14:02:03 #startmeeting fedora-qadevel 14:02:03 Meeting started Mon Jul 25 14:02:03 2016 UTC. The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:03 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:02:03 The meeting name has been set to 'fedora-qadevel' 14:02:10 #topic Roll Call 14:02:15 * mkrizek is here 14:02:39 * kparal is here 14:03:18 kparal: aren't you on PTO this week? 14:03:39 * garretraziel is here 14:03:46 * kparal is a double agent 14:03:54 . 14:03:57 * jskladan is here 14:04:06 tflink: I changed my PTO to Wed->Fri 14:04:20 ah, ok 14:05:24 well, let's get this party started 14:05:43 #topic Announcements and Information 14:05:57 #info new oauth2 based auth method has been deployed w/ recent database snapshot to qa.stg - tflink 14:06:39 any others? it turns out that I forgot to actually click the "create" button on the wiki page last night so I'm not sure if folks tried to add stuff or not 14:06:43 #info task-rpmgrill deployed to production - mkrizek 14:06:58 #info new task-example-gzip repo to showcase simple dist-git-style task, for flock presentation 14:06:58 #info task artifacts are now compressed - mkrizek 14:07:06 #link https://bitbucket.org/fedoraqa/task-example-gzip 14:07:09 #chair mkrizek kparal jskladan 14:07:09 Current chairs: jskladan kparal mkrizek tflink 14:07:21 #info new libtaskotron released and deployed - mkrizek 14:07:36 #info create_report and resultsdb directives now support 'file' as input parameter 14:07:45 #info new much improved bash directive 14:07:54 (now called 'shell') 14:08:16 is all that in the newly released libtaskotron? 14:08:30 mostly not yet 14:08:36 we'll need another release this week 14:08:45 k, just checking 14:09:04 we may want to do that sooner than later so there's time for that new release to make it past updates-testing 14:09:22 that brings us to a problem we identified today 14:09:31 minions still pull from tflink copr repo 14:09:32 oh? 14:09:41 and they are also built from it 14:09:43 ha 14:09:52 yeah, we may want to update that 14:09:55 so even if we push it to updates-testing or stable updates, it won't have any effect 14:10:01 s/may want/should 14:10:15 copr or infra repo? 14:10:16 i was thinking of the workshop - I assume that you'll be having folks install libtaskotron? 14:10:21 I'd like to do something about it this week, so that I'm able to showcase even minion execution on flock 14:10:50 that sounds like a good plan to me - probably should have been done already 14:11:19 mkrizek: I believe they pull from copr, for both building base images and updating libtaskotron when tasks are executed 14:11:44 tflink: yes, I'll ask people to install libtaskotron. having it in updates-testing is probably good enough 14:12:09 kparal: it must be infra, since minions are updated to 0.4.14 14:12:30 hmm, interesting. building base images is still black magic to me 14:13:14 we don't build in copr anymore but we still do put new packages into infra (so it's quicker that waiting for karma) 14:13:16 jskladan said this is in the kickstart: 14:13:17 repo --name="tflink-copr" --baseurl=https://copr-be.cloud.fedoraproject.org/results/tflink/taskotron/fedora-$releasever-$basearch/ 14:13:48 but isn't infra enabled as well? 14:13:58 the minions update from infra as a setup phase 14:14:16 when we spin them up - but that's taskotron's doing 14:14:40 oh I see, minions in our infra are configured to update from infra on spin up 14:14:41 either way, we should probably change that to the fedora repos, allowing for updates-testing 14:14:42 or, to be precise - we add the infra repo in the post-image-build steps specific to taskotron 14:14:48 but for regular people, it pulls from copr 14:15:00 but libtaskotron does update on the machine startup 14:15:04 if nothing changed 14:15:05 it could pull from copr 14:15:09 (and dnf works) 14:15:28 if there's a newer build in the fedora repos, wouldn't it pull from there? 14:15:29 ok, things work correctly in our infra, but won't work correctly for me (on flock) 14:15:50 tflink: we disable fedora repos to speed up spin up time :) 14:15:52 yeah, I think that the consensus is that it would be good to get all that updating from the same place 14:15:55 tflink: only enable the extra copr repo 14:15:57 ok, so let's remove copr from default taskotron.yaml and from minion kickstart 14:16:07 tflink: we might need to re-evaluate that :) 14:16:32 kparal: not adding the fedora repos or getting libtaskotron from fedora repos? 14:16:55 I'm actually not sure what is the best solution here 14:17:15 does anyone want to take a #action on this? 14:17:20 whether we should pull everything from fedora repos (and which ones), or whether we should have a separate copr repo for minions 14:17:42 i don't think that copr repo would be ok with infra 14:17:46 kparal why have separate repo for minions? 14:17:59 how are we going to test develop versions of libtaskotron? 14:18:08 *-testing? 14:18:27 either fedora or infra 14:18:54 and how will the minion know whether to update libtaskotron from stable updates, or even updates-testing? 14:18:58 i think that partially gets back to the questions of which images are produced and how folks get access to them 14:19:20 add updates-testing as a repo in conf? 14:19:36 I think we have 2 use cases - stable minions for production and general audience, and unstable minions for dev/stg 14:19:41 well, if there's a new version that it updates it, I am not sure I follow 14:19:54 *than 14:20:49 do we still use testcloud for pre-spinup configuration of repos, or do we currently bake the information in during image creation? 14:21:03 I know we changed it, but I don't know how 14:21:15 we don't use testcloud for that afaik 14:21:23 i thought that we were still using testcloud but it's been a while since I was in that code 14:21:48 ok, let's look at how it's currently done and talk what to do with it 14:21:55 *then talk 14:22:09 is this something we want to figure out now or wait until flock when many of us will be in the same room? 14:22:25 I would like to see ideally the same minion pulling just from updates in production env, and also from updates-testing in dev/stg env 14:22:44 https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/taskotron/taskotron-client/templates/settings.py.testcloud.j2 - we don't anymore 14:23:04 we don't add repos in testcloud anymore 14:23:05 I think this is something we can chat about in person, but I'd need at least some quick fix to make minions work well when used in public 14:23:11 we just add it in the postprocess 14:23:21 by working well, I mean updating to latest libtaskotron when I spin them up 14:23:42 they will update to the latest libtaskotron in updates kparal 14:23:46 the simplest fix is probably to build testcloud in our copr as well 14:24:01 not sure what that would fix 14:24:09 neither am i 14:24:36 but I'll renew my question of - is this something that we need to figure out now or can it wait until Flock/after-flock? 14:25:03 I think that if I provide a minion with "latest" libtaskotron at the time for kparal 14:25:07 quickfix now, proper fix later 14:25:14 isn't the simplest fix to push libtaskotron to fedora stable? 14:25:19 mkrizek: it is :) 14:25:28 then let's do that :) 14:25:30 once again, minions ignore fedora repos at spin up time 14:25:38 mkrizek: nope 14:25:46 not at image build time, though 14:25:50 ah 14:25:52 ^^ 14:26:02 ok, that would work 14:26:06 \o/ 14:26:12 (or to copr, as it is now) 14:26:16 when will the next batch of changes be ready for release? 14:26:21 but we need to build new libtaskotron first, then build the minions, then publish them 14:26:25 yeah, for this, copr would also work 14:27:01 I think we're mostly ready, but we might do some last touches tomorrow and build new libtaskotron tomorrow 14:27:06 sounds good 14:27:38 the pkgdb group stuff is working, isn't it? ie, we aren't hung up on one person for any part of the process? 14:28:04 the group is broken for bodhi submissions 14:28:12 but it works for dist-git and koji 14:28:24 have folks been added as maintainers directly, then? 14:28:38 or is mkrizek the only one who can create updates? 14:28:53 lmacken might have added our group into bodhi whitelist 14:28:59 * kparal checks older conversation 14:29:22 the bug is reported here: https://github.com/fedora-infra/bodhi/issues/872 14:29:30 either way, 14:29:48 .hello linuxmodder 14:29:49 linuxmodder: linuxmodder 'Corey W Sheldon' 14:29:50 kparal: bodhi has a hardcoded list of "important_groups" that it pays attention to... currently it's: 14:29:50 important_groups = proventesters provenpackager releng security_respons packager bodhiadmin virtmaint-sig kde-sig eclipse-sig infra-sig gnome-sig python-sig robotics-sig 14:29:51 I can add qa-tools-sig 14:29:51 fixed in ansible. 14:29:55 #action mkrizek to add at least one more person as maintainer of libtaskotron so we have backups 14:29:55 (kinda here) 14:30:04 I don't know if it has been deployed yet 14:30:33 i guess we'll find out when the release is done this week :) 14:30:50 good idea, I'll try to submit it 14:31:07 anything more on this? 14:31:24 not here 14:31:29 probably no 14:31:33 #action kparal to do new build and release of libtaskotron when needed-for-flock features are finished 14:32:05 ok, moving on 14:32:12 #topic Flock 14:32:21 How are we doing on preparations for Flock? 14:32:31 it sounds like most everything is on track? 14:33:12 the simple example task that I created should be enough materials for the workshop 14:33:29 I still need to prepare how I'm going to show it 14:33:44 I want to have different stages, increasing complexity gradually 14:34:10 we have picked an example that should work offline, mkrizek's idea 14:34:13 do you see any problems with having that done in time? 14:34:31 nope, if something's missing when I go on PTO, I simply ask mkrizek to finish it :P 14:34:47 interesting approach :-P 14:35:24 or cancel PTO 14:35:36 I already did, partially! :) 14:35:51 so overall I hope the workshop should be prepared 14:35:58 kparal is apparently only a limited team player ... 14:36:01 * tflink kids 14:36:18 sounds good 14:36:20 I'd like to know what's the internet access is going to be like, but I guess it's going to be a surprise, as always 14:36:42 my experience is that it's wise to count on no internet access when making presentations 14:37:15 anything else here? 14:38:05 nothing here 14:38:07 * tflink assumes silence == no 14:38:16 #topic Phabricator and qadevel 14:38:41 so, it sounds like there are still some bugs in the oauth2 stuff I wrote for phabricator 14:39:01 I'm going to see if I can get that figured out in the next couple of days 14:39:16 I had been hoping to move qadevel to qa.fedoraproject.org before Flock 14:39:44 so phab.qadevel.cloud.fedoraproject.org would become phab.qa.fedoraproject.org 14:40:18 stg is already like that, had been waiting for auth to not be horribly broken before moving production over 14:40:49 assuming that production ipsilon is updated in time, any objections to doing that this week? 14:41:36 nope 14:41:39 it might not be a good idea to do it on friday :) 14:42:33 kparal: because you'll be on PTO? 14:42:38 oh, 'not" 14:42:40 nevermind 14:42:56 part of it'll depend on when ipsilon is updated 14:43:09 I'll coordinate with infra on that 14:43:43 #info we're planning to migrate qadevel to qa.fedoraproject.org before flock 14:43:57 #action tflink to coordinate with infra on whether this will be possible or not 14:44:06 will we need to update links in docs? 14:44:14 possibly 14:44:22 because we need a new libtaskotron build to upload new docs, right? 14:44:23 my plan is to make redirects work transparently 14:44:27 alright 14:44:31 in that case, no issue 14:44:42 assuming the redirects work, yes :) 14:45:51 anything else on this? 14:46:13 nothing here 14:46:21 ok, moving on 14:46:39 #topic Docker Testing Status 14:47:19 i feel like this may have been resolved already and I just can't find it in my notes 14:48:50 does anyone know of any bits that we're missing to support docker image testing in taskotron? 14:48:59 other than concrete use cases, any ways 14:49:15 I'm not sure about the shell directive status 14:49:17 kparal? 14:49:57 lbrabec is unfortunately absent. but I'm not aware of anything particular we're missing. 14:50:08 our shell directive is much more useful now for docker testing 14:50:16 good to hear 14:50:50 I guess we'll need to see some real images to play with, in order to identify what more to do about this 14:50:51 * tflink is talking with releng folks later today about that, wanted to make sure he wasn't missing anything large or obvious 14:51:00 reporting from bash is now also reasonably easy 14:51:03 I'm not sure what the status of that is 14:51:18 we'll need trigger changes, of course 14:51:22 #action tflink to check on status of OSBS integration with koji/fedmsg 14:51:35 and probably some docker download directive 14:51:50 but all of that waits until fedora docker infra is in place 14:51:57 yeah, still not quite sure how that's going to work 14:52:08 it's mostly in place as of last week, AFAIK 14:52:15 the remaining bits may be done 14:52:32 we're almost out of time, anything else here? 14:52:41 so much for a quick meeting :-/ 14:53:02 nothing from me 14:53:21 #topic python package check 14:53:40 not sure what to call this but I was wondering what the plan was for the python check which was submitted in phab last week 14:54:11 https://phab.qadevel.cloud.fedoraproject.org/D948 14:54:58 is there a plan for that check? 14:55:21 the submitter approach us last week and said he had a nice useful python check, and whether we can execute it for all newly built python packages in koji 14:55:46 so we said "yeah sure, that's exactly what we can do" and asked him to send us the check for review 14:55:51 ok 14:56:01 do we know how we're going to handle triggering? 14:56:21 for the moment, we'll probably execute it on all packages, and it will ignore non-python ones 14:56:26 we'll just trigger for all koji builds 14:56:31 ok 14:56:52 just wanted to make sure someone knew what was going on :) 14:57:05 #topic open floor 14:57:23 is there anything that folks wanted to bring up that we didn't cover? 14:57:48 nope 14:57:58 nothing here 14:58:10 is anyone in need of tasks? 14:59:00 nope 15:00:25 ok, if there's nothing else, I'll light the fuse and we can head over to the qa meeting 15:00:44 * tflink will send out minutes shortly 15:00:48 thanks for coming everyone 15:00:51 #endmeetiong 15:00:53 #endmeeting