Friday, 2017-03-03

*** thrash is now known as thrash|g0ne00:38
*** jamielennox|away is now known as jamielennox00:43
*** Kevin_Zheng_ has joined #openstack-mistral01:20
*** Kevin_Zheng_ is now known as Kevin_Zheng01:21
*** gongysh has joined #openstack-mistral01:27
*** jamielennox is now known as jamielennox|away01:29
*** jamielennox|away is now known as jamielennox01:36
*** gongysh has quit IRC02:22
openstackgerritEmilien Macchi proposed openstack/mistral master: CI test - never merge  https://review.openstack.org/44089802:24
*** catintheroof has joined #openstack-mistral02:30
*** zhurong has joined #openstack-mistral03:22
*** zhurong has quit IRC03:24
*** fultonj has quit IRC03:30
*** catintheroof has quit IRC05:00
*** zhurong has joined #openstack-mistral05:59
*** jaosorior has joined #openstack-mistral06:44
*** jrist has quit IRC07:16
*** jrist has joined #openstack-mistral07:16
*** zhurong has quit IRC07:26
*** zhurong has joined #openstack-mistral08:04
d0ugalrakhmerov: the TripleO check was removed?08:10
d0ugalIt is here in the "gate", but not in the "check" https://review.openstack.org/#/c/434230/08:10
rakhmerovd0ugal: hi, I'm actually not aware of this08:25
rakhmerovhave no idea why it happened08:25
d0ugalok, I'll look into it!08:25
d0ugalI think before it wasn't on the gate and was check only?08:25
d0ugalMaybe I don't remember correctly but I'll see what I can find out08:25
rakhmerovnope, it was in both08:26
d0ugalrakhmerov: I guess EmilienM might know, he was testing something ~6 hours ago. https://review.openstack.org/44089808:37
d0ugalso I'll check with him when he is online later.08:37
rakhmerovok, thanks08:37
*** shardy has joined #openstack-mistral08:51
*** jpich has joined #openstack-mistral08:54
*** jrist has quit IRC09:14
*** jrist has joined #openstack-mistral09:14
d0ugalrakhmerov: wait until you see the patch I am going to push up soon :)09:20
d0ugalrakhmerov: I have reduced the mistral test suite from ~220 seconds to ~80 seconds in total09:21
rakhmerovWoooow!09:22
rakhmerovreally09:22
rakhmerovI just +2ed one relatively small change09:22
rakhmerovlooking forward09:24
openstackgerritSpencer Yu proposed openstack/mistral master: Add idempotent_id decorator to tempest testcases  https://review.openstack.org/44101709:24
d0ugalrakhmerov: yeah, so I found out that every unit test is loading and registerting all ~1300 openstack actions. That is a huge number of database inserts when they are only actually tested in one unit test...09:32
d0ugalI'm just waiting for https://review.openstack.org/#/c/434230/ to merge09:32
rakhmerov?????09:32
rakhmerovwhat??????09:32
d0ugalbecause I think if I push the patch now it will restart the CI run09:33
rakhmerovevery unit test is loading all actions???09:33
d0ugalrakhmerov: yup09:33
d0ugallol09:33
rakhmerovoh my....09:33
rakhmerovhow could we overlook it? :)09:33
d0ugalrakhmerov: so it will be solved even more when we move them to mistral-extra, but for now I am just using the new test_mapping.json file in all the tests.09:33
rakhmerovooh, yes09:34
d0ugalhttps://review.openstack.org/#/c/434230/4/mistral/tests/resources/openstack/test_mapping.json - this file (but I needed to add a couple more)09:34
rakhmerovyeah, I was just sitting and thinking that once we move openstack actions into mistral-extra it will be much easier to support and test them09:34
*** zhurong has quit IRC09:34
d0ugalyup09:34
rakhmerovyeah, so cool!!09:35
rakhmerovthat's awesome Dougal09:35
rakhmerovI'm glad you found it :)09:35
rakhmerovseems like https://review.openstack.org/#/c/434230/ should be merged soon09:36
d0ugalyeah09:36
d0ugalI am watching it in zuul09:36
rakhmerovd0ugal: we need to have a planning meeting in IRC after the PTG09:37
d0ugalyup, sounds good.09:37
d0ugalCan we just do that in the regular meeting slot?09:37
rakhmerovI'm going to sort out all my notes and update all LP tickets, after that I'll write an email and we can set up a meeting09:37
rakhmerovyes, we can09:37
rakhmerovI thought we might need more time though09:37
d0ugalyeah, probably09:39
d0ugalrakhmerov: Somebody wants to start using the ansible actions I wrote in TripleO. Do you think they could live in mistral-extra too? or is that only going to be for openstack actions?09:40
d0ugalWe could just move them into a TripleO repo - but given they are generic it would be good to let others have access.09:41
rakhmerovwell, IMO they could live in mistral-extra, why not09:41
d0ugalokay, cool09:42
d0ugalso over time mistral-extra could become a large library of actions.09:42
rakhmerovlet's talk about it on a weekly meeting09:42
d0ugalyup, sounds good09:42
rakhmerovyes, you're right09:43
rakhmerovthat's the plan09:43
rakhmerovI really would like to get these actions into order09:43
rakhmerovthe project would benefit a lot from this09:43
d0ugalAgreed.09:43
rakhmerovespecially, I'd love to find a good way to test openstack actions09:44
d0ugalyeah, that will be tricky but we should try09:44
d0ugalWe test some via tempest now I think?09:44
rakhmerovwhat I'm thinking about is to create sort of like a small framework that would help: 1) easily create new action packs 2) deal with versions of services 3) easily install needed action packs09:45
rakhmerovd0ugal: right, tempest09:45
d0ugalso actually this file will need to move to mistral-extra too eventually? https://github.com/openstack/mistral/blob/master/mistral_tempest_tests/tests/scenario/engine/actions/v2/test_openstack_actions.py09:45
d0ugalyeah, that is a good idea.09:45
d0ugalI shall have to think about this :)09:45
rakhmerovwe talked to thrash|g0ne at the PTG, he said it's possible to create CI jobs that run on a cron pattern. So I thought we could create a good test harness for OpenStack actions and run it not every commit but say once a day09:46
rakhmerovso, there are options09:46
rakhmerovwe'll be moving step-by-step to that09:46
d0ugalyeah, good idea09:54
rakhmerovd0ugal: I'm looking at https://review.openstack.org/#/c/440646/3 now09:57
rakhmerovhow does the verification lead to reduced test duration?09:57
openstackgerritMerged openstack/mistral-lib master: Updated from global requirements  https://review.openstack.org/44038309:57
d0ugalrakhmerov: I don't understand :)09:57
rakhmerov:)09:58
openstackgerritMerged openstack/mistral-dashboard master: Updated from global requirements  https://review.openstack.org/44005909:58
rakhmerovhow come?09:58
*** zhurong has joined #openstack-mistral09:58
d0ugalrakhmerov: but I can show you...09:59
d0ugalrakhmerov: in a different test run, you can see it is here and very slow; http://logs.openstack.org/30/434230/4/check/gate-mistral-python35/30dce76/console.html#_2017-03-03_07_22_30_99818709:59
d0ugalrakhmerov: with that patch it is no longer in the slowest tests: http://logs.openstack.org/46/440646/3/check/gate-mistral-python35/67bd2d1/console.html#_2017-03-02_16_48_22_55449509:59
d0ugalrakhmerov: I took the extra code from the test above because I noticed it was very similar, but much much faster.10:00
d0ugalThis test: https://github.com/openstack/mistral/blob/master/mistral/tests/unit/engine/test_policies.py#L487-L50310:00
d0ugalrakhmerov: My guess is that if a workflow is running, when the cleanup that stops the engine happens it is really slow at stopping if there is a workflow still running - does it maybe wait for it to finish?10:01
d0ugalbut really I don't understand :(10:01
rakhmerovhm...10:01
rakhmerovinteresting10:01
rakhmerovI want to understand why it happens10:02
d0ugalrakhmerov: yeah, I was planning to ask you about it10:02
rakhmerovok, I'll look at the cleanup code10:02
rakhmerovok10:02
d0ugalrakhmerov: when I was investigating, I commented out these two lines: https://github.com/openstack/mistral/blob/master/mistral/tests/unit/engine/base.py#L81-L8210:05
d0ugalrakhmerov: that made the test much faster too (but it caused lots of other tests to fail)10:06
rakhmerovyeah, of course10:06
rakhmerovbut this code has to run10:06
rakhmerovotherwise tests won't be isolated from each other, they may reuse the same infrastructure (engine, executor)10:06
d0ugalyup, I understand10:07
rakhmerovthat's why we run it with "True" also, to make sure they fully stopped10:07
rakhmerovok10:07
d0ugalbut it suggests that is the slow part.10:07
d0ugalbut I don't know why it is so slow, only for this test10:07
rakhmerovhm..10:11
rakhmerovok, getting back to the test I'm reviewing (additional checks). I'm not sure on 100% but seems like you're right. If we don't wait till wf is completed then when we try to kill engine and executor service there's an extra waiting loop somewhere10:13
rakhmerovmy guess is that it's in either oslo.service or in oslo.messaging10:14
rakhmerovdidn't find what else can do that, seems like we don't do anything like this ourselves explicitly10:14
rakhmerovI think it's essentially a "for" loop somewhere with a huge sleep()10:14
d0ugalI'll take a look in oslo10:14
d0ugalyeah, that is my guess10:14
d0ugaland for shutdown that is probably okay - usually stop doesn't need to be fast10:15
rakhmerovok, I'm going to give +2 now although I don't fully understand the reason10:15
d0ugalI'll do some more investigation10:15
openstackgerritDougal Matthews proposed openstack/mistral master: Updated from global requirements  https://review.openstack.org/44005810:17
d0ugal^ I just done that to kill the zull job. the tripleo CI run had failed.10:18
d0ugaland it was blocking 434230 merging as it was ahead in the queue10:19
d0ugalbut it seems doing that caused 434230 to start again :(10:19
d0ugalso I guess I should have waited for the timeout10:19
openstackgerritDougal Matthews proposed openstack/mistral master: Reduce the number of with-items and retried in the concurrency test  https://review.openstack.org/44063310:20
openstackgerritDougal Matthews proposed openstack/mistral master: External OpenStack action mapping file support  https://review.openstack.org/43423010:20
openstackgerritDougal Matthews proposed openstack/mistral master: Verify the retry policy when passed in via variables  https://review.openstack.org/44064610:20
openstackgerritDougal Matthews proposed openstack/mistral master: Remove the delay from the direct workflow rerun tests  https://review.openstack.org/44062510:20
openstackgerritDougal Matthews proposed openstack/mistral master: Every unit test creates and registers every OpenStack action  https://review.openstack.org/44103810:20
d0ugalrakhmerov: ^ that last one should speed things up the most.10:20
rakhmerovok10:21
rakhmerovthat's ok, no worries10:21
d0ugalWhat is Istvan's IRC nick?10:22
*** zhurong has quit IRC10:25
*** gongysh has joined #openstack-mistral10:34
*** gongysh has quit IRC10:37
d0ugalrakhmerov: http://logs.openstack.org/38/441038/1/check/gate-mistral-python35/5bf1c47/console.html#_2017-03-03_10_33_49_53558410:38
d0ugalstill some work to do, but getting there :)10:38
openstackgerritMerged openstack/python-mistralclient master: Updated from global requirements  https://review.openstack.org/44009910:39
rakhmerovok10:41
rakhmerovmgershen, d0ugal: pls review https://review.openstack.org/#/c/437058/10:43
d0ugaldone, I think there is an issue in the coveragerc10:46
*** jaosorior is now known as jaosorior_lunch11:01
openstackgerritRenat Akhmerov proposed openstack/mistral-specs master: Update folder structure  https://review.openstack.org/44107111:33
*** jkilpatr has quit IRC11:38
*** thrash|g0ne is now known as thrash11:41
*** jaosorior_lunch is now known as jaosorior11:41
openstackgerritMerged openstack/mistral master: Update docs for tasks function  https://review.openstack.org/43309611:49
openstackgerritDougal Matthews proposed openstack/mistral master: Every unit test creates and registers every OpenStack action  https://review.openstack.org/44103811:52
openstackgerritRenat Akhmerov proposed openstack/mistral-specs master: Update folder structure  https://review.openstack.org/44107111:53
openstackgerritRenat Akhmerov proposed openstack/mistral-specs master: Update folder structure  https://review.openstack.org/44107111:54
thrashd0ugal: just rebased on top of your changes... *big* improvement.11:56
d0ugalthrash: whoop :)11:56
d0ugalthrash: Have you fized all the timezones yet? ;)11:57
thrashd0ugal: not yet...11:57
thrashd0ugal: I've got 1 test failing atm11:57
d0ugalthrash: k, I'm just about to take the dog out but let me know if I can help this afternoon11:58
thrashd0ugal: gonna push up what I have atm11:58
d0ugalcool11:58
d0ugalI'll start by taking a look at that when I get back then11:58
thrashd0ugal: awesome11:59
*** jkilpatr has joined #openstack-mistral11:59
openstackgerritBrad P. Crochet proposed openstack/mistral master: Use utcnow() in expired executions policy test  https://review.openstack.org/44108212:02
openstackgerritBrad P. Crochet proposed openstack/mistral master: Fix get_next_execution_time  https://review.openstack.org/44108312:02
*** zhurong has joined #openstack-mistral12:05
*** jkilpatr has quit IRC12:07
openstackgerritRenat Akhmerov proposed openstack/mistral-specs master: Add Pike specs into index  https://review.openstack.org/44108912:10
*** jkilpatr has joined #openstack-mistral12:21
openstackgerritMerged openstack/mistral master: Remove output from list action executions API  https://review.openstack.org/43383112:24
*** shardy is now known as shardy_lunch12:25
openstackgerritBrad P. Crochet proposed openstack/mistral master: Fix get_next_execution_time  https://review.openstack.org/44108312:25
openstackgerritBrad P. Crochet proposed openstack/mistral master: Use utcnow() in expired executions policy test  https://review.openstack.org/44108212:25
openstackgerritRenat Akhmerov proposed openstack/mistral master: Rename package 'workbook' to 'lang'  https://review.openstack.org/44109312:27
openstackgerritBrad P. Crochet proposed openstack/mistral-extra master: Set up mistral-extra with the ability to run unit tests  https://review.openstack.org/43705812:27
*** zhurong has quit IRC12:33
*** zhurong has joined #openstack-mistral12:41
EmilienMd0ugal, rakhmerov: hey12:48
EmilienMI did some random testing just to check zuul layout12:49
EmilienMnothing bad in tripleo ci etc12:49
EmilienMis there any issue ?12:49
rakhmerovhi12:49
rakhmerovI think it's good now12:49
rakhmeroveverything seems to work12:49
openstackgerritJeremy Liu proposed openstack/mistral master: Add idempotent_id decorator to tempest testcases  https://review.openstack.org/44022112:51
openstackgerritJeremy Liu proposed openstack/mistral master: Add idempotent_id decorator to tempest testcases  https://review.openstack.org/44022112:54
thrashrakhmerov: question for you... It has to do with one shot triggers...12:59
thrashrakhmerov: the value passed via the client has no TZ information associated with it.12:59
rakhmerovthrash: ok12:59
thrashrakhmerov: triggers.create_cron_trigger now makes a conversion to UTC.13:00
rakhmerovyeah, I see where you're going to..13:00
rakhmerovso, IMO this info should come from the client13:00
rakhmerovright?13:01
rbrady-afk /nick rbrady13:01
*** rbrady-afk is now known as rbrady13:02
rakhmerovotherwise not sure how we can convert it correctly13:02
thrashrakhmerov: I'm thinking yes... but... It would mean perhaps breaking cli/api backcompat13:02
thrashrakhmerov: yeah... it should definitely come from the client, the more I think about it.13:02
rakhmerovyeah13:03
thrashthe client should always send utc actually...13:03
thrashso, the server assumes it's utc and does nothing with it.13:03
rakhmerovthrash: so, I would say backwards compatibility is aside here because we obviously have a bug13:03
thrashrakhmerov: I think that eliminates the backcompat problem.13:03
rakhmerovand I believe that very very few people use these one shot triggers13:03
thrashadd a '--utc' flag to the client.13:04
thrashif --utc, it means that the datetime typed in is already utc, and the client should not convert prior to sending.13:04
thrashif no '--utc', client should convert to utc before sending.13:04
rakhmerovhm..13:04
rakhmerovgood idea13:05
rakhmerovyep, looks ok with me13:05
*** dprince has joined #openstack-mistral13:05
rakhmerovd0ugal: what do you think?13:05
rakhmerovthrash: so, I like what you're proposing. I guess it'd good to have input from d0ugal since, as I remember, he was already looking at a similar problem13:07
thrashrakhmerov: +1.13:08
rakhmerovthrash: you'd like to work on it yourself?13:09
thrashrakhmerov: the --utc flag would also apply to start time13:09
thrashrakhmerov: I would. :)13:09
rakhmerovok, np13:09
thrashI'm already very intimate with that part of the code. :)13:09
thrashand I've got plenty of experience with client code too.13:09
rakhmerovyeah, seems like it needs to be applied for all dates we're passing from client13:09
rakhmerovok, great13:10
*** fultonj has joined #openstack-mistral13:12
*** chlong__ has quit IRC13:14
d0ugalEmilienM: the multinode job didn't run for a couple of mistral patches - that's why I pinged but it seems back now?13:18
EmilienMd0ugal: which one?13:19
EmilienMit should run13:19
d0ugalthrash, rakhmerov - sounds good.13:19
d0ugalEmilienM: hmm, let me see if I can find it again :)13:19
EmilienMd0ugal: take 440221 for ex, it runs13:19
rakhmerovd0ugal: ok13:20
d0ugalEmilienM: patchset 4 here: https://review.openstack.org/#/c/434230/13:20
d0ugalEmilienM: but actually, I see now that was several days ago. I thought it was new13:20
d0ugalso maybe just ignore me and I need to drink coffee earlier13:21
EmilienMd0ugal: yes I fixed it yesterday13:21
EmilienMd0ugal: gimme a sec13:21
EmilienMd0ugal: https://review.openstack.org/#/c/440668/13:22
d0ugalEmilienM: aha, thanks. You are way ahead of me then :)13:22
d0ugalrakhmerov: I had given up with the timezone bug, so I am glad thrash is taking a look :)13:25
d0ugalthrash: if somebody is already passing UTC, but they don't add the --utc flag, wont that be a problem?13:26
thrashd0ugal: I would presume so.13:28
thrashd0ugal: actually, it probably wouldn't be...13:28
thrashd0ugal: right now, they would see a problem if they *did* pass utc, as it would just swing the time further back from their local time.13:29
thrashthe server side is doing a conversion that it shouldn't be.13:29
d0ugalthrash: right, that makes sense13:29
*** catintheroof has joined #openstack-mistral13:31
rakhmerovyeah13:32
rakhmerovsounds right13:32
openstackgerritMerged openstack/mistral master: Updated from global requirements  https://review.openstack.org/44005813:37
openstackgerritMerged openstack/mistral master: External OpenStack action mapping file support  https://review.openstack.org/43423013:37
*** catintheroof has quit IRC13:39
*** shardy_lunch is now known as shardy13:39
*** aspiers has quit IRC13:41
d0ugal^ yay, finally merged.13:43
*** zhurong has quit IRC13:48
*** dprince has quit IRC14:05
openstackgerritRyan Brady proposed openstack/mistral-lib master: Adds minimum common shared code for custom actions API  https://review.openstack.org/41141214:12
*** zhurong has joined #openstack-mistral14:15
*** chlong__ has joined #openstack-mistral14:22
*** dprince has joined #openstack-mistral14:33
*** zhurong has quit IRC14:33
*** jaosorior has quit IRC15:06
d0ugalrbrady: do you want me to try working on a Mistral patch to make it use mistral-lib?15:07
d0ugalthrash: with your patches all the tests pass for me15:07
d0ugalamazing!15:07
*** aspiers has joined #openstack-mistral15:14
openstackgerritBrad P. Crochet proposed openstack/mistral master: Fix get_next_execution_time  https://review.openstack.org/44108315:48
*** rbrady is now known as rbrady-afk16:18
*** jkilpatr_ has joined #openstack-mistral16:28
*** jkilpatr has quit IRC16:31
thrashd0ugal: woot!16:51
*** jpich has quit IRC17:24
*** toure_ has joined #openstack-mistral17:40
*** zhenguo has quit IRC17:43
*** toure has quit IRC17:43
*** shardy has quit IRC17:44
*** chlong__ has quit IRC18:04
*** chlong__ has joined #openstack-mistral18:20
*** weshay has quit IRC18:52
*** jkilpatr has joined #openstack-mistral18:56
*** rbrady-afk is now known as rbrady18:59
rbradyd0ugal: sure.  thanks18:59
*** jkilpatr_ has quit IRC19:00
*** weshay has joined #openstack-mistral19:01
openstackgerritBrad P. Crochet proposed openstack/python-mistralclient master: Add --utc flag to cron trigger create  https://review.openstack.org/44135019:16
*** zhenguo has joined #openstack-mistral19:33
*** chlong__ has quit IRC19:55
*** chlong__ has joined #openstack-mistral20:09
*** dprince has quit IRC20:50
*** chlong__ has quit IRC20:57
*** jamielennox is now known as jamielennox|away21:29
openstackgerritBrad P. Crochet proposed openstack/mistral-extra master: Set up mistral-extra with the ability to run unit tests  https://review.openstack.org/43705821:29
*** toure_ is now known as toure21:44
*** catintheroof has joined #openstack-mistral22:29
*** thrash is now known as thrash|g0ne22:47
*** jkilpatr has quit IRC23:08
*** jkilpatr has joined #openstack-mistral23:41
*** gongysh has joined #openstack-mistral23:57

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!