! 16:00:17 hi 16:00:18 mep 16:00:19 hi 16:00:21 hi 16:00:24 Good morning. 16:00:29 Hello! 16:00:30 Hello :) 16:00:39 o/ 16:00:43 hi 16:00:47 o/ 16:00:51 hi 16:00:52 Today is my last time in this cushy chair in the meeting! 16:00:56 it has been a fun ride 16:00:58 :( 16:01:03 :'-( 16:01:05 :( 16:01:12 sniffles 16:01:18 maybe I'll fell in for the next PTL sometimes, who knows! 16:01:24 fill* 16:01:42 all is good though :) 16:01:51 Thanks for the last year! 16:01:56 thingee: if you think you're going to ride off into the sunset you're mistaken :) 16:02:00 diversity is good in leaders 16:02:07 jgriffith, +1 16:02:08 :) 16:02:29 #topic announcements 16:02:50 So I don't really have any... 16:02:56 oh! 16:03:07 RC1 should be cut today 16:03:16 #info Cinder RC1 should be cut today 16:03:17 Isn't master opened for Mitaka? 16:03:17 #link https://review.openstack.org/#/c/226850/1 16:03:35 dulek: yes that would mean mitaka would be open after that patch lands and ttx does the cut 16:03:53 :) 16:04:02 thingee: so there won't be an rc2 then correct? 16:04:18 thingee, are we pushing os-brick 0.4.1 for L ? 16:04:18 we need to not forget remove -2 after master unfreeze 16:04:21 diablo_rojo: there could be, if users, ops, other tests notice issues with rc 1 16:04:28 hemna: 5.0 I believe now. 16:04:48 thingee: That would be off of stable/liberty though, right? 16:04:50 hemna: os-brick 0.5.0 will be available 16:05:00 hemna: it should be os-brick 0.5.0 16:05:04 thingee, ok coolio. I didn't see the patch up for it 16:05:24 smcginnis: the upper constraints was set, but not requirements.txt for os-brick 0.5.0 16:05:25 hemna: Doug Hellmann anounced it yesterday in openstack-dev ML 16:05:32 thingee: Ah 16:05:38 just keep that in mind 16:05:38 e0ne, sweet. 16:05:56 nova has an rc2 for translations, not sure if cinder is in that same boat 16:06:15 mriedem: not...yea :D 16:06:17 mriedem: probably 16:06:29 /yea/yet/ 16:06:38 (depends how much work the translators will actually put in the coming days :) 16:06:51 any other announcements before we move on? 16:07:12 < crickets > 16:07:23 o/ 16:07:33 I did want to give a quick plug while I have everyone's attention... 16:07:54 so I know you all look forward to the openstack newsletter weekly! 16:08:04 and there is a new section to help with a common problem people give.. 16:08:09 http://www.openstack.org/blog/2015/09/openstack-community-weekly-newsletter-sept-12-18/ 16:08:21 "What you need to know from the developer’s list" 16:08:36 +1 16:08:44 nice 16:08:47 summary of the dev list... quick important points to keep you updated without having to read the entire thread 16:08:49 thingee: that's a great addition 16:08:56 +1 16:09:05 Sounds great! 16:09:31 give me feedback please! :) 16:09:35 ok lets gets started! 16:09:54 agenda for today! 16:09:57 https://wiki.openstack.org/wiki/CinderMeetings#Next_meeting 16:09:59 short 16:10:07 maybe with design summit topics 16:10:17 oh and nova volume bug by mriedem 16:10:20 got it 16:10:31 #topic Cinder API races bugs 16:10:33 geguileo: hi 16:10:36 Hey 16:10:43 Sorry, forgot to remote that 16:10:44 too late 16:10:47 geguileo: :) 16:10:47 :) 16:10:53 We agreed in cinder channel to not include it 16:10:57 #info first patch 16:10:59 #link https://review.openstack.org/#/c/218012 16:11:14 So nothing to discuss really 16:11:21 #info This follows dependency chain 16:11:23 #link https://review.openstack.org/#/c/221442 16:11:26 oh 16:11:34 #info nothing to discuss 16:11:42 :) 16:11:48 Sorry 16:11:51 no prob 16:12:12 #agreed in the cinder channel it was agreed this would not be in L 16:12:25 #topic Nova volume bugs 16:12:28 mriedem: hi 16:12:36 hey, so just a quick note about nova volume bugs 16:12:41 #info there are lots of nova/cinder bugs 16:12:44 https://bugs.launchpad.net/nova/+bugs?field.tag=volumes 16:12:45 #link https://bugs.launchpad.net/nova/+bugs?field.tag=volumes 16:12:45 yeah 16:12:50 i was poking through those last friday 16:12:56 it's at 102 16:13:06 i suspect some are closed now 16:13:11 some could be issues in os-brick 16:13:11 mriedem: I actually went through a bit of nova bugs to triage originally to that tag. 16:13:14 :) 16:13:15 mriedem, thanks for raising this one. I'm starting to go through that list. I think some of them might already be fixed using os-brick 16:13:36 i was also noting how many multipath bugs there were 16:13:42 which i'm assuming is in no way CI tested 16:14:00 mriedem, some of the cinder 3rd party CI tests use mpath 16:14:01 some don't 16:14:01 my ci tests multipath FC and iSCSI... not sure if anyone else is though 16:14:14 #info some might actually already be fixed. Maybe also since os-brick. 16:14:15 ah, that's good to know 16:14:26 but I don't think any of the multipath tests actually look at the state of the host after the detach is done 16:14:27 b/c there was a bug i think that actually said fc attach with multipath failed 16:14:45 the problem i have with looking at some of these is i don't have an environment to recreate 16:14:55 so was hoping to lean more on the cinder people that had these environments 16:14:59 mriedem, I'm working on some os-brick patches that will allow us to test before detach and after detach to make sure the volumes are gone. https://review.openstack.org/#/c/199764/ 16:15:41 #info mriedem noticed many multi-path bugs. patrickeast has Pure CI testing multi-path with FC and iSCSI 16:15:49 mriedem, I'd like some feedback on the spec if possible. that would help cover one of the missing pieces. 16:16:12 we can test iSCSI and FC multipath here 16:16:27 we also test iSCSI and FC multipath 16:16:38 ok 16:16:41 i don't have anything else really 16:16:45 was just raising awareness 16:17:01 since scottda is the nova liason, he and I can start looking through those bugs as well. 16:17:08 yup 16:17:12 * hemna volunteers scottda for more work :) 16:17:19 thanks 16:18:01 scottda: might be good to start prioritizing what we want to focus on for Mitaka. I know some of these already come from your initial list of bugs. 16:18:02 scottda, no problem, I was pretty sure you were getting bored anyway :) 16:18:09 hemna: I also noted that we could get some of my team helping out on those as well. 16:18:27 scottda: prioritize and a group...we might not be able to do all 100 bugs... but some managable goal for Mitaka I think would be good 16:18:50 Yes, We'll gather bugs , prioritize and publish . I'll get something going by next week's meeting. 16:19:23 #action scottda to priortize and set some expectation for Mitaka 16:19:26 mriedem: anything else? 16:19:30 scottda: Would you keep me in the loop? I'm the nova bug czar since 2 months or so. 16:19:36 https://etherpad.openstack.org/p/Mitaka_Priorites_for_Nova_Volume_bugs 16:19:53 #info markus_z would like to stay in the loop, being the nova bug czar 16:19:54 sure markus_z 16:20:00 scottda: thanks 16:20:04 thingee: nope 16:20:07 thanks 16:20:25 #topic Design summit topics 16:20:29 Regarding bugs, I have this one in the pipeline: https://review.openstack.org/#/c/209847/ 16:20:33 #link https://etherpad.openstack.org/p/cinder-mitaka-summit-topics 16:20:35 whops, too late :( 16:21:16 so we got some agreements from last week 16:21:35 let me just skim action items from last meeting in this area... 16:22:08 markus_z: I like that proposal BTW 16:22:22 patrickeast: you have an action for making an etherpad on experimental API research? 16:22:45 jgriffith: thanks, I'm currently figuring out how to make that an project wide blueprint. 16:22:59 patrickeast: didn't see a link for it though on the topic etherpad 16:23:35 cricket 16:23:40 ok, we'll circle back 16:24:13 dulek: you have an action for an etherpad for nova AZ issues 16:24:15 thingee: ah, sry, stepped away for a minute.... I do still intend to do that, have been side tracked with some other bugs 16:24:34 patrickeast: ok, I'll make a note for the next meeting organizer 16:24:51 #action patrickeast to make etherpad on expermental API research 16:25:12 wake up everyone :) 16:25:19 thingee: Yeah, that haven't happened, I was waiting for DuncanT to post his findings and discussion. 16:25:37 DuncanT: how's that coming along? 16:25:55 thingee: Added it to TODO for tomorrow morning, DuncanT will add there I think. Sorry about that! 16:26:14 dulek: no problem, just making sure we're ready for the summit! :) 16:26:21 patrickeast: ping me when you do that 16:26:40 patrickeast: me too 16:26:45 #action dulk to post an etherpad link under the topic proposal availability zone issues. 16:26:55 bswartz: scottda: will do 16:27:02 #action DuncanT to post his findings and discuss with dulek 16:27:06 * bswartz is already waist-deep in experimental APIs and their issues 16:27:23 rhedlind and eharney how is the driver api doc coming along? 16:27:41 thingee: not really moving yet, but i'll be getting things going more this week 16:28:00 ok, next meeting we'll circle back 16:28:14 #action rhedlind and eharney to collaborate on a driver api doc for cinder dev docs 16:28:32 hemna: renewing spec return VolumeIsBusy for ing states 16:28:44 https://review.openstack.org/#/c/149894/ 16:28:52 I started updating that guy yesterday 16:28:59 very nice! 16:29:02 working on some feedback from dulek and will push up a new version today. 16:29:18 #info spec renewed for VolumeIsBusy with ing states 16:29:21 #link https://review.openstack.org/#/c/149894/ 16:29:50 more reviewers would be good 16:29:54 scottda: you're still set to present that spec to Nova in their next meeting :P 16:29:58 We need a Nova spec for that as well 16:30:10 hemna: I'll be around, but may reply with delay. Anyway - ask what you need. :) 16:30:19 Yes, but I'd like to have a Nova spec ready for Nova meeting. 16:30:26 scottda, ok I can put that together. 16:30:52 cool, thx. Perhaps mriedem has input or can help with the NOva spec? 16:30:57 #action hemna to put together a nova spec for scottda to present VolumeIsBusy being returned for ing states at next nova meeting 16:30:59 whew 16:31:01 thanks 16:31:15 scottda, yah I was hoping to get some feedback/help from mriedem on that as well 16:31:24 i remember -1ing a cinder spec at one point 16:31:29 jgriffith: POC's for ABC class proposals 16:31:45 mriedem, heh yah, that url ^^ 16:31:52 mriedem: We will have cinder-api-microversions for this. YOur -1 was because of API changes 16:32:01 IIRC 16:32:06 thingee: nothing this week 16:32:10 jgriffith: I think we wanted this ready for your topic at the summit 16:32:10 * jgriffith still has a month :) 16:32:22 jgriffith: just checking in 16:32:26 roger 16:32:49 ok, I won't circle back :D 16:33:31 jgriffith: I'm assuming you had discussions with eharney on that and your other topic of should cinder be the next vipr? 16:33:59 thingee: not really... will need to try and link up with him next week 16:34:10 thingee: I think we're both a bit busy this week 16:34:15 jgriffith: ok just reminding :) 16:34:18 :) 16:34:20 will do 16:34:24 summit will be here before you know it 16:34:29 thingee: good thing, because I will certainly forget again :) 16:34:34 ok lets look at new topics 16:34:39 thingee, so do we only have 2 fishbowls planned (according to the etherpad ?) 16:34:41 https://etherpad.openstack.org/p/cinder-mitaka-summit-topics 16:34:48 hemna: yes! 16:35:07 so we have four fish bowl slots... we only have two proposed which I agree are fine..just need to have etherpads ready ;) 16:35:08 jgriffith: If I can find the time I would be interested in helping with the ABC/zope/etc POCs 16:35:16 dulek: dulek patrickeast ^ ;) 16:35:20 smcginnis: excellent 16:35:39 Wondering if "Future of Cinder, and getting back to our roots" should move to a fishbowl? 16:35:47 lets take a look at working sessions 16:35:56 jgriffith: I will be glad to discuss next VIPR too 16:36:14 So Milestone deadline ... not sure who put that up... could be a sprint 16:36:17 e0ne: noted 16:36:20 jgriffith: my bo cinder-without-nova could be related 16:36:27 anyone have thoughts? 16:36:48 e0ne, a session on attaching cinder volumes w/o nova could be good 16:36:52 i think jgriffith had a different one about deadlines below 16:36:58 thingee: we could discuss it in a cinder weekly meeting, imo 16:37:00 #info working session proposal: milestone deadlines 16:37:03 eharney: yes, I did 16:37:11 e0ne: I agree. 16:37:18 hemna: will you add a proposal? 16:37:18 I have two agreements 16:37:23 going... 16:37:25 going... 16:37:27 I don't know where that one came from either... but not sure it needs a working session slot 16:37:28 thingee: I think that is a good place for that discussion. 16:37:38 milestone deadlines is not my proposal but do have a proposal related to milestone deadlines -- I'll make a ML post 16:37:45 jungleboyj_: That being weekly meeting? 16:38:07 smcginnis: I was also wondering what he was agreeing to :) 16:38:18 :) 16:38:23 smcginnis: or sprint, don't need a whole session on it. 16:38:31 thingee: smcginnis Sorry, too many things going on at once. 16:38:45 #agreed moving milestone deadlines to sprint 16:38:52 thingee: ++ 16:38:53 jungleboyj_, a sprint session would be good for it. I'd like to see any/all deadlines decided upon asap and published. 16:39:03 hemna: ++ 16:39:21 hemna: The sooner that starts getting communicated, the better. 16:39:26 #info working session proposal cinder-api microversions 16:39:27 smcginnis, +A 16:39:35 smcginnis: agree 16:39:42 are we going to add any new minimum feature requirement 16:39:43 scottda: hi 16:39:45 hi 16:39:55 I've already put up spec and code 16:39:59 xyang: Any you think should become minimum? 16:40:01 so we already have the fishbowl session which is experimental api and micro versions 16:40:10 patrickeast: ^ 16:40:12 https://review.openstack.org/#/c/223803/ #spec 16:40:22 smcginnis: migration, retype, manage 16:40:28 #info spec for microversions 16:40:29 https://review.openstack.org/#/c/224910/ #code 16:40:37 #link https://review.openstack.org/#/c/223803/ 16:40:41 I thought we might need a session to iron out details 16:40:43 xyang: What would we add to the minimum? 16:40:48 #info code for microversions 16:40:50 #link https://review.openstack.org/#/c/224910/ 16:41:05 jungleboyj_: let me just add a note to the etherpad 16:41:08 xyang: Seems reasonable. 16:41:13 scottda: this is true. the fish bowl will be more user facing... working session can be more of a technical deep dive 16:41:20 i kind of like the idea of having another more deep dive in how the microversions will be implemented 16:41:21 patrickeast: does that make sense? ^ 16:41:25 also, FWIW, we seem to have a bunch of things that are looking like dependencies on the microversioning stuff landing early in M 16:41:27 xyang: Ok. 16:41:28 patrickeast: yay! 16:41:28 thingee: yep :D 16:42:00 #agreed Cinder-api microversion working session deep dive seems good! 16:42:03 scottda: thanks 16:42:11 so the more folks looking at scottda's spec and code the better. 16:42:15 scottda: and thank for already being ready with spec/code 16:42:36 sure. As hemna said, it's a dependency for a few things 16:42:39 #info working session proposal: Remove volume manager locks 16:42:40 hemna: hi 16:43:00 so, I don't think we need a fishbowl for this one as it's more of a deep dive 16:43:11 hemna: I agree. 16:43:13 this one has some dependencies that need to happen prior to this. 16:43:19 the microversion stuff that scottda is working on 16:43:30 as well as gorka's compare/swap changes 16:43:31 hemna: users shouldn't really know or care about this... just that stuff works 16:43:37 thingee, yah. 16:44:02 I'm still working on the review for the spec, so any eyes on it at this point would be good. 16:44:16 I'd like to get started on the code soon, but want to iron it out first in spec if possible. 16:44:21 hemna: my question though is what are we going to do differently here? I think last summit we had this similar topic, but it didn't really go anywhere. 16:44:40 thingee, I think there are a few small differences 16:45:01 mostly the above mentioned dependencies 16:45:19 Maybe a sprint session then? 16:45:22 and the addition of the new 'reading' states that gorka has mentioned as well 16:45:36 those might already be in place now thought, unsure. 16:45:45 the other related topic 16:45:47 thingee: hemna IMO removal of locks shouldn't need a session at all, just some code 16:45:55 is how to deal with the microversioning stuffs. 16:45:59 no spec either really 16:46:31 what to do when the version doesn't match the new code. how to deal with the existing volume manager locks, etc. 16:46:35 * jgriffith is no completely confused 16:46:36 that part I'm unclear on still. 16:47:00 anyway, we can take it offline in #cinder channel 16:47:08 hemna: We can talk. I *think* I've got a picture about how it will work 16:47:28 s/is no/is now/ 16:47:43 ok, just going to leave a note for the next meeting organizer to check back 16:47:45 seems to be multiple conversations taking place, none having anything to do with current topic? 16:47:53 scottda, coolio. microversioning API only changes seems straight forward, but stuff that bleeds into the volume manager is a bit more involved. 16:48:23 #action scottda will talk to hemna on volume manager locks to see if we need a topic in working session 16:48:24 jgriffith, it's all related. 16:49:05 ok 11 mins for sprints 16:49:06 there's a lot 16:49:35 one I want to focus on first are the potential to other session types 16:49:51 #info sprint proposal: Future of Cinder and getting back to our roots 16:49:53 jgriffith: hi! 16:49:57 hola 16:50:16 That might be a good fishbowl as you mentioned (or somebody mentioned) 16:50:38 so my initial question is the difference here with your working session cinder should/could be the next vipr 16:50:43 Seems like it might have wide interest. Or impact. 16:50:46 jgriffith: That makes sense to me. Seems like we want that to be open to a wider audience. 16:51:00 I know one session is user wide, the other is cinder dev wide 16:51:09 thingee: ahh... yeah, so they're different in my mind 16:51:38 thingee: one is stressing Cinder/OpenStack is NOT VMware and we are crazy with vendor features 16:51:56 thingee: the other is, how to make Cinder something more consumable outside of OpenStack 16:52:22 thingee: and what features to all these "new" abstractions have that we're missing and do they really matter? Or is it just marketing 16:52:46 features is not a good choice of words maybe... 16:52:52 but more along design 16:53:20 are there things we need to look at in our design to offer more flexibility for plugins that does not impact core API 16:53:30 that sort of thing... does that clarify at all? 16:53:37 jgriffith: ok, so the future of cinder is more focused on reference implementation 16:53:49 but a catchy title, as you would want... for a fish bowl :) 16:54:01 thingee: and what do we do with these things like multi-attach, CG's, replication etc etc etc 16:54:10 got it 16:54:14 thingee: as in... "don't do them" 16:54:24 or "do" 16:54:31 well, users might be like "I want all the features" 16:54:38 thingee: indeed! 16:54:46 Do all the things! Now! 16:54:54 but ... some of us will be like "lets give them what they want because it makes my company happy" :) 16:54:55 this sounds like the extensions discussion 16:55:03 hemna: it's not 16:55:14 it's about Cinder and what it should be 16:55:21 and some of us will be like "this hurts cinder core, how do we ensure these features work without the reference..." 16:55:34 regardless of the other components like extensions, contribs experimental API's and all that craziness 16:55:49 jgriffith: I agree, it's different from extensions. 16:56:18 hemna: it's more about a robust stable service that's maintainable/supportable 16:56:20 so I think the discussion will play out like that. 16:56:34 jgriffith: So - should we be adding new features that aren't core across all of not? Is that the point? 16:56:35 and actually "works" without being a mess of spaghetti code and weird hacks all over the place 16:56:49 jgriffith, I'm all for that :) re: fixing nova <--> cinder interactions. 16:56:51 smcginnis: yes, that's part of the point absolutely 16:56:52 s/of not/or not/ 16:57:05 * thingee is standing because there is a spider somewhere on his desk 16:57:10 LOL 16:57:15 thingee: you don't like spiders? 16:57:21 Hah! 16:57:27 * jgriffith doesn't like them either... can't tell what kind of mood they're in 16:57:27 jgriffith: not quick black spiders 16:57:31 only 2 minutes left 16:57:39 My wife would be screaming bloody murder. 16:57:49 ha 16:57:51 it's the eyes...too many of them.. 16:58:03 * hemna grabs the shotgun..... 16:58:08 :) 16:58:09 diablo_rojo: yeah, I don't trust anything with more than 3 eyes 16:58:09 ok... so fishbowl wise, I'm not sure the user can contribute to this. 16:58:13 can we not talk about bugs here?:) 16:58:25 thingee: yeah, may be a good point 16:58:30 it might be good to get user/ops feedback though, because we're essentially saying we want to move away from providing all the features 16:58:31 jgriffith: so people with glasses are untrustworthy? 16:58:42 thingee: and honestly we as a dev team may not be able to even agree on anything here anyway 16:59:01 diablo_rojo: those aren't eyes... just lenses :) 16:59:04 jgriffith: I think it's going to be the next interesting move for the cinder PTL 16:59:08 thingee: ++ I think user/ops will care about this 16:59:10 *hint hint* 16:59:18 thingee: indeed!! 16:59:33 IMO it really could be a make or break for Cinder 16:59:37 you want fame and glamour cinder ptl, this decision is it 16:59:38 jgriffith: okay good :) 16:59:42 No pressure future PTL :) 16:59:55 When are the election results? 16:59:56 Hah, fame and glamour. 17:00:02 ok we're out of time.. 17:00:08 thanks all for letting me lead the team for two cycles 17:00:11 been a pleasure 17:00:16 and I'm not disappearing 17:00:18 so no rumors 17:00:20 thingee: thank you for leading!! 17:00:21 Thanks thingee! 17:00:27 jungleboyj_: 13:00 UTC Sept 24 17:00:29 thingee, yah nice job man. 17:00:30 thingee: thanks! 17:00:32 thank 17:00:32 thingee: thanks for leading! 17:00:32 thanks thingee 17:00:34 thanks Mike!!! 17:00:38 thanks! 17:00:40 don't let that spider take you out :) thanks for the leadership 17:00:43 #endmeeting