14:01:41 #startmeeting freezer 14:01:42 Meeting started Thu Mar 10 14:01:41 2016 UTC and is due to finish in 60 minutes. The chair is m3m0. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:43 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:45 The meeting name has been set to 'freezer' 14:02:00 hello m3m0 14:02:14 the weekly Freezer meeting today will be delayed by 10 minutes because of the source code walk through session 14:02:24 hello 14:02:32 daemontool__: do you want to be the chair today? 14:03:53 ok thanks 14:11:52 ok 14:11:54 here we are 14:12:17 topis and meeting notes available at: https://etherpad.openstack.org/p/freezer_meetings 14:12:50 s/topis/topic/ 14:12:52 so 14:13:10 #topic Should we do a one-time sync of our liberty and mitaka branches ? 14:13:45 m3m0, slashme szaher frescof yangyape_ EinstCrazy zhangjn et all 14:13:53 o/ 14:13:55 ok 14:13:56 :0 14:14:00 :) 14:14:02 i'm here 14:14:12 This was raised by reldan 14:14:18 The idea is : 14:14:32 0/ 14:14:51 Our mitaka (current master) and liberty branch are very very close because we backported nearly every commit 14:15:16 I think we have 2 issues there 14:15:29 1 is that liberty and mitaka were branched in a close timeframe 14:15:34 that is not going to happena nymore in the future 14:15:38 the other one 14:15:39 In order to simplify support of the liberty branch, reldan proposed that we do a sync of the two 14:15:51 ok 14:16:04 I'd focus more on find an agreement 14:16:11 on how to do this in the near future 14:16:15 and stick with that 14:16:46 doing any rebase 14:16:49 take time 14:17:02 probably it is better to focus on new features 14:17:05 testing and doc 14:17:08 rather backporting 14:17:12 unless we have serius bugs 14:17:31 So you agree that we should sync them ? 14:17:41 nope 14:17:54 because it is not what we are going to do in the future 14:17:56 most probably 14:18:02 agree daemontool 14:18:10 Yes I agree, it won't happen in the future 14:18:11 so I'll start doing things how we'll do it in the near future 14:18:12 from now 14:18:24 s/I'll/I'd/ 14:18:40 I think we can impove our doc for user. 14:18:49 zhangjn, definetely 14:18:55 I agree 14:19:07 reldan, sounds resonable to you? 14:19:18 Can we then take two minutes to agree on the workflow 14:19:24 so from now on 14:20:00 I agree with any solution. 14:20:06 o from now on 14:20:14 nobady use freezer just look README.rst 14:20:25 we will just backport the bugs 14:20:25 I think this a problem. 14:20:29 doc improvement 14:20:35 and testing 14:20:55 I think we need to stick to backporting bugfix, Even if this means rewriting a full change to manage the bug in older branches. 14:21:10 well the bugfix has priority 14:21:13 for sure 14:21:41 we can maange this case by case 14:21:42 otherwise testing testing testing 14:21:57 zhangjn, that's for sure something we have to anyway :) 14:22:07 For my point of view: 1) We need good tests - now we haven't 14:22:12 +1 14:22:14 reldan, ++ 14:22:15 And actually it is all 14:22:26 ++ 14:22:30 +1 14:22:33 +1 14:22:36 +1 14:22:37 If I know that my code works with mysql, mongo, windows, linux for restore, backup 14:22:38 we can write tempest tests now 14:22:48 I may sleep well 14:22:49 as we have the plugin 14:23:01 unit test first 14:23:12 Otherwise we should invest tremendous amount of time for manual testing and bug fixing 14:23:41 zhangjn, unit test we have 14:23:48 what we need it integration tests 14:24:04 I would prefer functional tests 14:24:18 Like I don’t sure that we capable to do 100GB backup 14:24:20 and restore 14:24:25 Just I don’t know 14:24:57 +1 on integration tests 14:25:04 I think integration tests need a way to autodeploy the freezer first 14:25:06 until few months ago 14:25:25 the agent was capable of doing >500GB backups 14:25:28 +1 integration tests 14:25:42 anyway 14:25:44 we can test it in our lab 14:25:46 this is a different topic 14:25:55 +1 on functional testing 14:25:58 all good for the backportig thing? 14:26:06 I would like to have a system, that will do 100GB backup restore every day or at least once per week 14:26:16 integration test is the topic 3 14:26:35 I think the periodic tests can help us doing that 14:26:36 I’m fine with backporting, but it is not a root of our problems. It a symptome 14:26:41 ok 14:26:52 next 14:26:59 #topic python-freezerclient 14:27:03 m3m0, 14:27:16 does the tests workd there? 14:27:20 s/word/works/ 14:27:25 the ui is capable of using the freezerclient, I'm integrating the scheduler now 14:27:32 ok 14:27:34 but I'm having some issues with testr 14:27:43 so we need to have tests working 14:27:48 or the build will fail at repo creation 14:28:02 what's the issue you are having? 14:28:04 maybe is my workstation but I need one more day to finish it 14:28:25 I think I might be executing the tests incorrectly 14:28:37 the tests are the same as the apiclient right? 14:29:03 most of them yes 14:29:22 is the current code on your repo? 14:29:39 yes, we still need to create the pypi repo and the gihub one, any news? 14:29:49 I'll do that after the tests works 14:29:56 >S 14:29:59 :) 14:30:02 :s ok 14:30:06 we need to know the tests works 14:30:10 that's for sure 14:30:27 all good with this topic? 14:30:27 sure 14:30:31 yes 14:30:31 can we move forward? 14:30:41 next 14:30:44 #topic Integration tests 14:30:46 ok 14:30:53 so where we left? 14:31:02 ddieterly, are you doing some other stuff on this regards? 14:31:33 ok 14:31:39 so we need the following 14:31:41 i'm trying to get the tempest test to exercise the freezer api 14:31:50 ah ok 14:32:02 that's what I was saying 14:32:05 if i can get one example, then the rest should be easy 14:32:17 discuss some Test Case 14:32:23 ok 14:32:53 probably the test cases should start from the agent? 14:32:54 test all api or some api first. 14:32:58 i'm also trying to get the gate job to run the tempest tests for freezer 14:33:08 being executed by the scheduler so we have all the flow tested? 14:33:37 ddieterly, I can give you some test cases for the agent 14:33:45 Monday 14:33:49 ok 14:33:52 or Tuesday 14:33:57 Tuesday most probably 14:34:04 if anyone has test case 14:34:11 please provide them to ddieterly 14:34:27 i would prefer if you wrote them yourselves 14:34:48 once one test is working, then there will be an example that people can follow 14:34:51 should be easy 14:34:51 sorry then I didn't get what you were asking before 14:35:05 the hard part is getting the gates setup and getting one tests to work 14:35:10 ok 14:35:31 ok 14:35:37 devstack is not work? 14:35:56 where is the bug to setup freezer? 14:36:25 which bug? not sure I'm following 14:36:34 devstck should work now 14:37:13 ddieterly, ok let us know if you need anything 14:37:16 or anything we can do 14:37:30 will do 14:37:37 ty 14:37:39 I agree with ddieterly, once the skeleton for the integration tests is working, adding them should be boring .. I mean "easy" 14:37:39 next 14:37:49 #topic source code walkthru session for scheduler we need to reschedule 14:38:07 so the source code walkthrough it happen only partly 14:38:12 this morning because of technical issues 14:38:18 we need to reschedule that 14:38:22 when? 14:38:28 :( 14:38:46 zhangjn, EinstCrazy yangyape_ it is ok for you to do it tomorrow? 14:38:54 ok 14:38:56 or Monday or next Thursday? 14:39:04 vannif, sounds good for you? 14:39:27 share screen is not work. 14:39:49 everyday is fine :) 14:39:54 I think the network is terrible for us. 14:39:56 yes 14:40:07 so let's do this tomorrow 14:40:09 same time? 14:40:14 you didn't see the sceeen ? 14:40:14 let's use web ex 14:40:32 vannif, please let's make sure 14:40:36 it always start in mandarin for me. idk why 14:40:38 the layout is english 14:40:41 no I don't see the share screen 14:40:42 lol 14:40:42 ojk 14:40:53 so tomorrow Morning please 14:40:55 let's do that 14:41:15 and also find a way of recording the session, so to be uploaded to youtube along with the previous one 14:41:23 hmmm. so. either I try it starting from *now*, or I record a demo and then we only handle questions in the meeting 14:41:39 doc instead of walkthough. 14:42:14 zhangjn, like documenting the code? 14:42:21 or the user documentation 14:42:22 ? 14:43:07 code comment 14:43:28 update the use document 14:43:37 we can review this patch 14:43:39 I would prefer code refactoring over code comment. If something is not clear - let’s make it clear 14:44:05 yes I think anyway 14:44:12 the documentation in the classes and functions docstrings 14:44:19 can be improved 14:44:24 If you can check our code and place //todo - not clear in places where it is not clear 14:44:25 so the documentation can be generated automatically 14:44:39 and send a pull requiest with such comments - it will be really cool 14:44:59 like #todo: nothing clear here 14:45:05 yes 14:45:10 ok zhangjn ty 14:45:21 Thank you! 14:45:42 this is a good way to learn and impove our freezer. 14:45:53 +1 14:46:12 ok 14:46:15 next 14:46:28 #topic What needs to be done for Mitaka 14:46:36 1) python-freezerclient package 14:47:07 2) try to get rysnc code in, if by next Tuesday is available with tests and doc, it will be in, otherwise Newton 14:47:10 3) ? 14:47:31 starting from Tuesday next week we should focus at least 1 week only on documentation improvement 14:47:35 and testing 14:47:39 stabilization 14:47:42 all of us 14:48:00 until 25th of March 14:48:38 anything more? 14:49:00 next 14:49:06 #topic Summit 14:49:10 so we have 1 talk approved 14:49:47 Congratulations 14:49:48 and 3 slots for the design 14:50:02 so we can get together and have dev and arch conversations 14:50:11 and discuss together many things 14:50:15 f2f 14:50:30 I hope to see all of you in Austin 14:50:36 anything to add? 14:50:50 next 14:50:54 #topic DR bp: https://review.openstack.org/#/c/278242 14:50:54 daemontool__: only my personal opinion here. I think that the rsync code integration would make more sense in newton with th plugable engines 14:51:08 But that does not really matter 14:51:14 the engines are already pluggable 14:51:38 probably there are more code we can abstract 14:51:45 but the basic abstraction is done 14:52:19 about the engines, yes. I'd like to see the documentation on how it works, the file formats, the structure of any metadata etc. before any code implementation actually 14:52:32 the rsync I mean 14:52:47 vannif, ok 14:53:12 the implementation already started many months ago 14:53:13 :) 14:53:22 but it make sense 14:53:23 yeah, I know :P 14:53:32 to provide the necessary information 14:53:44 but if anyone wants to check the code, help with the implementation ... 14:53:58 Added as a topic for the midcycle meetup 14:54:13 having an idea of how it is supposed to work is fundamental 14:54:26 the ongoing review is here https://review.openstack.org/#/c/290461/ 14:54:33 still work in progress 14:55:12 ok 14:55:23 so by Monday we should have a first implementation 14:55:26 with the metadata 14:55:29 that can be changed anytime 14:55:34 and we have something 14:55:36 more tangible 14:55:38 we can discuss on 14:56:13 vannif, I'd like to have some feedback 14:56:21 not on the rsync code itself at this stage 14:56:27 but on the way the incremental_engine 14:56:30 is organized in the code 14:56:39 what can be abstracted and more 14:56:59 but IMHO, we have many things to do 14:57:16 so 1 engineer for now full time is more than enough on that 14:57:27 but reviews needs to be done anyway 14:57:33 yes. but in the end it will be a piece of art :) 14:57:44 as everything :) 14:57:45 so 14:57:51 let's move forward 14:57:59 #topic DR 14:58:05 frescof, are you around? 14:58:11 I see you updated the dp 14:58:12 bp 14:58:23 https://review.openstack.org/#/c/278242 14:58:36 can we please add all our considerations there? 14:58:36 Still 14:58:41 the same problem 14:58:53 slashme, ok 14:59:00 write it down there 14:59:19 I'd like to have some discussion reported there 14:59:27 then we'll ask for feedback on the openstack ml 14:59:34 m3m0 jonaspf and me started thinking about a viable solution 14:59:45 so the community can see our discussion 14:59:54 slashme, fantastic 15:00:11 We'll be happy to explain the draft at the midcycle and then get the opinion of the comunity 15:00:19 ok 15:00:26 we have one topic left 15:00:35 let's do that in #openstack-freezer please 15:00:43 #endmeeting 15:00:49 m3m0, close the meeting please 15:00:51 :) 15:00:52 thanks 15:00:57 tkx 15:00:59 thx 15:01:01 #endmeeting