16:00:23 #startmeeting Cinder 16:00:24 Meeting started Wed Feb 24 16:00:23 2016 UTC and is due to finish in 60 minutes. The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:26 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:28 The meeting name has been set to 'cinder' 16:00:29 o/ 16:00:29 Hello 16:00:30 Hi! 16:00:33 Ping: dulek duncant eharney geguileo winston-d e0ne jungleboyj jgriffith thingee smcginnis hemna xyang tbarron scottda erlon rhedlind jbernard vincent_hou kmartin patrickeast sheel dongwenjuan JaniceLee cFouts Thelo vivekd adrianofr mtanino yuriy_n17 16:00:33 hi 16:00:34 o/ 16:00:35 hi o/ 16:00:36 hello.. 16:00:39 hi 16:00:39 hi 16:00:39 hi 16:00:42 o/ 16:00:42 hi! 16:00:43 hi 16:00:44 Hello! 16:00:47 o/ 16:00:49 hi 16:01:00 hello 16:01:02 * bswartz lurks quietly 16:01:04 o/ 16:01:06 hi 16:01:10 #topic Announcements 16:01:11 Greetings and salutations 16:01:20 hi 16:01:20 Good day all 16:01:20 We're very quickly coming up on deadlines. 16:01:26 #link http://releases.openstack.org/mitaka/schedule.html Mitaka deadlines 16:01:29 wait.. that would imply I"m leaving 16:01:31 Hello :) 16:01:45 bswartz: don't be creepy 16:01:50 :) 16:02:04 Hello! 16:02:12 We need to wrap up os-brick ASAP. 16:02:19 hey 16:02:27 Also need to make sure whatever we need in the client gets some focus. 16:02:48 We can always realease a newer client, but it would be nice to have everything we need in there for the "final" Mitaka cut. 16:03:08 So.. review, review, review. :) 16:03:23 #topic Oslo Releases 16:03:32 Not sure who put this one on the agenda. 16:03:38 smcginnis: That was me. 16:03:51 jungleboyj: Ah, good. Take er away. 16:03:55 smcginnis: Just wanted to share that we should now have the last oslo library releases out. 16:04:12 Possible exception to oslo-privsep. 16:04:20 We didn't have any bugs there that I am aware of. Oh, yes. 16:04:47 * smcginnis probably should have put os-brick/privsep on the agenda 16:04:54 What happens if there's a critical issue in oslo.sth? 16:05:05 Are fixes and global-req bumps allowed in RCs? 16:05:13 I am planning on working to get what is left of incubator pulled out. 16:05:19 dulek: I believe we have in the past. 16:05:37 smcginnis: Yes, I think so. 16:06:02 jungleboyj: Awesome, would be good to get that cleaned up. 16:06:08 jungleboyj: e0ne was working on pulling out imageutils, but he failed. 16:06:16 smcginnis: Agreed. 16:06:20 jungleboyj: I don't remember why exactly. 16:06:28 dulek: I wondered about that as I was looking at what was left in there. 16:06:55 I will work on getting as much cleaned up as I can over the next few weeks. 16:07:36 jungleboyj: Thanks, I guess we can get as much pulled out as we can for now and clean up later. 16:07:40 That was all I had to share from Oslo. 16:07:45 jungleboyj: Thanks! 16:07:57 #topic Replication and microversions 16:08:08 * jungleboyj tries not to laugh 16:08:25 So these were some of the higher priority items we identified for Mitaka. 16:08:32 I think they are both very close. 16:08:58 microversions looks good to me - I'd have +2d it except I try to avoid doing that with code from my own team 16:08:58 If we can get some cores to sign up for reviews on those, that has worked well in the past I think. 16:09:12 DuncanT: We're on the same team? 16:09:12 smcginnis: if dulek is ok with it now I say merge it! BEFORE eharney 's patch preferably :) 16:09:21 haha 16:09:21 Hah 16:09:34 jgriffith: I agree 16:09:42 at least I didn't loose another db migration :) 16:09:50 I've done limited testing on both, and I'm fine with getting them in now. 16:09:57 smcginnis: FWIW I do plan to follow up with more cleanup and "old" code removal 16:10:00 jgriffith: I'll look into the new patchset, but it should be cool. 16:10:04 but wanted to get the base in 16:10:11 yes. it looks fine now 16:10:17 I'm sure we'll find _some_ issue, probably with both, but that should just be a bug fix. 16:10:28 smcginnis: yes, agreed 16:10:31 The overall direction and behavior looks right to me. 16:10:37 smcginnis: perfect is the enemy of the good IMHO 16:10:44 jgriffith: +1000 16:10:50 :) 16:10:56 jgriffith, thought list_replication_targets was going away. 16:10:59 jgregor: ++ 16:11:09 i'm ok with just bumping my rpc version and setting depends-on jgriffith's patch if we think it would help everything slide in together 16:11:10 Swanson: it did 16:11:17 Moving on before Swanson derails things... 16:11:17 :P 16:11:19 smcginnis: So should we name a date for reviewers to "speak or hold your peace"? 16:11:20 scottda, I presume we have a client patch to test microversions as well ? 16:11:28 and is that also ready? 16:11:28 Swanson: I think you're seeing the old V1 or V2.0 version in there 16:11:30 but mine is ready to approve like... now, so it's kinda up to whichever is first 16:11:37 hemna: I'm still adding changes to the client for new /v3 endpoint 16:11:43 dang 16:11:43 ok 16:11:47 scottda: Sure. If anyone has any major concerns, comment on those patches now. 16:11:55 * DuncanT has been testing with curl 16:11:55 I can rework it with the rpc bump after eharney 's lands.. no big deal 16:11:58 Otherwise I'd actually like to see them go in today. 16:12:02 previous client patch worked, but I'm trying to get keystone to ACK the new endpoint ATM 16:12:18 * DuncanT will look at replication this evening 16:12:53 #topic Cinder-Nova API changes 16:12:56 scottda: Hey 16:13:01 hi 16:13:13 We've identified issues with cinder-nova: 16:13:29 need to be able to update Cinder's connector info for live migration & evacuate... 16:13:45 need to pass in instance_id and host for multi-attach.. 16:13:53 need to pass in connector at attach time.. 16:13:58 probably some more. 16:14:14 and update the connector 16:14:18 So, we're attempting to corral all this stuff and come up with a cross-project spec for fixes. 16:14:32 hemna: Yes, that's up there ^^^ 16:14:34 There is a ML topic on this as well 16:14:38 scottda: so I don't think you need all of that 16:14:39 scottda: Will be nice to get that captured in one place. 16:14:40 for anyone that cares 16:14:56 scottda: I'm working on a patch now that I can submit later today 16:15:20 scottda: there's *another* misunderstanding/misuse of detach and terminate connection similar to what we had on attach/initiatlize 16:15:29 jgriffith: OK, well It'd be great to save a bunch of work. But there seem to be a bunch of IRC discussions and ML posts around all this. 16:15:51 scottda: but IMHO the correct fix is to build the volume_attachment entry using the connector during initialize_connection 16:15:56 jgriffith, well as long as the connector information is available to drivers that need it at terminate_connection time 16:16:07 scottda: then you don't need it any more on the detach/terminate because you "already" have it 16:16:36 jgriffith: That sounds great. We'd still need to be able to update connector info for nova migration and evacuate, I think. 16:16:37 hemna: yes, it will be 16:16:39 jgriffith, we still need to update it for live migration 16:16:53 scottda: you can be greedy if you want 16:16:59 jgriffith, look forward to seeing the patch :) 16:17:04 hemna: scottda ok, I'm out then 16:17:05 thanks for that 16:17:10 I'm not going to boil the ocean here 16:17:35 hello all..(joined late :( ) 16:17:36 pick a problem and solve it.. I can't solve all of the 3par problems either 16:17:48 jgriffith: Your patch sounds helpful, but there are a few things that need doing and I'm just trying to tame the complexity of it all: 16:17:49 https://etherpad.openstack.org/p/cinder-nova-api-changes 16:17:52 because I don't even know/understand what all of them are 16:18:07 jgriffith, it's not just 3par that needs the connector at terminate 16:18:09 So, anyone interested , sign up and pay attention. 16:18:09 #link https://etherpad.openstack.org/p/cinder-nova-api-changes 16:18:23 I think it will be good to have this captured somewhere. 16:18:31 hemna: why don't you pick which topic you want to discuss here? 16:18:33 We've had too much speculation on what different backends need. 16:18:46 And what nova does or does not need and what cinder does or does not need. 16:19:03 At least having it captured in a spec helps us get closer to actual data. 16:19:03 hemna: I'm fixing the incorrect code in Cinder that causes us to not have connector info at terminate_connection 16:19:26 great, like I said I'm looking forward to seeing the patch 16:19:27 jgriffith: That sounds like it will solve a whole lot of problems. 16:19:33 I'll pull it and play with it. 16:19:44 hemna: I don't know about the live-migration issues for 3par etc... but I'm not going to spend hours reworking code to have it rejected for a specific drivers evacuate or live-migration problem 16:19:50 hemna: ok 16:20:18 Yep, let's work it out once we have code to look at. 16:20:19 FTR I'm not trying to reinvent the wheel or solve all of the worlds problems at once. Just one thing at a time please 16:20:46 scottda: Anything else to add for Cinder-Nova? 16:21:14 jgriffith: It has been hard to communicate with NOva about this. That is part of what we are trying to do. 16:21:25 smcginnis: No, that's it. 16:21:35 scottda: Thanks 16:21:43 #topic os-brick with privsep 16:21:51 I wanted to bring this up here. 16:22:01 We have a patch out there waiting a requirements update. 16:22:13 And honestly I think that's going to come through too late. 16:22:21 But I want to make sure folks are aware of it. 16:22:38 * DuncanT thinks that this provsep patch is totally broken anyway 16:22:38 smcginnis, did you see Angus' ML post ? 16:22:43 I've had a LOT of pressure to make sure this is available in os-brick in Mitaka. 16:22:49 hemna: Yep, I think that was good. 16:23:02 so I plan on pulling his latest and test it locally again 16:23:07 it's been very unstable 16:23:16 The main point of contention is we currently have a locking with Nova. 16:23:18 I'm concerned about breaking the world right now 16:23:28 hemna: Thanks, let me know what you see with the latest. 16:23:31 hemna: Me too. 16:23:40 this is a big change with big risk 16:23:40 So I'm extremely hesitant to allow this in. 16:24:00 But I do understand the concern that right now we need to sync up rootwrap filters with cinder and nova. 16:24:15 So we have a compatiblity problem between versions. 16:24:25 But that was there in Liberty. 16:24:28 smcginnis: But I think it's not an issue for Mitaka, since we didn't add any new filters to Nova. 16:24:34 so we technically have 2 days to release client libs 16:24:36 And it's not going to be any worse in Mitaka 16:24:37 according to the schedule 16:24:41 scottda: Right 16:24:55 So saying it here - I don't think this will go in. 16:25:05 I think we are safer by not landing this just now 16:25:09 It's too late and there's not enough comfort level with it. 16:25:10 and try and get it in early N 16:25:27 so save the effort and look at it then? 16:25:27 But I know that will be challenged, so just making it clear. 16:25:28 I don't think there is any urgent reason to get it in. 16:25:41 we could always back it out, I suppose 16:25:48 I agree, early on in N and we can get runtime on it. 16:25:52 but I think there are cinder and nova patches that go with this change 16:26:00 hemna: Yes 16:26:07 hemna: Another good point. 16:26:25 yup 16:26:29 That's all I had to say about it. Just wanted to raise it here so everyone is aware. 16:26:35 #topic Open Disucssion 16:26:44 Anyone got anything? 16:26:54 Quick question on the spec tracking page 16:27:06 Once something is totally merged, can we just delete it off that page? 16:27:20 (or abandoned complete as in one case) 16:27:20 Yeah, good point. 16:27:24 That should be cleaned up. 16:28:28 * dulek removed rolling upgrades, last +2/+A already happened. 16:29:17 smcginnis: midcycle meetup plans? 16:29:20 dulek: Removed? 16:29:29 bswartz: Start planning for the next one? 16:29:43 I just wondered if any progress was made towards planning 16:29:47 Dublin this time? 16:30:02 jungleboyj: From etherpad. :D 16:30:08 I would still like to do a Europe midcycle for our non-US folks. 16:30:13 if it's international more time is needed to plan travel 16:30:19 dulek: Ah! Thank you. 16:30:20 Might be the last midcycle before it turns into something else. 16:30:24 Unfortunately. 16:30:25 Context is everything. 16:30:30 We can host in Israel if people want 16:30:31 bswartz: Good point. 16:30:40 DuncanT: Hah, that could cause issues I think. :) 16:30:46 I would like to, but I'm sure HPE Roseville wouldn't get travel approval at all for it. 16:30:47 Though I would love to. 16:31:12 Intel@Gdansk may be also an option, it's quite nice here in the summer. 16:31:14 hemna: None? 16:31:18 nope 16:31:21 smcginnis: I don't think changing the way summits & design sessions work will eliminate the need for face-to-face time, i.e. mid-cycles. Just my opinion... 16:31:24 I would need to confirm with management though. 16:31:37 In fact if I talk to NetApp Tel Aviv, we can get decent wifi on the beach.... 16:31:37 scottda, +1 16:31:44 scottda: +! 16:31:45 DuncanT, nice! 16:31:51 scottda: In some views of the proposal, the design summit becomes the midcycle. 16:31:51 lol +1 16:31:58 Depends how that all plays out. 16:32:04 * patrickeast wanders in late 16:32:06 what about india? 16:32:14 anyone interested? 16:32:15 smcginnis: We'd still need mid-cycles in between, call them what you will. 16:32:17 wait -- our tel aviv office is on the beach? 16:32:25 don't we have any companies in Hawaii? 16:32:37 sheel: I have my visa, but I don't know if everyone can get through that process in time. :) 16:32:45 scottda: I agree. 16:32:53 And I really really hope we don't lose those. 16:33:06 smcginnis: ok 16:33:20 But if vendors need to send folks to the Summit, the design summit, _and_ a midcycle... 16:33:22 bswartz: I think so, yeah 16:33:33 bswartz: It's either yours or a company here you own 16:33:35 smcginnis, +1 I don't really care for the proposed changes myself. 16:33:52 hemna: I agree. Going to make things even harder. 16:33:54 hemna: I have a lot of... concerns. 16:34:05 LA is an option. there are some openstack related companies around that might be able to host. 16:34:11 Seems like a tops-down decision 16:34:20 thingee: +1 16:34:28 and specifically in my hometown pasadena where everything you could want is in walking distance. 16:34:40 Well, for now, this is a good notice. 16:34:42 thingee: +1 16:34:56 Anyone interested/willing to host the next midcycle, start asking internally I guess. 16:35:06 An international mid-cycle will have smaller participation from IBM I am sure. 16:35:13 Especially non-US that would be in a good, convenient location for folks. 16:35:20 People who have concerns should shout up on the list, otherwise the change will happen 16:35:24 Thinking the usual late july early August time? 16:35:37 jungleboyj: Yeah, that sounds good to me. 16:35:38 jungleboyj: yea 16:35:51 smcginnis: You should get Dell to do it so people could stay in the cities. 16:36:09 I really would like to see, assuming they stick around, we have 1 out of 5 or so midcycles internationally if possible. 16:36:28 smcginnis: right, may be i need some idea what requires in hosting one...silly question though... 16:36:31 diablo_rojo: Better than this neutron midcycle in February in MN. ;) 16:36:42 Thought this one actually worked out OK. 16:36:47 smcginnis: Truth. Also in Rochester.. 16:37:12 sheel, working wifi 16:37:21 kmartin: done 16:37:29 and? 16:37:38 sheel: Easy for people to get to 16:37:38 :) 16:37:41 sheel: A room capable of holding 35 or so people, good wifi, a near by hotel 16:37:47 projectors, larger rooms... 16:37:56 DuncanT, by a beach.... 16:38:00 sheel: Free gourmet food 16:38:01 hemna: ++ 16:38:13 and some might include "good beer" in that list 16:38:21 kmartin, +A 16:38:23 DuncanT: don't forget the importance of the dance scene 16:38:26 kmartin: Loves beating a dead horse. 16:38:31 I think all seams ok except beach 16:38:33 sheel: you take care of visa for everyone:) 16:38:33 kmartin: 90%+ of survey respondents always do 16:38:34 Really the requirements are near a mostly convenient international airport, a room for 25-30 people, and working network. 16:38:48 xyang: hehe 16:38:53 xyang1: hehe 16:39:06 smcginnis: it can be arrange 16:39:12 smcginnis: I will discuss 16:39:12 sheel: Where are you again. On the beach in Goa? :) 16:39:17 smcginnis: intenarlly 16:39:18 smcginnis: Also tables that can be arranged in a U or something like that is better than separate round tables. 16:39:29 Raleigh was good though. cheap flights 16:39:40 sheel: Cool, thanks! 16:39:47 smcginnis: nopes...in delhi.. 16:39:49 Good dancing in Raleigh too 16:40:22 Raleigh was good. I'd do Ft Collins any time. CA could be nice if it's not in the bay area. 16:40:22 diablo_rojo: dance video? 16:40:23 Raleigh did work pretty well also. 16:40:33 smcginnis: Love Fort Collins! 16:40:42 Ft collins is always available. I just don't want to hog it. 16:40:44 xyang1: I have video of dancing but jungleboyj didn't drink enough to dance. 16:40:48 I do love Ft. Collins 16:40:50 :) 16:40:53 +1 16:40:56 +1 16:41:06 diablo_rojo: I am disappointed:) 16:41:12 OK, well folks think about it and get some wheels turning. 16:41:22 Anything else before I call it? 16:41:25 xyang1: I will get him line dancing at some point and get you a video. 16:41:47 xyang1: I am sorry. Next time. 16:41:48 diablo_rojo: looking forward to it 16:41:51 Thanks everyone! 16:42:05 And now that conversation is forever immortalized. 16:42:05 #endmeeting