14:00:11 #startmeeting Glance 14:00:12 Meeting started Thu Apr 9 14:00:11 2015 UTC and is due to finish in 60 minutes. The chair is nikhil_k. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:14 o/ 14:00:15 The meeting name has been set to 'glance' 14:00:15 o/ 14:00:19 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:00:21 o/ 14:00:24 o/ 14:00:39 o/ 14:01:07 cool, looks like a decent turn out for a small agenda 14:01:19 A couple of things before we start 14:01:26 * flaper87 won't be able to put much attention to the meeting today. Helping out in a workshop right now 14:02:04 o/ 14:02:08 I would like to encourage everyone to keep this meeting small as I understand that many folks have been busy with fixing things in their cloud at this phase of the cycle. 14:02:12 o/ 14:02:20 We are closing in on RCs 14:02:37 o/ 14:02:40 Though, I would be happy to chat and address any concerns after the meeting 14:02:44 \o 14:02:54 Let's get started 14:03:06 #topic RC plans 14:04:07 #link https://launchpad.net/glance/+milestone/kilo-rc1 14:04:18 I had a chat with ttx and looks like the important bugs tagged for RC1 are fixed 14:04:53 I've asked this one to be pushed to RC2 : https://bugs.launchpad.net/bugs/1434237 14:04:55 Launchpad bug 1434237 in Glance "glance-manage db_export_metadefs fails with NoSuchColumnError" [High,In progress] - Assigned to Ashish (ashish-jain14) 14:05:22 Although, the idea is that we will not have RC2 if things are stable 14:06:20 we do however, need to address any bugs that are found while testing the RC1 tarball so I have encouraged him to consider RC2 possibility quite seriously 14:06:55 RC1 may be tagged as early as today or as late as mid-next week 14:07:07 Any concerns/questions? 14:07:37 nope 14:07:56 newbie question, How is the RC1 testing done? 14:08:04 tempest on gate only? 14:08:29 The intent is that people will test them in thier environment 14:08:43 Not all things are covered (can be covered) in tempest 14:09:21 nikhil_k, agreed! people file bugs in launchpad tagged with RC1? 14:09:33 So, you download the tarball , deploy it and test it with tempest and other tests (including manual functional test) especially negative tests 14:10:13 ajayaa: No, you file a bug and bring to bug supervisor's attention to get it tagged 14:10:26 then ttx and I will have to discuss the possibility 14:10:32 cool. Thanks for the clarification. 14:10:52 bug supervisors can be found at : https://launchpad.net/~glance-bugs 14:10:56 #link https://launchpad.net/~glance-bugs 14:11:22 Moving on. 14:11:31 #topic Kilo review priority, blockers 14:11:39 #link https://bugs.launchpad.net/glance/+bugs?field.tag=kilo-rc-potential 14:12:12 That should be our focus right now. Plus any Critical/High bugs filed in the near future. 14:12:44 I would encourage everyone to start identifying anything that we need to consider for RC2 14:13:45 Please try to review High priority bugs that are pending review soon and let's start removing them from the list if they do not look to make it in Kilo 14:14:17 Any concerns/questions? 14:14:25 all sounds good 14:14:48 One more thing that everyone should be aware about: 14:14:51 #link https://review.openstack.org/#/c/171227/ 14:15:14 It opens Liberty development. We can do so after Kilo looks all set. 14:15:35 I would like to take vote from everyone in the next meeting for the same. 14:15:50 Thanks. 14:16:06 Moving on. 14:16:20 #topic Pending reviews 14:16:38 https://review.openstack.org/166909 14:17:01 That's tagged for Kilo. Lets try to sort it out soon. 14:17:41 That's it from my end. 14:18:15 Looks like no one has reviews to bring up today? 14:18:23 There is merge conflict on the patch and last patch-set landed on 27th March. 14:18:42 How can the reviewers help in this situation? 14:18:42 yep, it also has a -1 14:19:22 I was trying to gauge if we should omit it now and fix in L1 14:19:44 It looks quite trivial to fix that -1, but I am not sure about the merge conflict 14:20:31 cool 14:20:35 #topic Open Discussion 14:21:23 Everyone excited about Vancouver? 14:21:27 We have been pushing for NoSql backend for glance based on Cassandra for sometime now. 14:21:53 But there are some apis which kind of assume that the backend is a relational DB. 14:22:07 I have proposed a session around this in etherpad. 14:22:40 One example of such api would be arbitrary sort key in list images api. 14:22:42 ajayaa, yeah, we need to discuss it there 14:22:42 ajayaa: what does the mongodb backend do in those cases? 14:23:18 kragniz, If there is a mongodb backend for glance, that's news for me. 14:23:20 :) 14:23:44 ok, we do not have nosql support yet 14:24:06 The basic idea is, if your data is distributed among many nodes then it very expensive do sorting. 14:24:16 Also, the db code needs a major refactoring to make this happen. It might affect some queries that I'm hesitent about. 14:24:59 There's more feedback but I will try to bring it up at the discussion. 14:25:24 feedback from your side, nikhil_k? 14:25:30 ajayaa: I have some experience with cassandra and your proposal seems very exciting to me. However, it seems hard to me to build a proper abstraction layer which will make these backends interchangeable 14:25:31 ajayaa: duh - I should have looked at what gridfs actually was 14:25:35 ajayaa: thanks for your proposal though. Yes, on feedback. 14:26:07 nikhil_k, looking forward to the summit. 14:26:11 ajayaa: are you involved with the keystone effort on this? 14:26:35 (I'm under the impression this was being discussed for keystone) 14:26:36 mclaren, We are the ones who have proposed Cassandra backend for Keystone. 14:27:14 We have a working POC for Keystone with Cassandra which seamlessly works across regions. 14:27:15 ok. Personally I'd be interested to see how that goes. We could then make a more informed decision around if it makes sense for glance 14:27:31 (regions in s small lab setup though* :)) 14:27:45 based on kragniz's comment I'm wondering what the backend proposal is 14:28:21 I didn't understand how gridfs came into picture! 14:28:23 I think we need to be careful to not take on too much for one cycle also 14:28:32 ok 14:29:02 ajayaa: gridfs is an abstraction on top on mongodb 14:29:34 Alright, seems like there is plenty food for thought! 14:29:41 kragniz, We don't like mongodb too much the reason being the differentiation of config servers and normal server. 14:30:06 To me personally, C*(Cassandra) is a truly distributed fault tolerant db. 14:30:34 okay, We can have this discussion somewhere else or at the summit perhaps. 14:30:44 summit sounds good 14:30:51 ajayaa: Yes, that would be great. With some examples to shared would help.. 14:30:58 share* 14:31:03 ajayaa: I'm arguing for mongodb! I don't know very much about either 14:31:09 summit would be good 14:31:18 Anything else? 14:31:31 summit sounds good, if you have anything to share before (that perhaps would help us to get up to speed in the summit) please send any refs to the mailing list 14:31:39 +1 summit discussion 14:31:41 == jokke_ 14:31:46 jokke_, sure. 14:32:13 one more thing 14:32:47 how does bug day sound, lets say RC1 + 3 working days? when ever that RC1 gets tagged 14:33:14 +1 14:34:13 If nothing more.. 14:34:15 I think we would have anything Major (like totally rendering glance unusable) by that and well enough time to evaluate the likelyhood of us needing rc2 14:34:24 jokke_: sounds good 14:34:49 agreed 14:35:06 let's kick e-mail to the ML advertising the day when we get the tag 14:35:06 Thanks all! 14:35:09 thanks 14:35:14 thx 14:35:21 #endmeeting