16:04:51 #startmeeting Fedora QA meeting 16:04:51 Meeting started Mon Jan 8 16:04:51 2018 UTC. The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:04:51 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:04:51 The meeting name has been set to 'fedora_qa_meeting' 16:04:55 #meetingname fedora-qa 16:04:55 The meeting name has been set to 'fedora-qa' 16:04:59 #topic Roll Call 16:05:02 Hi 16:05:06 .hello tflink 16:05:06 Happy New Year everyone!* 16:05:07 Hi adamw! 16:05:08 * siddharthvipul1 is here 16:05:12 tflink: tflink 'Tim Flink' 16:05:13 .fas lailah 16:05:15 * sumantrom[m] is here 16:05:15 * kparal is here 16:05:15 Kohane: lailah 'Sylvia Sánchez' 16:05:20 * tenk is here 16:05:25 Happy New Year everyone! 16:05:25 * coremodule is here 16:05:29 * Kohane is here 16:05:31 * (who isn't on the lunar, Julian, Mayan or Illuminati calendar) 16:05:48 LOL 16:05:52 Happy New Year Everyone :) 16:06:12 funny story, i live in vancouver where like half the people are chinese so i already do two new years 16:06:26 now i started playing tennis with a guy who's on the Julian calendar so now i get three, heh 16:06:34 Bonne annee !! 16:07:17 how's everyone doing? 16:07:21 adamw: now you just need to find people who are on the Mayan and Illuminati calendars! 16:07:42 tflink: well, if i knew anyone on the illuminati calendar of course i couldn't tell you...;) 16:07:55 fair enough 16:08:08 adamw: that must be confusing... 16:09:00 Kohane: eh, you just great everyone with Merry Happy New Sam Dim Christmas Fai Lok Year and hope for the best :P 16:09:31 xD 16:09:36 Hi tenk! 16:10:08 * sumantrom[m] waves at coremodule tflink and kparal 16:10:35 allllrighty 16:10:45 let me remember what the hell i put on the agenda all the way back on friday night... 16:10:56 #topic Previous meeting follow-up 16:11:23 we had one action item: 16:11:24 * pschindl_wfh is late but here. 16:11:29 "sumantro to create f28 milestones in pagure (and close old tickets/milestones as appropriate)" 16:11:33 sumantro, did that go ok? 16:11:37 morning pschindl 16:11:50 Yes that's sorted! 16:12:21 excellent 16:12:30 #info "sumantro to create f28 milestones in pagure (and close old tickets/milestones as appropriate)" - this is done 16:12:40 anyone have any other follow-up from last time? 16:13:51 I can't remember what we discussed last time. 16:13:56 Or maybe I couldn't attend. 16:14:02 :-/ 16:14:10 Kohane: you can find the logs etc. on meetbot 16:14:18 Kohane: https://meetbot-raw.fedoraproject.org/teams/fedora-qa/ is a handy URL to have around 16:14:29 Oh, good, thanks 16:14:33 I'm so clumsy with IRC 16:14:35 you can find the records of *almost* all the meetings ever there (some very old ones aren't there) 16:15:11 Oh, okay, thanks 16:15:12 blocker bug meetings are at https://meetbot-raw.fedoraproject.org/fedora-blocker-review/ , at least the ones since we started using a dedicated channel (older ones are harder to find) 16:16:51 Good. Thanks adamw ! 16:17:27 alrighty then, moving on 16:17:37 #topic Meltdown / Spectre status / discussion 16:17:48 so, since this is the big topic lately, i figured it might be a good idea to give it some space at the meeting 16:18:53 the status as I understand it is that we have shipped the fix/mitigation for Meltdown to F26, F27 and Rawhide; we have not yet shipped any fix/mitigation for any form of Spectre 16:19:06 * satellit_ listening late to join 16:19:34 i believe kernel team considers it slightly less urgent to SHIP ANYTHING QUICK for spectre and wants to work with upstream to ship the right things 16:19:47 jforbes: if you're around, is that accurate? 16:20:09 (i'm assuming everyone knows what the hell i'm talking about here, if not, please yell) 16:20:21 Yes, that is correct 16:21:01 thanks 16:21:06 adamw: I should have patches for variant 1 in the next builds. Variant 2 is a bit more difficult because we are dependent on microcode that intel hasn't shipped yet (a day or 2) and updates to gcc 16:21:32 so should we expect first a kernel update for variant 1, then a combined kernel/microcode_ctl/gcc update for variant 2? 16:21:39 jakub said nothing would happen with Fedora GCC until those patches were posted upstream and discussed. They were just posted, so we will see 16:22:51 Yes, well, they won't be one update, if we build the kernel with retpoline and the compiler doesn't support it, the kernel will taint. So gcc first, then kernel. I expect microcode will happen before gcc does 16:23:33 well, you can build gcc, create a buildroot override for it, then build kernel with that gcc and submit a combined update. but they don't really *have* to be updated together, i guess. 16:23:55 it'd just go faster that way as you don't have to wait for gcc to be pushed stable before building kernel. 16:24:34 But variant 2 is extremely hard to exploit, and made harder by how quickly our kernel moves. AFAIK an exploit has to be crafted for a particular kernel build and CPU combination 16:24:45 roger. 16:25:07 so for us, folks, the status is we have nothing to do right now, i'll try and stay in touch with the kernel team so i can flag up updates for testing as they appear. 16:25:18 Yes, I will do an override when that happens, but the updates aren't bound in any other way. The kernel doesn't care if that gcc is installed on the system, only that it was used to build the kernel. 16:25:23 gotcha. 16:25:57 #info Meltdown fix has shipped to Fedora 26, 27 and Rawhide. Spectre fixes are pending, first for variant 1, then variant 2. adamw will stay in contact with folks working on the fixes and flag up updates for testing as they appear 16:26:09 any other questions/concerns/thoughts on meltdown/spectre, anyone? 16:26:25 we're all so very screwed. 16:26:31 (hey, you asked for thoughts.) 16:26:31 Why? 16:27:10 Basically any kernel updates you see over the next couple of weeks, test and karma quickly. They aren't all tagged with the CVE bugs for them because we are only doing particular arches, etc. Like right now, meltdown is coverred for x86_64 only. PPC patches should be posted today, arm64 are in mid discussion. 16:28:45 ah, right. 16:29:18 #info all kernel updates that land over the next couple of weeks should be tested urgently - they may address aspects of Meltdown/Spectre even if not marked with the relevant CVEs (e.g. fixes for ppc64 / aarch64) 16:29:44 Kohane: pjones is the crusty old guy who's seen some things, man. he's seen some things. 16:29:52 heh 16:29:53 Oh... 16:29:56 I see 16:29:57 LOL 16:30:09 Hi crusty old guy, nice to meet you 16:30:17 uh, same, I think? ;) 16:30:54 xD 16:33:01 alllrighty then - thanks, jforbes. moving along 16:33:18 #topic Fedora 28 Change review and status 16:36:31 so, quite a lot of the Fedora 28 Changes are landing on devel@ ATM 16:36:45 yeah 16:36:46 i figured for the next few meetings we can have a topic to look through and take note of any significant ones 16:37:06 Sounds sensible 16:37:53 Adamw roger 16:38:21 it would of course have been sensible for me to look through the list and have some primed and ready to go... 16:38:26 ...so of course i have not done that! 16:38:33 that sounds like work 16:38:33 we'll just DO IT LIVE, man 16:39:17 Yeah!! Let's do it live! 16:39:19 LOL 16:39:54 so, one that got proposed back in december is https://fedoraproject.org/wiki/Changes/ReduceInitialSetupRedundancy 16:40:26 that'd very obviously hit the release criteria, so assuming some version of it gets accepted/implemented, we'll need to re-check all the possible first boot flows on all the release-blocking images 16:40:28 that might be a great test day 16:40:38 * adamw looks meaningfully at sumantro :P 16:41:04 Yes we will have it ! 16:41:16 I am waiting for mass rebuild :) 16:42:51 gret 16:42:53 great 16:42:56 * adamw looks for others 16:43:56 https://fedoraproject.org/wiki/Changes/StratisStorage may be relevant to release validation if support gets added to the installer 16:44:16 aside from that it'd again be a good candidate for a test day, i suspect, if the support is advanced enough to be worht it 16:45:26 I don't know what is a Stratis Storage but I like that we're reducing setup redundancy. 16:46:09 there's a description of Stratis on the Change page 16:46:15 Kohane: it sounds good, but the devil is in the details :) 16:46:52 I know, I know. 16:47:05 But at least it's something I can test and fully understand 16:47:07 https://fedoraproject.org/wiki/Changes/StrongCryptoSettings could well cause some messes, worth keeping an eye on. not entirely sure how we'd go about running a test day for it, as it could affect an awful lot of stuff 16:48:26 https://fedoraproject.org/wiki/Changes/ThunderboltEnablement seems like another test day candidate, assuming we know anyone with testable hardware... 16:48:49 (man, i should have been #info'ing all of these) 16:49:00 Actually, yes... 16:49:19 https://fedoraproject.org/wiki/Changes/AnacondaModularization will obviously affect validation, but we'll naturally test it as part of the usual validation testing 16:50:02 oh yikes, https://fedoraproject.org/wiki/Changes/AuthselectAsDefault will again affect validation, both anaconda and freeipa use authconfig 16:50:07 this is gonna be a fun cycle! 16:50:20 you bet! :) 16:50:22 Looks like! 16:50:47 BTW, I don't have any Thunderbolt hardware as far as I can tell, so that's something I cannot test. 16:51:11 oof, and https://fedoraproject.org/wiki/Changes/AArch64_Server_Promotion just showed up today...entirely missing the bit where we would need someone with aarch64 hardware to test it. 16:51:36 well, pwhalen is a co-proposer, i guess. 16:51:38 * pwhalen raises his hand 16:51:43 What kind of computer is that? Never heard of before. 16:51:49 Kohane: 64-bit ARM 16:51:56 aarch64 is the 'official' arch name for it 16:52:10 pwhalen: are you really volunteering to run all the Server tests manually? 16:52:21 pwhalen: doing that is a gigantic pain, which is why i spent weeks teaching openqa to do it... 16:52:29 Ah, okay. Thanks adamw 16:52:38 Now it sounds familiar 16:53:04 openqa works for aarch64 as well, but between myself and others we can also cover the manual testing 16:53:07 pwhalen: it might be easier/better to add some aarch64 workers to openqa, we've already added ppc64 so we've kinda proven the process 16:53:14 pwhalen: we'd need hardware to test in openqa 16:53:26 right, we should have some hw for it 16:53:39 pwhalen: do we have hw in phx? 16:53:56 yes, already in place 16:54:01 (note for others: phx is where the main fedora data center is, including our openqa deployment) 16:54:07 pwhalen: aha, great. i'll follow up with you on that then 16:54:17 great, thanks 16:54:24 #action adamw and pwhalen to discuss adding aarch64 workers to openqa 16:54:55 #info see logs for call-outs of several specific Changes for validation testing / Test Day purposes, sorry for not individually #info'ing them 16:56:06 OK, we're running out of time, so moving on quickly - we can have this topic again next week for any further discussion 16:56:07 #topic Test Day status 16:56:17 sumantro: i guess the status is that you're waiting for a bit later in the cycle at present? 16:56:36 yes 16:56:40 exactly 16:57:16 we have the bits in place and many lined up as you marked but then we need to have the mass rebuild to complete 16:57:24 roger, indeed 16:57:42 #info most test day planning is waiting for later in the cycle, after the mass rebuild and when the Changes start to land 16:57:46 #topic Open floor 16:57:51 any other business, quickly? :) 16:58:00 note, we have blocker review in #fedora-blocker-review at the top of the hour. 16:58:09 quickly: HoF 26 and the HoF period (to short for me) 16:58:21 i think coremodule was meant to be doing the HoF, right? 16:58:25 HoF? 16:58:32 Kohane: Heroes of Fedora 16:58:36 Oh 16:58:49 I thought it was related to software, sorry. 16:58:58 Kohane: we usually run some blog posts (in various places, lately it's been the community blog) which list out major contributors to testing 16:59:00 * Kohane smiles nervously ashamed 16:59:01 for me the period is to focus on the final period and not relevant for people doing test all the year 16:59:06 Kohane: don't worry :) 16:59:41 tenk: indeed, it seems like only 27 Final HoF was run 16:59:48 not Beta or pre-Beta 16:59:57 i talk abouy the HoF 26 :) 17:00:04 tenk: i think we did a full set for f26... 17:00:15 or did we only do alpha? mm 17:00:26 hmm, yes 17:00:30 we did alpha I suppose 17:00:37 https://communityblog.fedoraproject.org/tag/heroes-of-fedora/ - it looks like every milestone from f26 alpha to f27 final is missing 17:00:46 coremodule: do you think you could do some fill-in posts? 17:01:09 and about the time to focus? 17:01:30 tenk: we really want people to be testing all the time 17:01:59 tenk: even for validation testing, it's really useful to know the status of the Final tests early in the cycle - it's much better to know something is broken and have it fixed in advance, than to find out late 17:02:43 tenk: of course, if you have, say, a week or two that you can "spend" on really focused testing...it's always useful to have folks around for the time right around go/no-go 17:02:47 yes of course but for me the time is to focus on the late test. Not o the early validation testing 17:03:46 adamw: ok i understand, push people to do some test during this period if they can only do few weekds of test 17:04:11 yeah, 17:04:22 OK, we're a bit over time, so let's wrap it up :) 17:04:25 thanks for coming everyone 17:04:26 * adamw sets the fuse 17:05:03 adamw thanks for hosting 17:05:45 see you over in #fedora-blocker-review for some blocker reviewing fun times! 17:05:54 See you there! 17:06:12 #action adamw to check with coremodule if he can write HoF posts for F26 Beta / Final and F27 Beta 17:06:19 #endmeeting