18:07:16 #startmeeting cue 18:07:17 Meeting started Tue Apr 14 18:07:16 2015 UTC and is due to finish in 60 minutes. The chair is sputnik13. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:07:18 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:07:21 The meeting name has been set to 'cue' 18:07:31 :) 18:07:37 there, now it's going all under cue 18:08:07 ok, our second meeting 18:08:27 Here the logs from last meeting which contains some action items: http://eavesdrop.openstack.org/meetings/cue/2015/cue.2015-04-07-18.02.html 18:08:30 my action item from previous meeting was: davideagnello to discuss cue's integration testing approach 18:09:02 yes, davideagnello please talk about what happened and where we are with your action 18:09:21 I am currently writing some sample tests with the Rally framework as 'plugins' 18:09:37 #topic integration testing approach 18:09:40 since the scenario or integration tests would be located in Cue 18:10:37 the second action item is to write a sample scenario test in Tempest, after which a decision would be made 18:11:08 does tempest allow you to have tests located outside the tempest repository as well? I think you mentioned but it slips my feeble 18:11:10 memory 18:11:14 :) 18:11:19 scenario testing approach would be simulating use-cases, including negative 18:11:44 davideagnello: it would be leveraging the cue client for this right? 18:12:03 esmute: yes 18:12:19 integration tests would ideally hit the API directly 18:13:15 sputnik13: they can, it requires some configuration to get them running which is in our action items to get running. 18:13:38 ok 18:13:49 davideagnello: I am about to submit a patch to refactor and separate the unittests from functional. Once Im done this, i'll help you with your investigation with tempest 18:14:03 with rally, in order to get tests running outside the repo they have to be copied to the appropriate location, namely: ./rally/plugins/ 18:14:21 would that be cue/rally/plugins? 18:14:37 rally looks in this location for your plugin scenarios, contexts, runners, etc. 18:15:24 esmute: thank you, what is remaining with tempest is to write some sample tests 18:16:27 sputnik13: the test code and yaml/json files would live in our repo. when running the tests, they would have to be moved to ~./rally/... for rally to actually execute them 18:17:19 symlink would work just as well I assume, like how we have the devstack plugin symlinked from our repo to devstack 18:18:20 sputnik13: it should, will try running my sample tests like this as well 18:18:37 so davideagnello you're going to do the rally tests, esmute is going to do the tempest tests, and we'll have more information on which to go with by next week, right? 18:19:00 sputnik13: yes, targeting by the end of this week 18:19:37 #action davideagnello to implement simple rally tests and provide findings on suitability for use in writing integration and scenario tests 18:19:56 #action esmute to implement simple tempest tests and provide findings on suitability for use in writing integration and scenario tests 18:20:07 did I do that right? :) 18:20:42 hmm, no feedback from the bot, let's just say that's correct and move on 18:21:07 #topic end-to-end conductor functional test 18:22:37 I think everyone knows, but for the sake of posterity... from looking at the test "coverage" as far as an end-to-end test, it was agreed that we already have a test that provides sufficient coverage from the API to the jobboard 18:23:09 so, there is already a functional test that directly exercises the API's POST method to initiate a create cluster flow 18:23:39 which results in a create cluster job being posted to the taskflow jobboard 18:24:26 what we're missing is a test to ensure when we spin up a conductor that the conductor is started properly and starts pulling jobs off the jobboard 18:24:42 so the latter is the functional test that I am currently working on 18:25:13 for a single unified test that does both I believe we need to treat it more like an integration or scenario test, which will be done once we decide on which framework to use 18:25:50 oh, shoot, I should have asked whether there are any questions before changing topics... sorry :) 18:26:02 sputnik13: would it be possible to use all the necessary fixtures to do an end-2-end test? 18:26:16 esmute what do you mean? 18:27:46 you mean use fixtures to do a single complete end to end test? 18:27:46 esmute: the issue is that an end-to-end test will require a separate process for the worker/conductor to take jobs and run them 18:27:52 i was wondering if we could levarage the fixtures (or create more) in order to have tests that will start from the api, post a job to the jobboard and get it consumed 18:28:54 could the same process do both? I remember us talking about it but it is also slipping my feeble :P 18:29:01 esmute: yes we can do that, but I think these things should be tested separately at least as far as unit/functional tests are concerned... the thing with test cases is the more they cover the less information you generally have as far as exactly what failed when failures happen 18:29:04 if we are going to have integration tests that will do that, there won't be much value added 18:29:58 so the idea is to have functional tests that does the posting to the jobboard and other tests that test the consumption 18:29:59 technically speaking, we could do all of our integration and scenario tests with fixtures, it really depends on what we want to accomplish 18:30:25 esmute: right, since those are in fact two separate "functions" in cue, one is by the API and the other is by the worker 18:30:52 at least that's what I'm telling the voices in my head to justify my thought process :D 18:31:59 ok.. thats fine.. we will need integration tests anyways and it is makes more sense to do them with int-tests then it is fine 18:32:31 right, as you said we would need to do tests like that in integration/scenario tests anyway 18:32:38 wassup folks 18:33:22 so, rather than do it in both functional and integration, I think it makes sense to have the smaller tests in functional that verify each leg of that particular "flow" and have the integration test verify the whole flow together 18:33:48 ok.. sounds good 18:34:11 vipul welcome fearless leader 18:34:29 Wozniak just got off stage.. he like to talk 18:34:37 how are things at openstack live :) 18:34:43 the woz 18:35:26 lol 18:35:30 we're presenting in a couple of hours.. the Cue logo will be prominently displayed 18:35:48 niiice 18:36:09 if there are no more questions/issues, this segways nicely in to the next topic :) 18:37:46 no more questions or issues? 18:37:54 for this or the previous topic (sorry)? 18:37:54 nop.. next topic 18:37:57 ok 18:38:07 #topic cue stickers 18:38:42 so just to give some quick feedback.. 18:38:51 from the folks that notice these things.. 18:38:51 ok 18:39:02 the shadows are kinda off in the design you sent last 18:39:19 ah right 18:39:24 the top of the lower bar should be lighter 18:39:26 not darker 18:39:28 :) 18:39:31 i just realized 18:39:34 so if the light source is at to top which is where it seems to be 18:39:37 right 18:39:44 I will let him know 18:39:51 do we have a final color selected? 18:40:17 I don't think so... vipul mentioned he would like it a darker red, pink is offensive to his sensibilites :) 18:40:22 :P 18:40:38 a richer red would be nice.. just my opinion.. 18:40:39 :D 18:40:40 http://snag.gy/nEquR.jpg 18:40:47 that is how the logo currently stands 18:40:50 vipul: im impressed you saw that level of detail 18:41:04 esmute: i didn't.. others did 18:41:37 vipul: any other feedback? 18:41:38 others? 18:41:44 from OS live? 18:41:55 Yes.. while i was creating slides 18:42:17 sputnik13: that's the only feedback.. otherwise people like it 18:42:37 cool 18:42:42 I passed on the feedback 18:43:02 ok we only have about a month.. so need to get these printed out soon 18:43:23 and a new version with the tweaks we asked for should be coming our way today or tomorrow 18:43:24 sputnik13: who is making these adjusting to the logo? 18:43:40 s/adjusting/adjustment 18:44:24 esmute: a guy I worked with at my previous company is a graphic designer by training (although he does web ui/ux now), we asked him to do the logo 18:45:42 so once we OK the tweaks, he's going to get a printed proof done to double check the cutouts, and then have the final printed and sent to us 18:45:53 ok 18:45:56 cool beans 18:46:12 we talked briefly about t-shirts, do you want to get something done there? 18:46:42 I think we just get the logo from him and have Angela take care of those 18:46:53 ok 18:47:24 yeah I wasn't thinking of asking him to do the t-shirts, I was more asking are we going to make t-shirts for the team :) 18:47:45 oh yea i think we should.. and make esmute wear it 18:47:52 as you said we have 1 month now, so we should probably get it going soon 18:48:22 i would wear a shirt with that logo... just that logo though... no other odd additions 18:48:24 :p 18:48:40 hah, I have some thoughts on what the t-shirt should look like 18:48:54 I'll get the t-shirt stuff moving 18:49:12 it would be really good to get more visibility at the conference 18:49:26 #action sputnik13 getting the cue t-shirt finalized 18:49:38 :) 18:49:54 #action esmute wear t-shirt for entire conference 18:50:08 oops.. i forgot to add no weird toons/geeky stuff 18:50:14 hah, too late 18:50:29 ok that's it for previous actions 18:50:37 open discussion 18:50:44 ok i got one 18:50:45 oh wait, any questions/issues? 18:50:49 with previous topic 18:51:00 if not, lets go to open discussion 18:51:06 PEPE8 import grouping checkers. Do we need them? 18:51:10 they are so annoying. 18:51:16 #topic open discussion 18:51:23 PEPE8 import grouping checkers. Do we need them? 18:51:34 is it the import grouping? 18:51:45 i dont understand that order pep8 wants it 18:51:47 I don't think we have any pep8 defined ourselves 18:52:04 so i keep swiching things around until i get it right 18:52:07 meaning we get it from the openstack common requirements... or something like that 18:52:11 afaik 18:52:25 so, yes we need to keep it... vipul any thoughts? 18:52:46 we can omit some pep8 validations if we want 18:53:13 these are annoying.. or am i the only one that thinks that? 18:53:19 import grouping isn't that bad 18:53:37 it's system libs, 3rd party libs, then project libs 18:53:39 s/libs/modules 18:54:07 so anything that's standard python module, you put at top in alphabetical order 18:54:18 ok 18:54:28 then anything that's a 3rd party module (likely installed via pip or requirements.txt) 18:54:42 then any cue modules 18:54:43 so 3rd party ares things like oslos and other openstack commons libs 18:54:46 yeah 18:55:08 I sometimes get confused whether it's stdlib, 3rd party, project 18:55:13 or stdlib, project, 3rd party 18:55:21 haha 18:55:22 ok 18:55:25 but that's the general grouping 18:55:33 and they need to be in alphabetical order 18:55:39 if its just me, ill learn the grouping order.. its fine 18:55:56 yeah it's annoying but once you figure out the ordering it's not that bad 18:56:09 and there's some merit to it 18:56:27 when you want to see what modules are used, it's easier to find things in a sorted list 18:57:37 ok, other topics? 18:57:39 it's lunch time :D 18:58:37 alright, should we adjourn the meeting? 18:58:38 cool.. we took the whole hour as well 18:59:16 alright, thanks everyone 18:59:20 #endmeeting