00:10:16 #startmeeting CongressTeamMeeting 00:10:17 Meeting started Thu Sep 17 00:10:16 2015 UTC and is due to finish in 60 minutes. The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:10:18 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:10:21 The meeting name has been set to 'congressteammeeting' 00:10:34 Who's here? 00:11:05 hi 00:11:17 hello time 00:11:23 hello Tim 00:11:27 how is going? 00:11:27 Hi 00:11:32 hello 00:11:34 RuiChen, Su, veena: hi 00:11:35 hi 00:11:46 hi 00:11:52 here 00:11:57 pballand, jwy, alexsyip: hey 00:12:18 I created several slides regarding our congress implementation here at Symantec 00:12:43 is there a way I can share the slides with anyone interested? 00:12:48 Su: great! 00:13:01 my pleasure 00:13:02 Su: That'd be great. 00:13:27 I don't know how you want to share the slides. They're not confidential, are they? 00:13:40 If not, maybe put them on a google drive? 00:13:42 completely not 00:13:55 good idea 00:14:44 Su: once you do that, let us all know on the ML, and we can get you feedback. 00:15:00 BTW, the slides I used at Vancouver as an intro are online... 00:15:20 https://drive.google.com/file/d/0B5VvD3PSoDPaLTVIWG1NNDhQRFE/view?usp=sharing 00:15:28 here is the URL 00:16:06 Here are the slides from Vancouver: 00:16:07 #link https://docs.google.com/file/d/0ByDz-eYOtswScTlmamlhLXpmTXc/edit 00:16:23 I already implemented the domain level quota policy. However, I did see there might be a limitation on congress right now 00:16:28 Su: let's have everyone look through the slides after the meeting and get you feedback. 00:16:41 sounds great 00:16:41 Su: what's the limitation you found? 00:16:46 can not guys see it? 00:16:54 #action Everyone will get Su feedback on slides 00:16:59 Su: yes we can see the slides. 00:17:07 the limitation is about variable aggregation 00:17:32 I saw an open ticket for that. But it seems like it is incomplete 00:17:44 Su: can you give a concrete use case: what do you want to aggregate? 00:17:52 https://review.openstack.org/#/c/140253/ 00:18:15 for example, we need to figure out how many cpus are being used for a particular domain. 00:18:30 Su: use case makes sense. 00:18:53 Su: the code was never completed; it's a pretty fundamental change, and I don't think the changeset you found is on the right track. 00:19:30 we can easily have tuples like cpu(project1, 4, domain1) cpu(project2, 3, domain1) and cpu(project3, 5, domain1). The aggregated cpu usage for this particular doamin should be 4+3+5=12 00:19:47 Got it. Aggregates have been on the wishlist for a while. 00:19:52 We need someone to implement it. 00:19:59 okay 00:20:22 i am not an expert of datalog. It will be wonderful if someone can help with it 00:20:31 shivharis needs to leave @5:30, so let's get to his agenda item. 00:20:41 Su: understood; I'll keep an eye out for someone who can help 00:20:50 shivharis: do you want to give an update on your use-case work? 00:20:52 I am currently aggregating the result manually 00:20:52 sounds good 00:21:53 another question I have is: how long does congress keep the data collected by datasource driver? 00:22:31 oops 00:22:33 (shivharis was on #congress but not seemingly on this channel. Waiting for him to enter.) 00:22:36 I remember congress does not store such data, right? but it seems like congress does store such type of data for a little while at least 00:22:48 Su: let's resume this once shivharis finishes his update 00:22:54 ok i am on, i was on the wrong channel 00:22:56 okay 00:23:34 I have started to put together a VM with devstack/openstack/congress 00:23:54 also i have set up the tutorial example on this VM 00:24:40 this was done based on last weeks AI that took for building a 4 usecase VM to be possibly installed on a tunb drive 00:24:51 thumb drive 00:25:15 this all seems to be up and running - i am open to 00:25:48 ideas for other usecases and would welcome if anyone else would like to colloborate on this 00:25:48 shivharis: cool. Have you touched base with alexsyip? 00:26:09 yes, alexsyip and I communicated 00:26:32 How is your work with usecases related to alexsyip's HOL? 00:26:40 we we doing similar work and will leverage work/ideas etc 00:26:56 Sounds good. 00:27:11 shivharis: is the VM available for us to look at? 00:27:23 we both are pretty much at the same point 00:27:23 Or: what's the best way to get you feedback? 00:27:43 i will put this vm at a public site tomorrow for anyone to download 00:28:08 shivharis: are you planning to have a doc that goes along with it that walks through the use cases? 00:28:33 yes, that will be a good idea, will do that 00:28:51 May I have our usecase included as well? 00:29:04 any UI would be necessary? 00:29:08 shivharis: This is all great! 00:29:21 Su: yes very welcome to include your usecase 00:29:24 shivharis: I think using the Horizon UI is fine. 00:29:29 the implementation is already completed 00:29:33 cool 00:29:57 Su: maybe you can help shivharis get your use case into his VM. Maybe make your VM public or give him instructions? 00:30:00 Su: ML will be a good place to start communicating, i can initiate 00:30:12 shivaharis: thanks. Note it's 5:30p. 00:30:46 i need to run today, sorry... I will catch up reading logs 00:30:55 thats all fromme 00:30:56 shivharis: thanks and sorry we started late. 00:31:17 #topic PTL elections 00:31:28 see you. Look forward to seeing the instruction docs. 00:31:42 I should have done this last week, but we're in the midst of PTL candidacy right now. 00:31:52 So if anyone wants to run you need to submit your name now. 00:32:01 I think the deadline is tonight. 00:32:11 There have been a bunch of emails on the ML. 00:32:21 If anyone wants to talk about what it entails, let me know. 00:32:29 I put myself up as a candidate. 00:32:40 Questions? 00:32:41 I am interested 00:33:26 is there a place I can submit the application? 00:33:35 or I can just send an email to someone? 00:34:17 Su: sure. I don't have the link handy, but the basic idea is you push code to the 'openstack/elections' repo. 00:34:50 okay. I will figure it out 00:34:56 The "code" you push is a textfile contains the email you send out to the ML detailing why you want to be PTL. 00:35:09 noted 00:35:27 I'll spend 1 minute looking for the link. 00:35:54 okay, thanks 00:36:15 Info on being a PTL, which was well-received by current PTLs. 00:36:16 #link http://lists.openstack.org/pipermail/openstack-dev/2015-September/073986.html 00:36:47 Info on process 00:36:48 #link http://lists.openstack.org/pipermail/openstack-dev/2015-September/074037.html 00:36:56 Any other questions? 00:37:12 yes 00:37:17 back to my previous question 00:37:28 how long does congress keep the data collected by datasource driver? 00:37:36 Su: until it gets restarted 00:38:16 so all of such data was stored in congress server memory, right? 00:38:26 I do have a couple other agenda items that we need the group for. 00:38:42 Su: let's save these clarification questions about how congress works til the open-discussion at the end. 00:38:48 Su: sound good? 00:38:55 no problme 00:38:58 Su: thanks. 00:39:06 #topic Gate problems 00:39:10 you are welcome 00:39:31 Right now we're having trouble getting code in b/c there are tempest tests that are failing. 00:39:43 The tests aren't ours; they are glance tests. 00:39:56 I've been unable to replicate the failures locally. 00:40:17 There are 2 I've noticed failing, each of which are testing glance's notification capabilities. 00:40:39 So I tried to change the tempest tests running in our gate to only be those relevant to congress. 00:40:47 https://review.openstack.org/#/c/223855/ 00:41:03 But we're getting pushback in that no one else seems to be having these problems. 00:41:37 thinrichs: looks like they are timeout issues right - fixtures timeout 00:41:59 So I'm trying to work with the infra team (ianw who wrote the review) to figure out what's happening. 00:42:13 ramineni: agreed. 00:42:23 thinrichs: recheck solving them 00:42:38 ramineni: when I ran the tests on my own machine they always finish in < 3 seconds. 00:42:48 In the gate, they're running for 500 seconds and timing out. 00:43:01 thinrichs: ya, 00:43:06 ramineni: yesterday they seemed to be failing > 50% of the time. 00:43:30 The worry is that there's something that conflicts with glance in some way. 00:43:31 thinrichs: ya, its pain to do multiple rechecks everytime .. 00:43:35 Do we have any code to receive the notifications? 00:43:51 zhenzan: no. 00:44:01 thinrichs: so i don't know what's going on, but it doesn't seem to be something that affects everyone from poking at logstash. that's why i think it's something specific. we can discuss in another channel 00:44:21 ianw: thanks for chiming in. 00:44:26 what about ceilometer? Do we enable ceilometer in gate? Is it using the latest code? 00:44:55 ianw: makes perfect sense. I tried to ping you a couple of times this morning (for me) and clark told me you had a different notion of morning from AU. 00:45:04 ianw: maybe tomorrow we can sync up? 00:45:11 I talk with ceilometer guys, they don't face the same issue in their CI. 00:45:39 It'd be great if we could get some eyeballs on this. 00:45:42 Any volunteers? 00:45:58 I remembered that I saw this issue before, together with Lianhao Lu, a ceilometer core 00:46:07 but several months ago 00:46:44 yes, if Lianhao can help to check, it's great 00:47:21 it's around June 24 00:48:38 anyway, I can also take a look today 00:48:39 Ian pointed me to a helpful tool that lets us search over all logs in the gate 00:48:41 #link http://logstash.openstack.org/ 00:48:46 zhenzan: that'd be great. 00:49:08 #topic Tempest plugin 00:49:22 zhenzan: you've been working on a tempest plugin. Want to report? 00:49:28 yes 00:49:32 (And does that help with this issue in the gate?) 00:49:42 just pushed patch set 3 00:50:06 the patch set 2 has an issue to conflict with local unit tests 00:50:29 #link zhenzen's patch https://review.openstack.org/#/c/223933/ 00:50:48 the code is based on Manila's experience 00:51:02 but, the test cases are still the same 00:51:26 so we still have to skip some test cases 00:52:17 with the tempest plugin framework, we don't need to seek for merging our tempest code into the main tempest tree 00:52:25 skipping test cases is fine. 00:52:33 and it's easy to run just congress tests 00:52:41 Does this plugin help us with our gate problem? 00:53:33 with our failure cases skipped, I think the answer is yes. 00:53:56 but the effect is just like what you do in https://review.openstack.org/#/c/223855/ 00:54:31 if we still run all tempest cases, we may still meet the glance failure 00:54:36 Ok. It's good to be using the proper plugin framework in any case. 00:54:57 That's my last agenda item. 00:55:15 Anyone have anything they need to discuss with the group? 00:55:18 (5 minutes left) 00:55:34 thinrichs: have one quick question 00:55:57 ramineni 00:56:03 ramineni: shoot 00:56:19 i have raised couple of patches seeing python-3 gate failures -- -https://review.openstack.org/#/c/223406/ 00:56:34 but im seeing error like this - gate-congress-requirements http://logs.openstack.org/06/223406/1/check/gate-congress-requirements/9f98a1c/ : Incompatible requirement found; 00:56:57 i can see mox3 is part of global-requirements.. is there anything else needs to to? 00:57:55 I think you need to use the same versions as global-requirements. 00:57:56 Maybe 00:58:13 thinrichs: thanks, will try that 00:58:17 So if they have mox3>=0.7.0, maybe you need to use that too 00:58:22 Just a guess. 00:58:31 Su: do you want to ask your questions now? 00:58:38 * ramineni gives it a try 00:58:41 Or we could move over to #congress and do that, since we have 2 minutes left. 00:59:29 Apologies once again for the shortened meeting. 00:59:54 I'll be around for another 10-15 minutes in #congress if anyone wants to chat. 01:00:05 Bye 01:00:13 bye 01:00:18 bye 01:00:19 bye~ 01:00:23 #endmeeting