16:00:48 #startmeeting Fedora QA meeting 16:00:48 Meeting started Mon Mar 5 16:00:48 2018 UTC. The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:48 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:48 The meeting name has been set to 'fedora_qa_meeting' 16:00:51 #meetingname fedora-qa 16:00:51 The meeting name has been set to 'fedora-qa' 16:00:55 #topic Roll Call 16:00:58 morning folks, who's around? 16:01:00 * sumantro is here 16:01:07 * kparal is here 16:01:10 .hello2 16:01:11 frantisekz: frantisekz 'FrantiĊĦek Zatloukal' 16:01:12 morning adamw :) 16:03:18 morning morning 16:04:44 .hello2 16:04:45 tflink: tflink 'Tim Flink' 16:04:53 hi tim 16:05:00 #chair sumantro tflink frantisekz 16:05:00 Current chairs: adamw frantisekz sumantro tflink 16:07:07 welp, let's get at it 16:07:14 #topic Previous meeting follow-up 16:08:57 #info "adamw to kick things again re aarch64 openQA workers" - well, i didn't exactly formally kick this, bad me...I *did* notice someone has been working on this, though, as stuff for the proposed worker machines showed up in infra ansible. will check in again this week 16:09:06 #action adamw to check in again on aarch64 openQA workers 16:09:14 that's the only action item from last week...any other follow-up? 16:11:12 guess not! 16:11:19 #topic Fedora 28 status, freeze, schedule, major features 16:11:33 sooo, yeah, this is fun! 16:11:40 :) 16:11:55 :D 16:11:57 Hi, I will have some remarks but I will wait for the open floor. 16:12:32 hi there 16:12:48 so, uh. it took a long time to get f28 to compose at all. 16:13:21 then we *did* get a couple of composes, 0301.n.1 and 0302.n.0. we got a validation event for one of those. 16:13:27 (so please do go test that) 16:14:10 now, well, for some reason (reason still being figured out, but dollars to donuts it involves Modularity), composes are taking about a day and a half to complete 16:14:20 * satellit late join 16:14:55 so we have the composes for the last three days for f28 and the last two days for rawhide all running together, and who knows what's gonna come of that. the Rawhide compose for 0303 just completed a few hours ago. 16:15:20 so, that's obviously a big problem. 16:16:15 note, for folks who want to keep track of compose status, dustymabe has set up a neat new initiative; there's a pagure repo at https://pagure.io/dusty/failed-composes/ where tickets get filed automatically for all failed composes, and we're trying to keep notes on the reasons and status in there 16:16:28 it's a bit easier to follow than #fedora-releng IRC (where this stuff usually gets discussed)...we hope 16:16:56 While at this, is gdm and gnome-shell starting? There were some issues with mutter, iirc. 16:17:08 other than that, in the recent composes, anaconda failed to run on live images, which is a clear blocker bug. that *ought* to be fixed as of today, but of course as things stand, we won't get official images with the fix for a day and a half. 16:18:09 rluzynski: well, it worked in openQA with the last completed composes: https://openqa.fedoraproject.org/tests/197131 , https://openqa.fedoraproject.org/tests/196796 16:18:33 rluzynski: and jens petersen mentions it working in https://bugzilla.redhat.com/show_bug.cgi?id=1547691 16:18:43 "Fedora-Workstation-Live-x86_64-28-20180302.n.0.iso 16:18:43 Live instance works fine for me" 16:19:43 there is also an obvious issue with KDE which i'm in the middle of filing right now; its default panel is completely screwed up (it appears in very weird form at the top left hand corner of the screen) 16:20:09 Thx, that's what I was asking about. 16:20:49 beyond blocker issues, one of the major Changes for F28 - add-on Modularity - theoretically should have appeared only in the most recent composes, so we're obviously pretty late on that. i'm not sure whether it's 'working' at all (or what we even define 'working' as, yet). 16:21:01 anyone more up to date on that than me? sgallagh, perhaps? 16:21:19 * sgallagh looks up 16:21:34 So, it turned up over the weekend finally. 16:22:10 The Rawhide compose now has all the right content there. 16:22:20 F28 is getting built in the next compose, I think 16:22:35 We still need to get a fedora-repos update out that can address this, but that's pretty easy. 16:22:48 I'm not sure where the mirrormanager work is; maybe nirik can answer that 16:22:57 sgallagh: all done 16:23:00 see infra list 16:23:00 awesome 16:23:33 adamw: Does that help? 16:23:40 when you say "all the right content", what does that mean in practical terms? 16:23:47 what "content" is there? 16:29:37 sgallagh: ^^ 16:29:44 * adamw does some infos in the mean time 16:29:50 sorry folks, busy morning, i'm multitasking a bit 16:30:03 Sorry, same here 16:30:14 #info Some Fedora 28 composes did succeed in 20180301 and 20180302, there is a new validation event 16:30:30 I mean it contains the RPMs from the modules listed in the variants.xml that we expect to be there. 16:30:33 #info Since then composes are taking an unreasonable amount of time for some reason, see https://bugzilla.redhat.com/show_bug.cgi?id=1551653 16:30:53 sgallagh: OK. and what modules are there, specifically? what can we try to test (once this fedora-repos update is done)? 16:31:15 #info There is now a tracker for failed composes: see https://pagure.io/dusty/failed-composes/ if you're interested 16:31:17 adamw: The best example is probably the Node.js streams 16:31:27 https://paste.fedoraproject.org/paste/dkWPpFFVOlRiV5yce-0Vpw 16:31:40 those are the ones that exist in the rawhide module repo right now ^ 16:31:55 #info There are several apparent blocker bugs in the composes that did complete, notably anaconda failing to run on live images and KDE's default panel being broken; come to the blocker review meeting next for more on those 16:32:28 adamw: The compose includes Node.js 6 (LTS), 8 (LTS) and 9 (latest) 16:32:47 #info The new 'add-on Modularity' content should be present in recent Fedora 28 composes, but is waiting on a fedora-repos update to be usable; once that is available, sgallagh recommends the differently-versioned Node.js streams as the most practical testable modules 16:35:31 #info there is also some sort of signing issue related to the compose, waiting for puiterwijk to file tickets on that to clarify it 16:36:18 adamw: short summary: autosigning for modules needs to be rewritten, so that's what I'm working on. 16:36:22 #info F28 Beta freeze and Bodhi enabling are scheduled in a few hours' time (2018-03-06 00:00 UTC) 16:36:29 puiterwijk: oh goody 16:36:42 puiterwijk: was 'allow f28 key for modules for now' a temporary workaround for that, then? 16:36:59 adamw: "allow modular key for f28 composes", yes 16:37:15 I can explain it all outside of the meeting 16:37:32 * sgallagh apologizes profusely to puiterwijk once again for the oversight that led to this fire-drill. 16:37:33 np 16:38:25 sgallagh: the fun part is: not only signing needs massive changes. Bodhi too, for the same reasons, but I just had a meeting about that so it's fun. 16:38:36 Anyway, sorry for derailing 16:41:23 hashtag info everything is terrible 16:41:35 sooo, that's about where we are with f28. 16:41:42 does anyone have any more fire to add to the giant pile of fire? :) 16:41:49 mattdm: ^^^ 16:42:29 * rluzynski waiting for Open Floor 16:43:44 we'll get there 16:43:56 #topic Test Day status 16:44:12 sumantro: got a quick update for us on what's coming up in test days? 16:44:43 i18n Test day coming up 2018-03-13 16:44:59 Kernel test day went good 16:45:40 yay 16:45:54 we have cloud test day coming up and few others in line 16:48:40 alrighty 16:49:07 #info Kernel test day on 2018-02-22 went off well with 56 folks joining in - thanks to everyone who tested! 16:49:43 #info i18n Test Day is coming up on 2018-03-13: please join in if you can - https://pagure.io/fedora-qa/issue/527 16:49:57 and with that, let's get to the promised open floor. 16:50:00 #topic Open floor 16:50:07 rluzynski: you had something for us? 16:50:43 So, it's about the dates/calendars issues which I have been working on. 16:51:41 It's an upsteam issue in glibc, fixed in January and it has already landed in F28. 16:52:36 sorry, can you explain the issue? not sure i recall it 16:52:54 The problem is that some apps must be adapted to this new change and I am unable to test every app around. 16:54:30 adamw: the dates were formatted incorrectly in several languages, mostly Slavic but also Greek etc. 16:55:55 So a new feature has been added to strftime function. Now 2 grammatical cases of month names are possible and application must select the correct one. 16:56:29 okay. and the consequence of not adapting would only be that the apps would continue to use the old, incorrect form? or something worse? 16:57:47 Kinda... Actually, the apps not adapted will display the incorrect grammatical form. 16:58:15 The change is visible in 7 languages only, atm. 16:58:36 Probably Czech and Catalan will be also added soon. 16:59:22 okay. 16:59:32 so, what can qa do for you regarding this? 16:59:39 Those 7 are: Belarusian, Croatian, Greek, Lithuanian, Polish, Russian, Ukrainian. 16:59:43 (side note: i'll kick off the blocker review meeting in #fedora-blocker-review soon) 17:00:33 Test the apps, mostly calendar apps, in those locales, and report bugs. 17:01:21 Should we finish and continue on email? 17:02:46 sure. thanks for the heads up 17:03:02 Ok, thx. 17:03:09 if you could post a summary of the issue and an easy way to test (what a 'wrong format' and 'right format' look like, for e.g.) on test@ that'd be great 17:03:24 Ok 17:03:57 #info there is a glibc issue with calendar formatting in certain languages and we would like to ensure as many apps as possible are adapted to use the correct format. rluzynski will post a testing guide to test@ soon, it'd be great if folks can help test out applications and find ones that need updating 17:04:01 anything else for open floor? 17:04:18 also, blocker review meeting is starting in #fedora-blocker-review right now, so come join the 'fun' :) 17:08:04 ok, thanks for coming, everyone! 17:08:06 #endmeeting