17:00:26 #startmeeting qa 17:00:27 Meeting started Thu Jan 31 17:00:26 2013 UTC. The chair is jaypipes. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:28 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:30 The meeting name has been set to 'qa' 17:00:39 good morning QAers 17:00:49 good morning 17:01:00 hi 17:01:11 jaypipes: morning 17:01:32 lots of open reviews :( 17:01:43 * jaypipes bad for not being able to do reviews :( 17:02:42 why not being able to do reviews? 17:02:54 chunwang: w$rk getting in the way :) 17:03:04 hi 17:03:29 ravikumar_hp: heey 17:03:37 Here 17:03:41 Hi 17:03:54 * jaypipes hoping that afazekas or davidkranz_ might be able to bring me up to speed on what's gone on in tempest over past week. 17:04:15 jaypipes: Mostly a lot of new code submitted. 17:04:36 jaypipes: And the formal proposal from sdague to gate all projects (except horizon) on full tempest. 17:04:43 yes, I saw that 17:05:02 jaypipes: I'm not sure why the reviews have piled up but we should discuss that. 17:05:21 davidkranz_: afazekas has pushed >50% of the current queue 17:05:27 davidkranz_: lack of reviewers/approvers is my guess ;) 17:05:31 I'm sure that has something todo with it 17:05:36 indeed! 17:05:43 jaypipes: Yes, but the question is why? 17:06:10 davidkranz_: people working on other things, I assume 17:06:21 Speaking for myself, I don't know anything about the boto stuff so did not feel I could review it. 17:06:33 understood 17:07:00 #topic performance of tempest 17:07:29 I'd like to get an update from sdague and others about the testtools conversion and improvments in performance. 17:07:36 jaypipes: OK, but I want to go back to the review topic later. 17:07:53 jaypipes: sdague can't make it today 17:08:11 mtreinish: do you have enough info about the topic to speak on it? 17:08:51 jaypipes: unfortunately, not really. I know that the testtools conversion got merged, but I don't know about the status of testr 17:09:12 ok 17:09:18 we will leave that to next week then 17:09:23 #topic Reviews 17:09:46 davidkranz_: I also don't know much about the Boto stuff which is why it's been difficult to do for me. 17:10:01 jaypipes: Tempest has become very broad. 17:10:20 davidkranz_: yes indeed. 17:10:28 jaypipes: our team is writting boto tests for Ec2 . 17:10:41 Jaypipes: do you think developers can review 17:10:47 ravikumar_hp: which developers? 17:11:00 Nova dev team that supports Ec2 .. 17:11:06 jaypipes: I think we need to subdivide into groups that can review particular topics. 17:11:38 Or in some cases not try to verify every aspect of the tests which involves detailed understanding of the api. 17:11:47 davidkranz_: +1 17:11:48 ravikumar_hp: tough to identify those people... 17:11:56 This is not ideal but we need to change something. 17:13:41 How about this: 17:13:55 davidkranz_: perhaps we just need to institute tempest review days 17:14:15 If a submission goes more than 24 hours with no review, some one with less specific knowlege can review it. 17:14:41 davidkranz_: anyone should be able to review any patch at any time... 17:15:06 davidkranz_: if the reviewer is not a subject matter expert, they can still comment on style, robustness, and other things. 17:15:49 jaypipes: Right. So we could say that a core review +1 means "style, etc" is OK but not verified all the functions. 17:16:09 But what if no one feels comfortable giving a +2? 17:16:21 sure, that makes total sense, and that's pretty much what I've been doing 17:16:40 jaypipes: OK, then the issue may be (lack of) review days. 17:16:48 davidkranz_: if nobody feels comfortable +2ing then we need to bring in the devs from Nova and other projects to make a decision 17:17:16 So how long should a submitter expect to wait for an approval? 17:17:18 davidkranz_: but I still think review days are upon us 17:17:33 davidkranz_: right now... upwards of a week, which is unacceptable 17:17:39 I would like a 24 hour turnaround ideally. 17:17:45 davidkranz_: we should endeavour to keep it <48 hours IMHO 17:17:59 davidkranz_: for at least an initial comment on the review 17:18:21 jaypipes: Do you know how review days are set up on other teams? 17:18:32 davidkranz_: in addition, adding more people to oepnstack-qa-core would be helpful. 17:18:32 davidkranz_: jaypipes: yes . better less tahn 24 hrs as many reviews go through 4 or 5 patches 17:18:50 davidkranz_: it's up to the PTL in other dev teams... 17:19:10 jaypipes: OK, just looking for an example to follow. 17:19:12 davidkranz_: I think we need to have weekly review captains to herd everyone and get reviews pushed 17:19:48 jaypipes: Yeah. Is there a way to "request review" if you are not the submitter? 17:19:55 The button doesn't seem to work. 17:20:07 davidkranz_: if it doesn't work, it's a bug and notify mordred 17:20:08 * afazekas here 17:20:09 That's what the captain needs to do 17:20:30 The button works if I am the submitter but I can't add a reviewer to some one else's change. 17:21:13 jaypipes: Grr, I think I misunderstood the UI. Never mind. 17:21:19 davidkranz_: k 17:21:39 Should we maintain the captain list in an etherpad or wiki? 17:21:54 davidkranz_: yes, I think so. 17:21:57 davidkranz_: I was able to add reviewers .. 17:22:08 ravikumar_hp: I misunderstood the UI. 17:22:09 in UI for others submission 17:22:45 davidkranz_: so I would propose: 1) increase number of people on openstack-qa-core, 2) have weekly review captain on rotation, and 3) have qa-core members assigned to review days of week. 17:23:04 BTW, what's the qualification to be a patch reviewer of tempest? 17:23:10 jaypipes: Sounds good. 17:23:11 chunwang: nothing. 17:23:21 chunwang: anyone is encourage to do reviews 17:23:30 jaypipes: I would also have an entry for each project where people can put their names as knowing something about it. 17:23:37 chunwang: however only members of openstack-qa-core can +1 Approve a patch. 17:23:43 sorry, +2 Approve... 17:23:58 davidkranz_: excellent. can you handle creating those pages? 17:24:03 jaypipes: +1 , 1) 2) is fine 17:24:06 jaypipes: That list could be use by the captain to add reviewers when necessary. 17:24:18 davidkranz_: ++ 17:24:46 Sure. Etherpad is easier to edit but wiki is less subject to screwup. 17:25:03 ty 17:25:03 jaypipes: I'll go with the wiki. 17:25:08 k 17:25:38 jaypipes: got it 17:26:21 OK, so there are a number of folks that could be proposed to qa-core... I will be watching over the next couple weeks for folks who are actively doing reviews and following up. and I will propose those folks for qa-core 17:26:29 I will send out an email shortly saying that. 17:26:30 ravikumar_hp: Can you guys review the boto stuff today? 17:26:49 davidkranz_: will try . 17:27:09 Guess I'm captain this week :) 17:27:13 davidkranz_: indeed 17:28:25 davidkranz_: do you have the review id/subject handy ? 17:28:26 davidkranz_: link ? 17:28:57 ravikumar_hp: There are several of them in the queue. 17:29:11 afazekas: I think davidkranz_ is talking about your boto patches 17:29:19 mtreinish: That's right. 17:30:05 davidkranz_: ok, I thought I forget to review one change .. 17:30:43 afazekas: We need to review your changes. 17:31:12 afazekas: You are doing a lot of stuff and it would help is there were blueprints that provided some context for reviewers. 17:31:42 davidkranz_: small changes 17:32:11 email sent... 17:32:16 davidkranz_: bug fix and reformating mostly 17:33:16 afazekas: I understand but we need to be careful about changing code just because we like it better a different way. 17:33:17 afazekas: Along this note, I meant to ask if any of the reformats are related can you put them in a series with the same branch. Even just that would help with the context. 17:33:49 mtreinish: ++ 17:36:16 Any other topics? 17:36:33 davidkranz_: there is the coverage analysis I've been doing 17:36:35 davidkranz_: none from me. I'm eager to get on reviews... 17:36:42 I sent a mail out to the list about it 17:36:44 mtreinish: oh yeah... how is that going? 17:36:48 mtreinish: now these changes are mostly independent, but mostly have second step, but these next step codes are even not created 17:37:09 davidkranz_: yes we will like to discuss more about it 17:37:17 https://etherpad.openstack.org/coverage-analysis and https://etherpad.openstack.org/MissingTempestTests 17:37:19 afazekas: I don't understand that last comment... could you explain? 17:37:47 mtreinish: ah, well done. 17:37:56 jaypipes: it's going well, I've found some gaps, mostly on negative tests 17:38:20 I'm still not done, but this what I've found so far 17:38:21 jaypipes: I modified lot of different unrelated part of the tempest , without marking the comment to dependent to each other 17:38:38 afazekas: that makes reviews quite difficult... 17:38:49 jaypipes: As I said the bug fix commits will be continued some code cleanup 17:39:23 jaypipes: can I fix multiple unrelated bugs in one commit ? 17:39:46 afazekas: yes 17:39:57 afazekas: but only if they are small... 17:40:25 jaypipes: can I do code reformatting and reorganization, in bugfix commit ? 17:40:41 afazekas: what I was saying is that it is difficult to review stuff that is *both* fixing a bug AND changing formatting or style because you prefer something some different way in the code 17:41:31 afazekas: In addition, commit messages must be more descriptive :) 17:41:38 afazekas: I think code reorganizations should be discussed before code is submitted for review if it is significant. 17:41:38 afazekas: example: https://review.openstack.org/#/c/20683/ 17:41:58 afazekas: Fixes *what* about boto intialization? 17:42:08 now all bot test are skipped 17:42:12 boto 17:42:20 Reorganization is expensive both to reviewers and in lost knowlege of where things are and how they work. 17:42:43 afazekas: what I am saying is that you need to make the commit messages more descriptive, otherwise reviewers don't really understand what the patch is attempting. 17:42:58 jaypipes: And why. 17:43:18 OK 17:44:04 May be it was obvious just for me :) 17:44:28 for the ssh related test, I suppose they will be impacted when quantum applied, since the host cannot ping vm directly in quantum. Is there any change on the way to fix this? 17:44:28 afazekas: yes, always err on the side of being TOO descriptive :) 17:45:53 chunwang: I'm sorry, I'm not catching you... 17:46:12 chunwang: what about Quantum prohibits a VM from being reachable via SSH? 17:46:22 yes 17:47:01 chumwang: no, it doesn't 17:47:11 after quantum applied, the network related execution should add network uuid, since they are in different namespaces 17:49:16 like "ping " should be "ip netnc exec ping " 17:49:32 ssh is similar... 17:50:05 hmm, I think we will need some documentation about that... 17:51:37 OK, do we have any further discussion before we wrap up? 17:52:29 OK, wrapping up... 17:52:32 #endmeeting