16:00:01 #startmeeting Cinder 16:00:01 Meeting started Wed Oct 12 16:00:01 2016 UTC and is due to finish in 60 minutes. The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:02 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:05 The meeting name has been set to 'cinder' 16:00:06 Ping- 16:00:07 dulek duncant eharney geguileo winston-d e0ne jungleboyj jgriffith thingee smcginnis hemna xyang1 tbarron scottda erlon rhedlind jbernard _alastor_ bluex patrickeast dongwenjuan JaniceLee cFouts Thelo vivekd adrianofr mtanino yuriy_n17 karlamrhein diablo_rojo jay.xu jgregor baumann rajinir wilson-l reduxio wanghao thrawn01 chris_morrell stevemar watanabe.isao,tommylike.hu mdovgal 16:00:10 o/ 16:00:15 hi 16:00:16 o/ 16:00:18 hi 16:00:22 Hi! 16:00:23 hi 16:00:31 Hello! 16:00:33 Hi 16:00:46 hello 16:00:49 Hi 16:00:57 hi 16:01:00 Hi! 16:01:06 <_alastor1> o/ 16:01:22 #topic Announcements 16:01:41 Hello! 16:01:43 .o/ 16:01:48 I've added all topics from the summit planning etherpad to the schedule. 16:01:59 All sessions are also added with etherpads here: 16:02:07 #link https://wiki.openstack.org/wiki/Design_Summit/Ocata/Etherpads#Cinder Design Summit sessions 16:02:12 hi 16:02:34 If you are leading any of those discussions, please start adding info there to prepare. 16:02:48 ok, will do 16:03:02 Any extra topics can also be added to our contributor meetup etherpad: 16:03:11 #link https://etherpad.openstack.org/p/ocata-cinder-summit-meetup Contributor Meetup Etherpad 16:03:33 Just the afternoon on Friday, but we will handle that like the midcycle and just work our way through topics. 16:04:07 #topic Documentation needs 16:04:13 So... documentation. 16:04:25 Apparently I misunderstood what was being proposed in the past, or things have changed. 16:04:56 But just so everyone knows, for any patch that has DocImpact in the commit a bug is automatically created against Cinder when that patch merges. 16:05:32 It is up to us (ideally the person submitting the patch) to take that and follow through with adding anything needed to the config guides or other documentation affected. 16:05:55 But looking through, I think we are misusing that tag in some cases. 16:06:05 The general understanding was any config option change required it. 16:06:10 But that is not always the case. 16:06:26 Only if additional information needs to be added to go with the config option. 16:06:44 Otherwise the config options are automatically generated in tables when the docs are built. 16:07:06 So if having it listed in the table is sufficient, no need to put a DocImpact flag in there. 16:07:29 Good to know 16:07:42 Any questions/concerns? Or did everyone go to get coffee? 16:08:02 smcginnis: I answered your question in the ML about generating help based on microversion. 16:08:06 smcginnis: Did you see that? 16:08:16 scottda: Oh, great. No, not yet. 16:08:23 smcginnis: And how that affects generation of the docs 16:08:37 Is it worth requiring the doc patch to be at least up in draft before we merge, for complex features? 16:08:37 scottda: I'll take a look after the meeting. Thanks! 16:09:21 DuncanT: I'd hesitate to make that a hard requirement, but especially for complex features that might be a good pattern to follow. 16:09:50 We also have the api-ref in-tree now. 16:09:55 Not as a hard requirement, they usually end up painful, but as a suggestion where it feels necessary 16:10:09 We really should start to enforce including api doc updates along with the api changes themselves. 16:10:12 DuncanT: ++ 16:10:19 DuncanT: +1 16:10:25 smcginnis: +1 16:10:43 smcginnis: +1 16:10:55 any reviewers for the groups api doc? https://review.openstack.org/#/c/370526/ 16:10:57 There's also some talk of moving all docs in tree. I'm a little concerned about that, but something to consider. 16:10:59 smcginnis +1 16:11:20 Then all patches that have a documentation impact can include those updates along with the change. 16:11:55 I really don't like the idea of cramming more non-functional stuff into the repo, but there definitely are some benefits to that approach. 16:12:04 Not something I'm going to advocate for any time soon though. 16:13:16 I'll move on, but ping me in channel if you have any docimpact questions. I can at least try to answer. 16:13:29 #topic Deprecating various extensions, specifically qos-specs 16:13:32 thingee: Hey 16:13:38 o/ 16:14:03 hey so I talked about deprecating qos-specs in favor of just using volume type specs 16:14:03 xyang2 mentioned something about three releases for deprecating tables? 16:14:22 do we have winston here? 16:14:52 thingee: before talking about the deprecation I'd like to get winston's input as he implemented it 16:14:52 thingee: I wasn't around for that discussion, what is the justification? There's already a *lot* of stuff in volume type specs 16:15:18 DuncanT: because that's where it belongs and what it does anyway :) 16:15:19 I found these two original patches by winston: https://review.openstack.org/#/c/29737/, https://review.openstack.org/#/c/42521/ 16:15:22 but it would still be fine for me to just remove qos_specs_manage from contrib and give a warning if it's loaded in cinder.api.extensions? 16:15:50 DuncanT: so in all seriousness, the justification was that there tends to be some confusion on "which" method should be used 16:16:05 thingee: I would think that would be OK. 16:16:13 jgriffith: Is throwing everything into one pot really helping make things even slightly usable / understandable? 16:16:21 you can't just remove it 16:16:22 and I can leave the table around with some migration script to volume type specs. 16:16:30 there are drivers that have implemented it 16:16:38 DuncanT: To turn it around, why does that need its own api? 16:16:44 some drivers will be broken if you just remove them 16:16:47 DuncanT: we can chat on the side if it distracts the conversation but yes IMO optional things like that are better in one place rather than 2 16:17:02 xyang2: I think after 1 cycle being marked deprecated. 16:17:27 I'd like to get winston's comments on this 16:17:44 thingee: Yup, we should do it in a way guaranteeing backward compatibility, but it's possible to move that into another table quite quickly (only c-api can modify QoS specs, right?). 16:17:46 xyang2: Yeah, that would be good to get his input. 16:17:55 xyang2: yeah, that would be good. I do remember that part of the reasoning was to be able to modify the qos without retyping the volume 16:17:56 I'd really like to defer this decision until after spain, I'm not sure IRC has enough bandwidth to actually get a decent discussion 16:18:11 xyang2: but that doesn't really work that way either 16:18:19 DuncanT: +1 That makes sense. 16:18:19 ok so I'll throw the warning in the file itself to stop using it and instruction to use the migration script and use volume type specs. 16:18:19 ok with that aside, I wanted to talk about the other stuff in contrib 16:18:19 is there anything in there that really needs to be an extension? 16:18:24 whoa serious lag 16:18:28 * thingee catches up 16:18:31 jgriffith: It was also so the you didn't have to specify the whole of the QoS details for every type 16:18:40 jgriffith: there's a front end and backend 16:18:54 dulek: I doubt there are that many to move over, really. lets be realistic 16:18:56 xyang2: yes, I'm aware 16:19:00 jgriffith: in nova, there is qos specs which is front end 16:19:09 thingee: We'd discussed moving things from contrib -> core if they are supported by all drivers. 16:19:22 thingee: But maybe not things like consistency_groups 16:19:37 scottda: ok good to know, so there are quite a bit of things that I can move over. 16:19:38 Or replication? 16:19:49 thingee: Sure, I'll be happy to help work out the DB migration strategy if we want that. 16:19:53 scottda: thingee I'd like to talk about changing that whole structure we have altogether if we could 16:19:54 scottda: no need to move consistency groups. we will migrate them to groups any way. actually part of CG is already in v3 16:20:12 I think we're doing it wrong and should fix it sooner rather than later 16:20:55 smcginnis: ok I think I have the information I need, thanks 16:21:07 LOL 16:21:08 ok 16:21:11 smcginnis: I mean if we make a decision to get rid of it, we need to make sure all drivers change the implementation to use volume types instead. we need to give driver maintainers development time 16:21:26 xyang2: Oh, right. Definitely. 16:21:31 jgriffith: you mean this? https://bugs.launchpad.net/cinder/+bug/1627921 16:21:31 Launchpad bug 1627921 in Cinder "clean up API versions" [Undecided,Confirmed] 16:21:32 xyang2: I think we've tabled that and moved on to the contrib discussion 16:21:38 thingee: Thanks! 16:21:40 xyang2: I don't think there are too many. happy to do it myself 16:21:51 scottda: yes 16:22:04 thingee: it needs code change in the driver, not just removing it 16:22:14 scottda: so as someobody that's been writing things to consume our API's lately I think we have some room for improvement 16:22:17 This is probably one of those things we should at least have a quick discussion on Friday afternoon for. 16:22:22 thingee: the keys need to be changed 16:22:31 xyang2: not sure you're reading what I said. I said I would change the drivers myself 16:22:31 it's not that complicated and not too many instances as you're making it out to be 16:22:40 jgriffith: That sounds good to me at first pass. 16:22:43 thingee: once they are moved to volume types, I think they need to be scoped, etc 16:23:02 thingee: I think that is a change that need to be validated by the storage backend 16:23:07 scottda: ok, I guess at this point you and I should just talk about it in Barcelona maybe? 16:23:26 scottda: not like I'm going to work on it between now and then 16:23:48 jgriffith: I reckon so. It's probably going to take until P release anyway. 16:23:56 scottda: no way! 16:24:01 #topic Open discussion 16:24:02 O-3 targetted 16:24:29 We have a cross project session related to proprietary libs and things. 16:24:33 thingee: also documentation has instructions on using qos specs 16:24:40 thingee: driver doc 16:24:44 I've updated the etherpad we started with what I could find here: 16:24:50 https://etherpad.openstack.org/p/cinder-brick-driver-externals 16:25:06 External libs are left highlighted. 16:25:18 Please take a look and see if I missed anything that you are aware of. 16:25:45 smcginnis, I have talking to other CP Liaisons and we should have decent representation from other projects there 16:25:51 #link https://etherpad.openstack.org/p/cinder-brick-driver-externals Proprietary lib info 16:26:11 diablo_rojo: Great! I think it will be a good discussion if we have a good mix there. 16:26:43 smcginnis yeah I have ironic, Manila, and Nova for sure. 16:27:04 thingee: for example, this doc explains in detail how to use qos specs for the VMAX driver: http://docs.openstack.org/newton/config-reference/block-storage/drivers/emc-vmax-driver.html 16:27:43 Oh, any objection to skipping next weeks meeting? I'll be in Barcelona already, and I know travel will be starting for some others as well. 16:28:16 smcginnis: That is good for me. 16:28:24 smcginnis: +1 16:28:24 diablo_rojo: that session has the latest Tuesday afternoon spot 16:28:28 I can wait until Monday or Tuesday and see if any important topics come too, if that helps. 16:28:35 jungleboyj, bardia: Cool, thanks. 16:28:55 xyang2: it sure does 16:28:58 diablo_rojo: overlapping with booth crawl 16:28:59 I'll wait to send out something to the ML until early next week just in case anything important does come up. 16:29:10 xyang2, Oh noes. 16:29:18 diablo_rojo: dang it! 16:29:24 I think booth crawl is 5-7 16:29:28 Nooooo!!!! 16:29:32 No official party this year either I hear. 16:29:36 You are correct. 16:30:00 So no evening events? 16:30:10 jungleboyj: We'll do a Cinder one I guess. ;) 16:30:15 smcginnis: saving money? 16:30:16 I'm sure there's going to be smaller ones. 16:30:18 dulek: ++ 16:30:22 dulek: +A 16:30:28 Time to go find the next big thing that people are throwing money at.... 16:30:36 xyang2: I guess so. 16:30:37 DuncanT: :-( 16:30:40 DuncanT: ;) 16:30:43 DuncanT: :D 16:30:46 DuncanT :) 16:31:03 Well kids ... it's been fun. ;-) 16:31:05 Tentatively, Cinder meetup Tuesday evening? 16:31:23 DuncanT: For parties maybe you should be a wedding planner? 16:31:52 scottda: And dance instructor. :) 16:31:53 smcginnis: And Wednesday night and Thursday night and ... 16:31:59 jungleboyj: Sold 16:32:35 Don't we know someone in Spain that might help with social planning? 16:32:48 scottda: ;) 16:32:58 OK, anything else meeting worthy? 16:33:08 Or can I go get some lunch? :) 16:33:30 OK, thanks everyone! 16:33:42 #endmeeting