14:03:02 #startmeeting fedora-qadevel 14:03:02 Meeting started Mon Apr 24 14:03:02 2017 UTC. The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:03:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:03:02 The meeting name has been set to 'fedora-qadevel' 14:03:02 #meetingname fedora-qadevel 14:03:02 The meeting name has been set to 'fedora-qadevel' 14:03:02 #topic roll call 14:03:08 Let's try this again 14:03:52 * garretraziel is here 14:04:01 * mkrizek is here 14:04:05 * lbrabec is here 14:05:10 let's get this party started 14:05:16 #topic Announcements and Information 14:05:22 #info i386 repo issues should be fixed now (related to i386 being an alternate arch since F26) -- kparal 14:05:22 #link https://phab.qa.fedoraproject.org/T876 14:05:22 #info task-modularity-testing-framework should be now working fine on dev -- kparal, tflink, mkrizek, puiterwijk 14:06:19 has anyone followed up with the modularity folks to make sure everything is working on their end? 14:06:24 * roshi is here 14:07:20 kparal has been following up with them 14:08:07 ok, he added that to the meeting page so I assume that things are working well enough on that end 14:08:52 any other questions/comments? 14:08:58 none here 14:09:24 ok, moving on 14:09:32 #topic Test Invocation 14:10:35 #link https://fedoraproject.org/wiki/Changes/InvokingTestsAnsible 14:10:40 I started cleaning up the libtaskotron codebase for that today 14:10:42 #link https://phab.qa.fedoraproject.org/T939 14:10:49 cool 14:11:05 * tflink is looking forward to getting this done, it'll allow us to get rid of some bits that really need to go away :) 14:11:11 not that it'll be a trivial transition 14:11:31 but no more imagefactory, no more nested runner are all good things 14:11:47 what images are we going to use? 14:11:54 the ones produced by releng 14:11:59 cool 14:12:49 I really don't want to be building images and since we won't need libtaskotron on the images anymore, I think we can use the releng produced images now that all production releases should be getting new cloud base images nightly 14:13:05 mkrizek: which part? or just in general? 14:14:04 so we'd just boot guests on a bare metal host and ssh into those? 14:14:16 is that how we'd be getting rid of nested virt? 14:14:28 we'd still have nested virt 14:14:48 in general, looks like a lot of code will go away 14:14:51 but we'd be getting rid of the "libtaskotron spins up a vm and calls libtaskotron in that vm" part 14:15:05 ah 14:15:17 mkrizek: are you looking at getting rid of the minion code, then? 14:15:32 instead, we'd be calling ansible which would connect to that VM via ssh 14:15:42 yep 14:16:01 tflink: regarding that ^^^, I found this in the proposal "MUST execute the playbook locally with ansible_connection=local and host localhost" 14:16:21 huh, I missed that part 14:16:23 and there is "hosts: localhost" in the task examples 14:16:48 which is... strange requirement for "the testing system" 14:16:57 it's not a testing system, though 14:17:16 it's just a standard way to invoke tests with as little nailed down as we can get away with 14:17:56 * tflink is not sure why that's in there as a requirement 14:18:19 an important use-case, I understand. required, not so much 14:19:34 yeah, with that requirement we wouldn't get rid of the "libtaskotron spins up a vm and calls libtaskotron in that vm" part 14:19:56 :( 14:20:02 that part needs to DIAF 14:20:32 at the very least, move forward with the proposal to separate the executor from the minions 14:21:16 or we can hack inventory and do "[localhost] $ip_of_vm" :P 14:21:36 it seems silly to me to require an inventory file in that case 14:21:57 * tflink will follow up on that part 14:21:59 doesn't ansible-playbook need a inventory file? 14:22:13 thanks 14:22:15 yeah, but if it's always localhost, why not generate it? 14:23:03 #action tflink to follow up with ci folks on why localhost-only is a requirement for ansible-based tests 14:23:43 we need to have this ready by the end of may 14:24:04 I don't think it needs to be perfect by then but at least working 14:24:27 I also think that for the sake of our sanity, we should make this a hard switch and deprecate the task formulas at the same time 14:24:50 I'm only aware of a few people other than us who have written tasks 14:24:52 makes sense 14:25:03 +1 14:25:12 +1 14:25:22 and the translation should be relatively easy 14:26:13 * tflink realizes that he just used a bad word 14:26:31 wash your mouth out with soap 14:26:39 soap is worse 14:26:46 and then rinse with bleach to make sure you got it all :p 14:27:07 https://en.wikipedia.org/wiki/SOAP 14:27:11 * roshi almost typed soup, which doesn't really serve as a cleaner (depending on where you get it) 14:27:17 is bleach enough to kill soap? 14:27:42 in the wild, I think soap will always roam free somewhere 14:27:47 are there any questions, concerns or comments on all of this? 14:28:00 who's going to do what? 14:28:25 roshi: you're signed up to do it all 14:28:31 on top of what you're already doing 14:28:36 and I expect it by the end of the week 14:28:45 in reality ... 14:28:56 lol 14:28:59 I suspect that I'll continue porting the directives to work with ansible 14:28:59 * roshi shrugs 14:29:18 it sounds like mkrizek has started working on the other libtaskotron tweaks 14:29:45 we still need to rework our documentation, port the existing tasks and test everything 14:30:03 that's it? 14:30:05 :p 14:30:07 * roshi ducks 14:30:15 * tflink hears a volunteer 14:30:22 for ALL THE THINGS! 14:32:04 I also want to add a new entry point either in addition to or to replace runtask 14:32:18 but that's a ticket already and I think that's been discussed 14:32:52 any corrections or other concerns? 14:33:18 is 'tron' taken as a binary name? 14:33:25 no idea 14:35:31 deploying it all will be fun as well :) 14:36:06 yeah 14:36:18 roshi: any idea how close you are to the docker-from-git stuff? 14:36:42 I know the answer: docker! lots of docker! 14:36:52 that'll solve all the problems :) 14:36:54 still sussing out all the details 14:37:15 so no idea how close you are? 14:37:19 since what's in taskotron-docker doesn't match prod, I'm trying to understand all the bits 14:37:27 I hope by the end of the week 14:38:03 to make sure I've been clear: what's in taskotron-docker isn't what I'm looking for 14:38:12 right 14:38:15 is docker-from-git stuff for taskotron self-testing? 14:38:23 and local runs 14:38:25 basically, yeah 14:38:42 well, will be used for self-testing but also for local dev environments 14:38:50 are we going to get rid of taskotron-ansible-local then? 14:38:54 * mkrizek hopes so 14:39:00 not sure, it would make sense 14:39:11 * tflink hadn't gotten that far, though 14:39:13 I don't think it works anymore 14:39:26 probably doesn't, it's been a while since it was updated 14:40:15 anything else on this? 14:40:32 nothing here 14:41:17 ok, moving on, then 14:41:36 #topic ResultsDB performance/usage 14:42:07 it occurs to me that this was supposed to get started last week but some jerk forgot about the plans ... 14:42:11 * tflink glares at tflink 14:42:36 mkrizek: any thoughts on when a decent time to do it this week will be? 14:43:16 tflink: does tomorrow work for you? 14:43:43 depends on what time - I have meetings from 6-10 tomorrow 14:44:19 so, afternoon my time will work well enough 14:44:40 but that starts getting a bit late for you 14:45:33 let's sort it out after the meeting 14:45:52 sounds like a good plan 14:46:10 anything else on this topic? 14:47:45 * tflink assumes not 14:47:49 which brings us to: 14:47:52 #topic Open Floor 14:48:03 any other topics that folks want to bring up? 14:49:34 * roshi has nothing 14:51:12 * tflink sets the fuse 14:51:41 thanks for coming, everyone 14:51:45 * tflink will send out minutes shortly 14:51:49 #endmeeting