16:00:02 #startmeeting cinder 16:00:03 Meeting started Wed Sep 16 16:00:02 2015 UTC and is due to finish in 60 minutes. The chair is thingee. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:05 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:09 The meeting name has been set to 'cinder' 16:00:16 hello all! 16:00:16 hi 16:00:16 Hi! 16:00:16 hi 16:00:18 doink 16:00:19 hi 16:00:20 o/ 16:00:20 o/ 16:00:26 Hello! 16:00:26 hi 16:00:30 hi 16:00:30 hi 16:00:32 #topic announcements 16:00:44 hi 16:00:52 #info we have blockers for liberty release 16:00:55 hello :) 16:00:56 #link https://launchpad.net/cinder/+milestone/liberty-rc1 16:00:57 hi 16:01:06 Lets get these pushed through today 16:01:09 Hi 16:01:12 hi 16:01:20 o/ 16:01:20 if there's anything missing from that list, please ping me in #openstack-cinder to discuss 16:01:28 otherwise I'll give the OK 16:01:42 I'll work with e0ne on the cinder-all binary. I was helping him test that last night 16:01:54 hemna: thanks 16:02:01 hemna: I believe that patch has one +2 now 16:02:09 hey! 16:02:12 thingee: when are you planning to cut RC1 16:02:15 Happy Wednesday! 16:02:23 xyang: asap 16:02:23 thingee: Is this worth to get into RC1 - https://review.openstack.org/#/c/223602/ ? 16:02:40 thingee, ok. I had seen problems with it last night before I left. I'll follow up. 16:02:40 dulek: taking note, will discuss after meeting in #openstack-cinder 16:02:48 hemna: thanks 16:03:00 thingee: the fact that quota-get for a normal user is broken with the default keystone policy is a bug that should be on that list IMO 16:03:03 thingee: Got it. 16:03:30 DuncanT: ack, can you ping me on #openstack-cinder with the bug link, please? 16:03:38 thingee: I've verified dulek's patch with devstack 16:03:39 Aye 16:04:09 DuncanT: +1. do we have filed bug? 16:04:11 #info Cinder PTL nominations/elections happening 16:04:14 #link https://wiki.openstack.org/wiki/PTL_Elections_September_2015 16:04:28 just a reminder to folks that nominations are being accepted now 16:04:37 we have smcginnis stepping up for the position in mitaka! 16:04:44 So far Sean and myself have announced candidacy 16:04:44 DuncanT: https://launchpad.net/bugs/1491495 16:04:46 Launchpad bug 1491495 in Cinder "cinderclient.quotas.get() fails against Cinder service in a Devstack" [Undecided,Fix committed] - Assigned to Ivan Kolodyazhny (e0ne) 16:04:50 oh and DuncanT ! 16:05:20 o/ 16:05:25 I created a patch with nomination too 16:05:35 any other announcements? 16:05:37 DuncanT: Awesome! 16:05:58 e0ne: Cool! 16:06:10 I'm going to post mine here in a bit. 16:06:38 looks like we got a healthy election coming 16:06:44 alright lets get started 16:06:45 Wow! 16:06:59 Nice! I'm going to throw a dice to decide:) 16:07:04 agenda for today: https://wiki.openstack.org/wiki/CinderMeetings#Next_meeting 16:07:13 :) 16:07:26 xyang: :-) 16:07:28 going to switch the order 16:07:33 #topic Nova summit session for Cinder issues 16:07:35 scottda: hi 16:07:37 hi 16:07:50 #info Etherpad for session 16:07:52 #link https://etherpad.openstack.org/p/NovaCinderMitakaSession 16:07:52 I wanted to setup a session with Nova to discuss Cinder issues 16:08:11 I figured we could put all things Cinder <-> Nova in one session. 16:08:28 scottda, that's a lot :) 16:08:29 multi-attach, brick (rootwrap), api changes, bugs 16:08:34 well, it might be too much 16:08:35 40 minutes 16:08:50 And I haven't submitted the request for a Nova session yet. 16:08:52 scotttda: the rootwrap thing might be covered in a session with oslo 16:09:00 mriedem: roger that 16:09:05 i posted for nova to do the rootwrap thing already 16:09:14 mriedem, yah that's going to require oslo participation, so we should just do that there. 16:09:39 so probably focus on features/api changes in a nova session 16:09:51 so, if folks think differently, I could avoid doing a 0ne-session-for-all, but johnthetubaguy like that idea. It's a matter of topic scope 16:09:56 e0ne, if cinder-baremetal is going to use os-brick, then it's going to need rootwrap fun as well. 16:10:08 does multi-attach need it's own session? 16:10:14 keep in mind the amount of bikeshedding that will happen in a design session 16:10:20 hemna: you're right 16:10:24 mriedem, +1 16:10:29 so make sure this is focused 16:10:33 with homework assigned before hand 16:10:36 mriedem, scottda should we do the multi-attach session in nova ? 16:10:44 perhaps a separate session just for the bikeshedding? 16:10:48 :) 16:10:59 so the sessions with lots of topic do tend to fail 16:11:02 hemna: This is about a NOva session, not a cinder session. 16:11:03 scottda: ++ 16:11:41 johnthetubaguy, +1 16:11:41 now if there is a full story around improving integration that could be reviewed, maybe thats a good single session, at least thats where I was thinking 16:12:15 I was hoping once we have the list of things we want to talk about, it should become much clearer about what the best way forward is 16:12:15 ok, so maybe a single session about improving integration and any API changes being proposed.... 16:12:28 johnthetubaguy: fair enough. 16:12:45 I suspect a combination of already agreeing stuff in specs, and a session to cover some loose ends, might be a good approach 16:12:48 so perhaps cinder folk can put any additional ideas on that etherpad, and we can go from there? 16:13:06 scottda, url 16:13:07 yeah, provide any and all education/background up front 16:13:18 https://etherpad.openstack.org/p/NovaCinderMitakaSession 16:13:19 otherwise you're going to be explaining the cinder use cases to the nova people for half the session 16:13:22 I am good with that, a Nova-cinder topics, and we can work out what needs doing to make that work 16:13:25 mriedem: +1 16:13:26 scottda: Initially when this was discussed at the Cinder midcycle sprint, we wanted to have our API defined properly so others can consume it properly. Maybe appropriate for this session. 16:13:57 thingee: yes, including a spec for that. 16:14:38 OK, if you are interested, add to that etherpad and let me know. 16:14:52 johnthetubaguy: What's the time frame for requesting a NOva session? 16:15:01 yeah, discussing spec disagreements tends to be more productive, focused bikeshedding, a shed for only one bike (waffle waffle) 16:15:08 scottda: we don't have one right now 16:15:17 scottda: maybe point to your etherpad, and submit one now? 16:15:33 ok. I'll do that today. 16:15:43 #action scottda to propose nova session with etherpad 16:15:46 I think that's it for me 16:15:46 awesome, that will stop me forgetting about it at the last moment 16:16:13 #action scottda to follow up with spec on proper Cinder API usage 16:16:20 scottda: got a lot to do 16:16:35 yeah, I think hemna will also help with the api spec 16:16:41 * scottda volunteers hemna 16:16:57 :-) 16:16:58 scottda: thanks, anything else? 16:17:03 nope. Thanks 16:17:07 :) 16:17:21 #topic Design summit topics 16:17:23 #link https://etherpad.openstack.org/p/cinder-mitaka-summit-topics 16:17:26 scottda, I have it on my plate to rework the volume manager locks -> ING checks spec 16:17:48 Not much notice to get proposals added here, but we can take another look next week 16:17:50 hemna: OK, I guess this is separate. We'll talk.... 16:18:25 thingee: +1, it seems people need time to come up with the topics. 16:18:37 thingee: Last time it was harder to get this scheduled. 16:18:41 * thingee is noticing people adding things now :) 16:18:45 thingee: +1 16:19:23 patrickeast: hi 16:19:29 hey 16:19:38 Expermental/api features I think is a great topic 16:19:48 for the user input, what were you interested in 16:19:58 When I was looking in the morning all the sessions proposed for Friday would fit to the two other categories. I thought we will go for a beer at Friday then. ;) 16:20:12 mostly looking for feedback on if we did it, would operators want them 16:20:23 or maybe they prefer not to be exposed at all to that sort of thing 16:20:39 and if we do implement them, maybe they have suggestions/requirements for the how 16:20:44 patrickeast: we sould have a summit session, I have a proposal for a "different" model 16:20:54 patrickeast: using a contrib directory 16:20:59 jgriffith: The next ViPR !?! 16:21:02 jungleboyj: :) 16:21:09 o_o 16:21:16 :) 16:21:19 to both of those things 16:21:22 jgriffith: What has come over you man!?! 16:21:24 ;-) 16:21:42 contrib/vipr.py solved 16:21:42 manila just implemented experimental API. we can take a look of that 16:22:02 xyang: +1 16:22:12 * thingee takes note 16:22:52 patrickeast: I have took notes on things mentioned. would be good to research manila and jgriffith's different approach 16:23:09 thingee: +1 Sounds like experimental API is working into. 16:23:11 yea i'll take a look 16:23:21 I've already linked manila's patch in the etherpad 16:23:23 patrickeast: please post an etherpad on your findings and link to your topic 16:23:25 manila experimental is built on microversions 16:23:28 need micro versions too 16:23:29 thingee: will do 16:23:38 tbarron: +1 16:23:39 I'm working on microversions ATM 16:23:43 #action patrickeast to make etherpad on expermental API research 16:24:00 scottda: I saw you have a spec 16:24:04 scottda: great 16:24:11 not sure if we need micro versions in fishbowl 16:24:22 xyang: Yes, and I'm porting code from Manila this morning 16:24:26 I'd recommend working session 16:24:30 scottda: ^ 16:24:31 scottda: nice 16:24:34 ok 16:24:40 +1 16:24:49 scottda: then you will get experimental too 16:24:57 scottda: Rocks! 16:25:05 Well, I didn't port the experimental bits. 16:25:11 I thought that should wait 16:25:12 I'm not sure how we solve the dependency problem of patrickeast's fishbowl happening first then working session on micro versions 16:25:23 scottda, awesome 16:25:29 thingee: they should be one session 16:25:38 xyang: good idea 16:26:04 especially if you are using experimental from manila that needs microversions 16:26:54 #info Proposal Availability Zones in Cinder 16:27:03 dulek: DuncanT hi 16:27:08 hi 16:27:13 Hi 16:27:27 I definitely think this is good for fishbowl 16:27:32 for getting user feedback 16:27:36 operator rather 16:27:43 ++ 16:27:43 So basically we were talking about that multiple times on IRC, always the conclusion was - what do operators expect from Cinder. 16:27:44 thingee: +1 16:27:56 So this has come up on the mailing list recently, and of the two modes we defacto supported, one is currently broken (though the bug is a blocker). I've pinged a few operators 16:27:59 We don't know that, let's gather feedback. 16:28:05 thingee: can we do it after operators session? 16:28:05 thingee: ++ 16:28:20 I've three replies so far, all saying the same thing, plus the HP answers 16:28:31 I'll write them up soon 16:28:37 thingee: is the cinder team thinking about possibly getting a session at the ops track to get feedback or is this cinder design summit fishbowl session? 16:28:47 That should give us a starting point for discussion 16:28:58 e0ne: might be difficult to schedule this around ops sessions. Fishbowl will be taking place tuesday and ops happens tuesday-wednesday 16:29:02 If there is a list of questions that need feedback, I'll be glad to pass them to the Product WG too. 16:29:27 ShamailT: we've done it in the past and it was a flop ... although with discussions with the foundation, might've been a schedule issue 16:29:42 that's a bummer they are done on the same days. It would be nice if the ops stuff happened first, so we could bring some of the feedback into the fishbowl and sprint sessions. 16:29:49 thingee: IIRC we were the only ones there :( 16:29:51 if cinder does AZ's, don't repeat the issues that nova has had 16:29:55 jgriffith: yup 16:29:58 thingee: gotcha 16:30:18 we tried in Vancouver and almost no one showed :( 16:30:26 mriedem: ? 16:30:30 dulek, DuncanT would appreciate an etherpad with what Nova has done ... maybe connect with mriedem 16:30:35 hemna: That was a bummer. 16:30:37 dulek: https://review.openstack.org/#/c/223802/ 16:30:38 Yeah, Vancouver ops sync up was a dissapointing turn out 16:30:44 mriedem: Cinder already does AZs, with two different configs 'supported' 16:30:47 dulek: speak with bauzas in the nova channel 16:30:59 we've just had some really gross bugs with AZs 16:31:10 thingee: I've got the problem described in the spec linked in description, getting the info into the etherpad isn't a problem. 16:31:14 ultimately user error, but really not good usability 16:31:45 scottda: nova ops sessions at summit are also very sparse, it's mostly nova people talking to each other 16:31:59 :( 16:32:05 #action dulek to post an etherpad link under the topic proposal with previous nova AZ issues in mind 16:32:45 Okay, that's probably all. 16:33:08 dulek DuncanT thanks 16:33:21 that leaves us two more slots for fishbowl 16:33:30 #info two more slots available for fishbowl 16:33:39 working session 16:34:00 cinder<->driver interface 16:34:01 eharney: hi 16:34:10 hey 16:34:12 glad to see this up and seems appropriate for working session 16:34:34 thingee, +1 16:34:36 i don't have a lot of specifics on this yet, but based on various chatter over the last cycle, it seems like something that needs some work and definition 16:35:20 eharney: makes sense to just document the driver api now and have the review or merged doc be included with your session 16:35:25 don't think things need to wait 16:35:43 hemna: I'll submit a patch to fix the CG interfaces before the summit 16:35:47 if things need to be decided, they don't need to be documented right away 16:36:02 sounds good 16:36:18 eharney: do you have bandwidth for proposing that to the cinder dev docs? 16:36:18 xyang, :) 16:36:21 before the summit 16:36:26 hemna: :) 16:36:29 It would be good to talk about the worts on out interface 16:36:35 thingee: hard to say at this point 16:36:56 xyang, that would be awesome. we'll need to file bugs against drivers after that lands to get them fixed. I can help with documenting which drivers are doing it because of CG. 16:37:05 can anyone collaborate with eharney to get a doc patch going? 16:37:10 part of this is collecting up knowledge on some of the things like drivers that don't use interfaces exactly in the intended way, etc 16:37:11 hemna: sure 16:37:17 and i'm not sure how big the scope of that kind of thing is yet 16:38:01 I can help 16:38:06 #info cinder<->driver interface working session 16:38:22 rhedlind: thanks 16:39:00 #action rhedlind and eharney to collaborate on a driver api doc for cinder dev docs 16:40:12 #info proposal for c-vol A/a working session 16:40:17 dulek: hi 16:40:26 hello again 16:40:31 so cross project distributed lock topic should be happening before this, yay 16:40:39 Ah, definitely. 16:40:45 #info cross project distributed lock topic will happen before this 16:40:54 I think we all know what's this about. 16:41:04 And I don't think there's a clear action plan for Mitaka. 16:41:16 So I would expect that to be the aim of the session. 16:41:24 Based on discussions in the cinder midcycle sprint, I think work can just begin with for -ing state work, no need to discuss things more 16:41:39 thingee, +1 16:41:53 it would be great if those patches were ready and linked with the etherpad topic 16:42:13 thingee, do we need a spec for the -ing checking, or a bug ? 16:42:45 thingee, we have this old one that started talking about it...https://review.openstack.org/#/c/149894 16:43:03 Isn't -ing checking changing the behavior of the API and therefore invalidates API contract? 16:43:20 dulek, yes. I think we need microversioning in place prior 16:43:41 we'll be returning VolumeIsBusy in many places, and nova doesn't really expect that currently. 16:43:41 hemna: Okay! :) 16:43:42 hemna: yeah lets just renew that spec 16:43:49 * scottda types microversioning code faster.... 16:44:03 thingee, ok I'll post up a new version of that spec and add the deps for microversioning and we can churn on it. 16:44:16 So I would say - let's move that as a backup for the sprint and we'll discuss things there if needed. 16:44:19 scottda, is it passing jenkins yet? 16:44:37 I've not finished the port, but I should be done later today. 16:44:40 hemna, dulek I need some to spearhead renewing that spec and have our nova liason present this to the nova folks so there are no surprises 16:44:58 someone* 16:45:06 thingee, scottda and I will work on that asap. 16:45:11 Yes, when spec is ready I'll take it to Nova (meeting or IRC ) 16:45:29 thingee: I see hemna's volunteering, but I can also help to work on the spec. I think geguileo will be also interested. 16:45:49 #action hemna to renew spec on return VolumeIsBusy on ing states 16:46:03 #action scottda will present spec to Nova meeting 16:46:30 dulek: I'm sure hemna will be happy to delegate since he's a busy guy and all 16:46:47 :) 16:46:53 hemna: Just tell me what to do if you need. ;) 16:47:02 Okay, so I'm moving that to the sprint, right? 16:47:03 #info proposal for continuing abc work working session 16:47:04 jgriffith: hi 16:47:09 thingee: hola 16:47:34 I agree this is a good working session topic. 16:47:53 thingee: which one :) I added two :) 16:47:56 thingee ++ 16:48:03 thingee: OHH 16:48:05 never mind :) 16:48:06 my bad 16:48:10 jgriffith: one at a time :D 16:48:10 jgriffith: all of them:) 16:48:11 dulek, ok I'll post up a new patchset on that review and we can all go at it. 16:48:17 thingee: got ya... sorry 16:48:35 i'll continue to be involved in the abc work, seems like a good session 16:49:05 jgriffith: we should probably take a different approach from the cinder midcycle sprint to make headway 16:49:14 eharney: I'd like to run some ideas / POC's by you in the coming weeks if you're intrested 16:49:22 jgriffith: for sure 16:49:25 thingee: agreed 16:49:47 thingee: I'll have POC's, if nobody likes them we can swap something else into the slot :) 16:50:07 thingee: and I will not raise it again afte r this :) 16:50:17 #action jgriffith to propose more POC's 16:50:35 jgriffith: you have already done a poc :P ... but always welcomed 16:50:59 #action eharney and jgriffith will have discussion on pocs in the coming weeks 16:51:15 eharney jgriffith We'll still check back next week on progress if that's fine ^ 16:51:30 sure 16:51:41 note that there are still some pending patches cleaning up the current state of abc things out there, too 16:51:51 ok enough time for one more working session 16:52:16 #info Proposal for why fight it, cinder could should be the next vipr 16:52:21 jgriffith: hi again 16:52:32 thingee: :) 16:52:40 maybe appropriate for fishbowl? 16:52:42 lol 16:52:50 :) 16:52:51 jgriffith, are you feeling ok today? :P 16:52:56 hemna: ++ 16:52:59 thingee: so honestly I don't know/care necessarily where they land 16:53:00 :) 16:53:19 * thingee doesn't care either... it's the next ptl's problem j/k 16:53:25 lol 16:53:36 hehe 16:53:37 i'm not sure i really see the whole picture for this one, but part of my driver interface proposal may be asking some related questions from a different direction 16:53:51 eharney, yah seems like it might be related 16:53:52 hemna: jungleboyj :) So maybe a better title, but the point being there's no reason we souldn't be offering the same sort of functionality to make those things osbolete 16:53:53 if we're not already 16:54:05 jgriffith, +1 16:54:08 eharney: could be for sure 16:54:09 And the PTL nominations just went from five down to one 16:54:19 eharney: I'm fine with covering it ad-hoc in other discussions 16:54:22 jgriffith: I think it is actually an important discussion. 16:54:23 kmartin, lol 16:54:36 kmartin: huh? 16:54:52 jgriffith, 16:54:57 jgriffith: I think he was replying to my remark of me making the next ptl's job harder 16:54:58 hemna: oh... :) 16:55:07 Ahhhh!!! 16:55:09 got it! 16:55:10 sorry 16:55:15 :) 16:55:30 kmartin: You and your sarcasm. ;-) 16:55:40 5 minute warning, BTW. 16:55:43 jungleboyj: I agree... I don't know if it needs to be its own session or not, but I'll let others decide based on available time 16:55:53 but I certainly plan on discussing it 16:56:02 even if I hijack somebody elses talk :) 16:56:06 just kidding 16:56:12 jgriffith: I think that is great. It has a lot to do with keeping Cinder relevant. 16:56:13 jgriffith: can you work with eharney and decide next week if it can be merged in with the driver interface session 16:56:25 thingee: certainly 16:56:42 thingee: I can also fold it into one of my other proposals.. like "future of cinder" 16:56:49 it fits there as well and has some overlap IMO 16:57:13 The future is bright man! 16:57:27 #action jgriffith and eharney to discuss if the cinder should be the next vipr session should be folded in the cinder<->driver api session 16:57:46 jgriffith: If it's about what I think we've discussed in the past, that would be a great discussion. 16:57:58 jgriffith: might need more planning of making sure key people are there 16:58:00 jgriffith: did you see DuncanT's coment in the etherpad? 16:58:01 thingee: I think it is, yes 16:58:06 jgriffith: excellent 16:58:10 #topic open discussion 16:58:13 2 mins left 16:58:19 we'll return back to summit topics next week 16:58:24 this was good though 16:58:53 Good stuff. 16:59:12 So, how much longer do we have to approve and merge bug fixes? 16:59:26 can someone add to the summit topics etherpad how long these sessions are? 16:59:41 jungleboyj: at some point this week I'll make a cut. Ideally I'll give the word for tomorrow 17:00:02 I also removed some stuff from rc-1... 17:00:05 #endmeeting