15:59:48 #startmeeting Cinder 15:59:49 Meeting started Wed Jan 20 15:59:48 2016 UTC and is due to finish in 60 minutes. The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:59:50 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:59:52 The meeting name has been set to 'cinder' 15:59:53 o/ 15:59:53 o/ 15:59:59 Hello :) 15:59:59 Hey everyone. 16:00:08 Hi! 16:00:09 hi 16:00:13 hi 16:00:15 hi 16:00:21 hi 16:00:24 hi 16:00:27 hi o/ 16:00:37 hey 16:00:55 #topic Announcements 16:01:03 hi 16:01:07 Some general stuff first. 16:01:15 N and O releases now have names! 16:01:44 hi 16:01:49 Newton and Ocata 16:01:57 * dulek_ thinks Nameless and Om were best... 16:01:57 Wow, really? 16:02:04 dulek_: ;) 16:02:10 dulek_: +1 16:02:15 dulek_ +1 16:02:20 dulek_, +1 16:02:33 dulek_: :) 16:02:42 M-2 is here. Just waiting for some things to make it through gate before officially cutting the release. 16:02:53 At the current rate that should be some time in February. 16:02:54 :) 16:03:05 XD 16:03:18 Gate is 20 hours queue now.. 16:03:22 Hopefully later today or tomorrow I can submit the patch for the release. :) 16:03:38 Midcycle is next week. 16:03:46 Looks like a big group! 16:03:50 will we have meeting next week? 16:04:02 #link https://etherpad.openstack.org/p/mitaka-cinder-midcycle 16:04:09 e0ne: Good point. No 16:04:15 #info No meeting next week. 16:04:39 We'll probably be wrapped up in discussion there, so no IRC meeting next week. 16:04:46 I'll send out a notice to ML. 16:04:59 * mriedem sneaks in late 16:05:07 hi 16:05:28 Our spec list has morphed into a review focus list: 16:05:29 smcginnis: Nova meetup is next week, too, right? 16:05:32 #link https://etherpad.openstack.org/p/mitaka-cinder-spec-review-tracking 16:05:38 flip214: Yes, unfortunately. 16:05:51 We will have a hangout or something between meetups Wednesday morning. 16:06:03 We have a few topics to go over in our etherpad. 16:06:04 that probably means that there's no Nova IRC meeting too, grmpf. 16:06:07 thanks 16:06:12 I know nova has a few things as well. 16:06:30 Oh sure, now mriedem is hiding. :) 16:06:56 If anyone has a link to the nova midcycle etherpad, feel free to post it here too. 16:07:31 * thingee has some announcements 16:07:38 Just a note to warn folks, we will need to start cracking down on third party CI stuff very very soon. 16:07:41 https://etherpad.openstack.org/p/mitaka-nova-midcycle 16:07:43 or I could little diablo_rojo do it for cross-project specs 16:07:45 scottda: Thanks! 16:07:54 little = let 16:07:58 sorry need more coffee 16:08:00 thingee: I know she had a couple things. You too can arm wrestle for it. 16:08:03 thingee: :) 16:08:06 thingee: Yeah I have two to point out to people :) 16:08:09 *two 16:08:10 * thingee kicks back 16:08:30 thingee, diablo_rojo: Want to cover that now? 16:08:31 The first is on backwards compatibility for libraries and clients and the second is about having a common policy scenario more complex than just admin/non-admin. 16:08:50 #linlk https://review.openstack.org/#/c/226157/ 16:09:00 #link https://review.openstack.org/#/c/245629/ Common Policy Scenario more complex than admin or not admin 16:09:11 #link https://etherpad.openstack.org/p/mitaka-cinder-spec-review-tracking 16:09:16 Doh 16:09:17 #undo 16:09:18 Removing item from minutes: 16:09:37 Darn copy paste!!! 16:09:39 #link https://review.openstack.org/#/c/226157/ 16:09:43 diablo_rojo: Thank you! 16:09:44 There :) 16:09:58 thingee, diablo_rojo: Any other important CP stuff? 16:10:05 yes 16:10:08 hang on 16:10:15 The other two mentioned last week were approved so thank you to smcginnis e0ne and hemna_ for looking at those 16:10:32 Thats all :) Unless I am missing anything thingee? 16:10:53 Yes, so it has been approved by the TC for deprecating individual CLI's 16:11:04 #link https://specs.openstack.org/openstack/openstack-specs/specs/deprecate-cli.html Deprecate individual clis in favor of OSC 16:11:20 I believe we also have this on the midcycle topic list. 16:11:23 btw I learned that is a thing with the link command^ 16:11:33 thingee: we have to discuss cinderclient CLI at midcycle 16:11:49 I think this is something we will still need to discuss within the group, though it is approved. 16:11:51 just putting it in people's mind that the TC has come to a decision :) 16:12:00 thingee: Fancy link expert. 16:12:06 that's all 16:12:10 thingee: Thanks! 16:12:31 OK, just a couple more announcement/bookkeeping things, then we can move on. 16:12:44 For awareness: 16:12:49 #info Bug counts: Cinder - 487 open bugs, Python-cinderclient - 39 open bugs, OS-Brick - 14 open bugs 16:12:50 I will also miss you all at the midcycle. 16:12:52 :( 16:13:05 going to be at an offsite with the foundation that conflicts 16:13:09 thingee: Yeah, that will be weird. 16:13:23 Just so everyone has the right expectations - I will not be making coffee for you all. 16:13:25 thingee: Don't abandon us to go to the Carribbean :P 16:13:39 heh ;) 16:13:56 no thingee special brew this time :( 16:13:57 smcginnis: What!?! 16:14:03 * jungleboyj is out! 16:14:06 ;-) 16:14:07 OK, one more thing re: bugs. Pulling in some nice purdu graphs here if anyone cares: 16:14:13 #link http://cinderstats-dellstorage.rhcloud.com/ Bugstats 16:14:18 jungleboyj: ;) 16:14:29 #purdy 16:14:33 Can't type today. 16:14:50 And finally... 16:15:01 Nova can always use our help with volume related bugs 16:15:08 #link https://bugs.launchpad.net/nova/+bugs?field.status:list=NEW&field.tag=volumes Nova volume bugs 16:15:11 Oooh purdy. 16:15:33 So if you have some cycles, feel free to browse those for anything you can provide input on. 16:15:42 Alright, I'm done. 16:15:54 #topic Spes: Add-ServiceGroup-using-Tooz 16:16:17 dongwenjuan, janice : That you two? 16:16:29 * smcginnis copy pasted spes, that wasn't my typo this time! 16:16:30 yeah 16:16:35 ye 16:16:37 yes 16:16:56 #link https://review.openstack.org/#/c/258968/ Spec link 16:18:20 dongwenjuan, janice: Was there something you wanted to discuss about this? 16:18:34 in this spec, we proposal a feature of CinderServiceGroup 16:19:40 but, according to Huang's Comments, we don't make sure if this feature has availible 16:19:55 I'm agree with DuncanT: use cases are not clear from the spec 16:20:56 could you give us some suggestions for use cases? 16:21:15 late hi 16:21:25 e0ne, janice: +1, use cases would be fine. 16:21:47 * smcginnis notes tbarron's name on his list 16:22:07 oh 16:22:07 janice: I think that is what you need to provide us use cases so we know why we would want this in Cinder. 16:22:24 * tbarron hides head in shame 16:22:27 jungleboyj: +1 16:22:29 jungleboyj: +1 16:22:35 In Nova ZooKeeper was used to know of services that gone down faster 16:22:42 tbarron: You can get off his list buy buing him a shot of fireball 16:22:54 janice: So maybe this would be good to follow up in the cinder channel if there's not a specific issue to go over at the moment. 16:23:04 Only so much time in the meeting. ;) 16:23:05 If I recall correctly dongwenjuan's use case was connected with use of some external monitoring tool? 16:23:24 the benfits of using servicegroup is to get the service state immediately 16:23:28 diablo_rojo: :-) 16:23:32 OK 16:23:53 dongwenjuan: what circumstances does that help? ' 16:24:12 dongwenjuan, janice: OK if we move this to our channel? Or we can discuss more after we get through the other agenda items. 16:24:28 smcginnis: +1 16:24:29 DuncanT: It looks to me like they are trying to reduce the failed creates when a service is temporarily down? Not clear though. 16:24:45 ok, let mova the cinder channel. 16:24:52 OK 16:24:58 janice, dongwenjuan: Thanks 16:25:00 #topic Review requests 16:25:06 abhishekk: Hey 16:25:12 hey 16:25:15 #link https://review.openstack.org/#/c/257170/ 16:25:25 yes this is against, https://blueprints.launchpad.net/python-cinderclient/+spec/return-request-id-to-caller 16:25:28 abhishekk: Anything specific here, or just looking for visibility? 16:25:37 I need reviews 16:25:54 abhishekk: If you can answer my question in the review, I'm happy to +2 16:26:05 I have submitted 5 patches to implement these ans reviews are welcom 16:26:19 abhishekk: OK, unless there's some point to discuss about it, I think we can move on. 16:26:19 DuncanT: sure, I will do that 16:26:40 sure, smcginnis thank you 16:26:42 abhishekk: Captured in meeting notes, so hopefully that helps. 16:26:51 DuncanT: thank you 16:26:53 DuncanT: You had one too? 16:26:58 #link https://review.openstack.org/#/c/251843/ 16:27:11 yeah 16:27:21 Oh, Default=None 16:27:30 That probably is worth discussin. 16:27:34 I don't like this change, personally, but we should go one way or the other 16:27:35 ggggg!!! 16:27:52 :) 16:27:57 Probably with a hacking check too 16:28:09 Personally, I don't mind it being explicit. 16:28:19 * jungleboyj perks up 16:28:26 It's loaded at module load time, so there's not really a bug performance impact. 16:28:48 But I agree, if we want to enforce that, we should have a hacking check too. 16:28:52 I prefer explicit too, that means reverting to other change... 16:28:56 smcginnis: i'd prefer being explicit 16:28:57 smcginnis: +1 16:29:05 to explicit 16:29:18 DuncanT: Which other change? 16:29:50 I wouldn't think a hacking check for that would be too bad. 16:30:02 https://review.openstack.org/#/c/239704/ 16:30:08 DuncanT: Thanks! 16:30:09 https://review.openstack.org/#/c/239704/ 16:30:33 Ooops. :] 16:30:55 I'd vote for reverting that 16:31:07 DuncanT: +1 16:31:07 indeed, let's pull out the ballots 16:31:14 DuncanT: +1 16:31:23 Wait, I am confused. 16:31:34 So we want default=None or we don't want it? 16:31:37 we want it 16:31:41 explicit is better than implicit 16:31:44 i don't like that change... the code was probably written expecting None, what happens if the config module decides to change the default? 16:32:07 patrickeast: ++ Good point. 16:32:08 seems like there isn't anything gained by removing it except lowering the file size 16:32:12 OK... 16:32:14 patrickeast: They can't just change it like that, backward compatibility 16:32:16 i think it's a pretty solid contract from oslo.config that the default isn't going to just change 16:32:16 jungleboyj: We want it 16:32:20 I am all for being explicit. 16:32:29 DuncanT Good. 16:32:31 :-) 16:32:39 so if you remove default=None, does it set it to None as a default ? 16:32:43 #startvote Should we enforce explicitly setting Default=None in config options? Yes, No, Meh 16:32:44 Begin voting on: Should we enforce explicitly setting Default=None in config options? Valid vote options are Yes, No, Meh. 16:32:45 Vote using '#vote OPTION'. Only your last vote counts. 16:32:57 hemna: Yes 16:33:01 hemna: Yes, I believe so. 16:33:05 #vote no 16:33:09 yes 16:33:09 ok, well I like having it there, but that's just me 16:33:09 oslo.config defaults to None 16:33:11 #vote yes 16:33:13 #vote no 16:33:16 #vote yes 16:33:16 #vote Meh 16:33:18 #vote yes 16:33:21 oslo.config isn't going to change that default 16:33:21 #vote no 16:33:23 w/o breaking the world 16:33:29 #vote meh 16:33:30 #vote yes 16:33:31 mriedem: No, we shouldn't enforce because that's the default anyway? 16:33:32 #vote meh 16:33:32 wait... enforce? 16:33:34 #vote yes 16:33:35 #vote Meh 16:33:37 #vote meh 16:33:41 smcginnis: default=None is redundant 16:33:41 what does it mean to enforce ? 16:33:46 I don't understand the question actually 16:33:47 #vote no 16:33:58 so like... was meh the option for leave them alone and don't churn code removing/adding them? 16:33:59 we just prefer to not mess with the code that is explicit 16:34:02 requiring this as a hacking change is just saying that we aren't leveraging what oslo.config is providing as a library, which seems... odd 16:34:11 guitarza1: +100 16:34:14 I honestly don't see the point in that patch to remove the default=None 16:34:20 hemna: I'm fine if it's there or not, but given that the default is None anyway it is redundant. 16:34:24 other than someone wanting a code commit stat. 16:34:24 But more explicit. 16:34:30 hemna: +1 16:34:31 it's pointless really. 16:34:32 hemna: Exactly. 16:34:39 hemna: agree 16:34:46 But I don't think we need to revert or do any extra enforcement personally. 16:34:48 * DuncanT wants consistency at a minimum 16:34:50 ok -2 on that patch 16:34:51 done 16:34:52 next 16:34:58 If someone wants to be explicit, fine. If they don't, oh well. 16:35:06 DuncanT: so a proposed hacking check for "you must specify a default"? 16:35:14 smcginnis: that's my opinion too 16:35:15 DuncanT: That is the only think leaning me slightly toward doing anything. 16:35:20 DuncanT: ++ 16:35:24 *thing!!! 16:35:27 Dang it. 16:35:37 #vote Meh 16:35:39 you know sqlalchemy defaults nullable too right? 16:35:42 OK, last call. 16:35:48 for the sake of consistency, shall we vote for wether to revert Nate's change? 16:35:57 looks like someone wants to run up their -2 stats :) 16:36:00 there's probably lots on your sqlalchemy model code that is being explicit and redundant 16:36:01 :) 16:36:08 #endvote 16:36:08 Voted on "Should we enforce explicitly setting Default=None in config options?" Results are 16:36:09 Yes (5): DuncanT, winston-d_, jungleboyj, jgregor, diablo_rojo 16:36:10 Meh (6): hemna, smcginnis, kmartin, tbarron, geguileo, dulek 16:36:11 mriedem: True 16:36:12 No (4): mriedem, abhishekk, patrickeast, eharney 16:36:18 kmartin, I don't get to use it often, but this is as good as spiking the football. 16:36:22 so this hacking check would also enforce things like setting [] for ListOpt? 16:36:23 Apathy for the win! 16:36:28 smcginnis, :) 16:36:33 lol 16:36:35 XD 16:36:54 eharney: If that's a good thing to have, sure 16:37:03 DuncanT: I don't think it's a good thing to have 16:37:22 just wondering for the sake of completeness 16:37:36 eharney: I thing being consistent with the default is important. I've never noticed the [] thing 16:38:18 ListOpt defaults to None 16:38:21 I'd prefer is be explict, I find it more readable, but certainly I want consistency 16:39:05 everything defaults to None 16:39:44 mriedem: but you need to know that reading the code... explicit stating means one less thing to remember 16:39:59 ... 16:40:02 DuncanT: +1 16:40:17 the whole point of libraries like oslo.config is that it does the right thing and you don't have to explicitly do everything yourself 16:40:24 otherwise why use libraries for these things anyway 16:40:24 eharney: right 16:40:28 eharney, +1 16:40:43 eharney: +1 16:40:53 eharney: 'the right thing' is rather subjective. What should the default of an int option be? 0? None? What abotu a boolean? 16:41:13 eharney: that is absolutely the problem with libraries 16:41:21 people assume they do the right thing 16:41:28 Are we bike shedding too much? 16:41:31 never 16:41:33 smcginnis: absolutely 16:42:10 smcginnis: Probably. 16:42:14 Discuss this at the bike-shedding meeting during the mid-cycle? 16:42:19 smcginnis: I'll put up the patch, people can fight there... 16:42:30 there's a specified bike shedding meeting? :) 16:42:31 DuncanT: That sounds good. Thanks! 16:42:34 DuncanT: you have my +2 :) 16:42:49 guitarza1: Almost always ends up on the meetup agenda. ;) 16:42:53 winston-d_: :-) 16:43:05 Let's move on then... 16:43:08 #topic api-microversions is ready for review and testing 16:43:11 scottda: Hey 16:43:15 hi 16:43:18 I'd rather spend more time on this one. ;) 16:43:32 I think it would be good for folks to have a look before the mid-cycle... 16:43:41 scottda: +1 16:43:50 It can be complex when you first try to understand the whole microversion thing 16:43:57 scottda: do you have a plan to have it tested via CI? 16:43:57 #link https://review.openstack.org/#/c/223803/ Spec 16:44:08 And I can go over issues and questions at the mid-cycle 16:44:09 #link https://review.openstack.org/#/c/224910 Implementation 16:44:19 #link https://review.openstack.org/#/c/248163/ Client 16:44:26 e0ne: not with CI, but Nova is testing (or will soon) in Tempest 16:44:39 scottda: I agree. If we can go in to the midcycle with at least some foundation on this, that would help. 16:44:42 scottda: tempest on gates == CI 16:44:46 mriedem: Does NOva have microversion testing in Tempest going yet? 16:45:14 e0ne: Anyway, not yet for cinder microversions, but I'll work on that next. 16:45:24 scottda: i don't think so 16:45:30 oomichi: was working on that 16:45:35 scottda: sounds good 16:45:46 Tempest tests only become usefull when we actually bump the microversion with a change. 16:46:06 nova has functional testing for the microversion boundaries 16:46:08 in tree 16:46:25 scottda: the code for microversion support has landed in tempest 16:46:32 OK, I'll look at adding some functional tests as well. 16:46:46 the first test using it has been approved but bounced off the gate a couple times 16:46:52 pip 8 didn't help there 16:47:05 Anyway, please start looking at the code. It takes some time to soak it all in. 16:47:23 I'll put something on the mid-cycle agenda for microversions 16:47:31 That's all I've got. 16:47:36 scottda: Thanks! 16:47:47 scottda: fwiw: https://review.openstack.org/#/q/branch:master+topic:bp/api-microversions-testing-support 16:47:59 mriedem: thanks! 16:48:01 Definitely look into things before the midcycle if folks can. 16:48:32 #topic Open Discussion 16:48:44 OK, bike shedding time! 16:49:24 paint it black 16:49:46 Burn it down and get some more parking spaces 16:50:47 Put it in the reactor containment vessel 16:50:49 Nothing to discuss!? 16:51:01 did we decide if something was going on monday night? 16:51:03 Color doesn't matter there. Battleship grey will do! 16:51:08 patrickeast: +1 :D 16:51:17 When will you folks arrive? 16:51:19 patrickeast: I get in Monday afternoon. I'd love to meet up with everyone. 16:51:20 patrickeast: Beer 16:51:32 Anyone needs a ride? 16:51:37 yea i'm in the afternoon too 16:51:43 plenty of beer time 16:51:46 smcginnis: I will be aorund 16:51:46 Should we meet in the "official" hotel lobby at a certain time and head somewhere? 16:51:47 locals have a good monday night recommendation? 16:51:49 Me and e0ne will be there around 7 PM, any pointers where we should head? 16:52:04 smcginnis: +2 16:52:08 vincent_hou: your shiny new car can't wait to be loaded? ;) 16:52:09 smcginnis: +1 16:52:15 jungleboyj: You can scout out the good places. :) 16:52:25 right. I got a red car. 16:52:25 monday night at 11 PM, I'm sure I'll run into scottda or patrickeast out and about 16:52:36 jungleboyj: 'aorund' ? Have you started drinking already? ;) 16:52:39 smcginnis: Yes sir! Right on it sir! 16:52:47 dulek: We'll post in IRC and on the etherpad when we decide 16:52:47 diablo_rojo: :-p 16:52:53 e0ne, dulek: Will you have a mobile here, or would email be the best way to contact you? 16:53:03 scottda: Oh, good plan! 16:53:08 start drinking....dinner Monday 16:53:24 Meet in the lobby around 6:30-7? 16:53:42 sounds good to me 16:53:47 scittda: https://etherpad.openstack.org/p/mitaka-cinder-midcycle you mean put here? 16:53:50 create a Cinder google hangout ? 16:53:51 smcginnis: I'll send you may cell no 16:53:59 vincent_hou: yup 16:54:05 e0ne: Cool 16:54:14 hemna: good idea 16:54:44 OK, guess we're done here. Thanks everyone. 16:54:49 smcginnis: Sounds good to me. 16:55:03 #endmeeting