fedora-qa
LOGS
16:00:06 <adamw> #startmeeting Fedora QA meeting
16:00:06 <zodbot> Meeting started Mon Feb 13 16:00:06 2017 UTC.  The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:06 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:06 <zodbot> The meeting name has been set to 'fedora_qa_meeting'
16:00:14 <adamw> #meetingname fedora-qa
16:00:14 <zodbot> The meeting name has been set to 'fedora-qa'
16:00:17 <adamw> #topic Roll call
16:00:27 <adamw> ahoy hoy folks, who's around for exciting qa meeting times?
16:00:32 <roshi> .hello roshi
16:00:35 <marc84> hi everyone
16:00:36 <zodbot> roshi: roshi 'Mike Ruckman' <mruckman@redhat.com>
16:00:45 <tflink> .hello
16:00:45 <zodbot> tflink: (hello <an alias, 1 argument>) -- Alias for "hellomynameis $1".
16:00:49 <tflink> .hello tflink
16:00:50 <zodbot> tflink: tflink 'Tim Flink' <tflink@redhat.com>
16:01:03 <coremodule> .hello coremodule
16:01:04 <zodbot> coremodule: coremodule 'Geoffrey Marr' <gmarr@redhat.com>
16:01:07 * garretraziel is here
16:01:31 * a2batic is here
16:02:21 * sumantrom is here
16:02:36 <adamw> cmurf: morning, you around?
16:03:19 <cmurf> yes
16:03:58 * pschindl_ is here
16:04:01 <Southern_Gentlem> .hello jbwillia
16:04:02 <zodbot> Southern_Gentlem: jbwillia 'Ben Williams' <vaioof@yahoo.com>
16:04:51 <adamw> hi cmurf, southern gentlemen, everyone at the borg :P
16:05:06 <adamw> so! let's get right into it
16:05:18 <adamw> #topic Previous meeting follow-up
16:05:25 <adamw> this is gonna be short as it seems like we don't have any!
16:05:30 <adamw> #info no action items at last meeting
16:05:51 <adamw> can anyone think of anything we needed to follow up on which wasn't action item-ed?
16:05:51 <roshi> whee!
16:05:55 <roshi> good work everyong!
16:06:05 <roshi> s/yong/yone/
16:06:11 * roshi has nothing
16:06:39 <coremodule> For everyong, I want to be for everyong.
16:07:25 * kparal is here
16:08:04 <adamw> if this was 1953, i could've made a really insensitive joke there.
16:08:34 <roshi> lol
16:08:39 <adamw> aaallllrighty then
16:08:40 <adamw> moving along!
16:08:56 <adamw> kparal: morning, sorry about the calendar
16:09:12 <adamw> #topic Factory 2.0, CI, Modularity, no-Alphas, Atomic, oh my...the future of Fedora QA
16:09:20 <kparal> no problem
16:09:21 <adamw> so! any questions? :P
16:09:39 <adamw> i realize this is a big topic, and i don't have a specific agenda for discussing it; sorry if that makes things messy
16:10:10 <kparal> my only question is... is this fine?
16:10:15 <adamw> but those of us who were at devconf recently wound up thinking about a lot of big picture stuff and i wanted to find a way to get that out to 'the community' even if it's a bit sloppy
16:10:26 <adamw> kparal: https://thenib.com/this-is-not-fine
16:10:38 <cmurf> adamw: assume people have read the blog post and the recommended videos?
16:10:46 <adamw> cmurf: that might be a good place to get started
16:11:01 <adamw> that is, this blog post: https://www.happyassassin.net/2017/02/12/the-future-of-fedora-qa/
16:11:14 <adamw> and the videos linked there, especially https://www.youtube.com/watch?v=5gqccjyjwFk (Ralph Bean's Factory 2.0 talk)
16:11:44 <roshi> so what's the timeline for all of this
16:11:46 <adamw> so really i just wanted to open up a bit of space for us to talk about...Change :)
16:12:01 <cmurf> roshi: Thursday
16:12:04 <adamw> last Thursday
16:12:08 <roshi> as I look at it, I think it'll be a couple years before we're in the glorious future where we can just do testdays :p
16:12:37 <roshi> or dig down into the other 16,500 packages in the repos for polish
16:12:39 <cmurf> Well, the nifty part is Fedora now has a unicorn... but that's getting ahead of things.
16:12:40 <adamw> it seems like at least some of it is supposed to be happening quite fast. there seems to be some sort of plan to have at least an initial implementation of 'CI', roughly defined as changes to Atomic host getting tested on the commit level, within 2016
16:12:44 <adamw> (is that still right tflink?)
16:13:04 <tflink> so far as I know, yes
16:13:11 <adamw> the whole ball of wax is, yeah, i would guess further out than that, but it gets a bit crystal ball-ish at that point
16:13:32 <roshi> right, but getting a new machine installed and getting it to exactly where you want it are different things
16:13:36 <kparal> but we all know how software development planning turns out in most cases
16:13:49 <roshi> and I see the standing up of CI for atomic as closer to the installing a fresh machine
16:14:00 <roshi> but then there's a ton of work putting it to it's full use
16:14:09 <roshi> and that's if you're *not* debugging it as you build it
16:14:11 <adamw> to put it in release cycle terms: it seems there's a goal to have the Atomic CI stuff (which is really kinda like the current two-week Atomic process, but you know, less stuck together with duct tape and bash scripts) during the 26 stable cycle, which means having the work done by the time 26 goes out
16:14:37 <roshi> that makes sense
16:14:44 <adamw> there is also a plan to have a PoC for Modularity in 26, which would be a compose of Fedora Server using the 'modularity' bits. they have a codename for that, i forget what it is, anyone?
16:14:56 <tflink> boltron, I think
16:14:59 <adamw> note that really is a PoC, it's certainly not a release blocking deliverable and it may not be really 'usable'
16:15:00 <adamw> that's the one
16:15:21 <adamw> but then by 27 there seems to be a goal to have actually usable, non-PoC builds of multiple Fedora products
16:15:26 <adamw> er, Modularity builds
16:15:27 <kparal> is boltron somehow derived from taskotron? :)
16:15:35 <adamw> EVERYTHING DERIVES FROM TASKOTRON
16:15:52 <roshi> ^ that
16:16:04 <tflink> I think it comes from bolted together voltron but I don't honestly know so I'll default to what adamw said
16:16:16 <adamw> 27 is also the cycle where dennis is talking about no longer having Alphas
16:16:34 <tflink> and upstream based branching for packages
16:16:46 <adamw> i'm not quite sure what the intended timeframe for all the stuff about doing git branches on software versions rather than release versions is
16:16:52 <adamw> ah, that's 27 too?
16:16:53 <roshi> and that's not having Alphas because rawhide is stable as alpha would be, right?
16:16:56 <roshi> from the CI?
16:17:19 <tflink> that's what I understood from the factory 2.0 talk @ devconf
16:17:24 <adamw> roshi: yes. although we're actually kinda working towards that goal a bit...in parallel. that's kinda the intent of the stuff i've been doing lately with resultsdb
16:17:48 <adamw> roshi: i'm trying to get that whole setup to a point where releng could gate rawhide using a thing that just queried resultsdb
16:17:51 <roshi> ok, just wanted to make sure I had it right in my head
16:18:20 <adamw> so for that goal we don't need true 'CI', strictly speaking, as we can just test on the daily composes and not sync them to the public mirrors if they fail tests
16:18:26 <adamw> so we don't have to get down to the package commit level for that one.
16:18:47 <adamw> (or form a proper feedback loop to the packager who actually broke the thing.)
16:19:12 <adamw> man, that blog post is still long, isn't it? i should've cut it into sections. you should've seen it with the original first section, jeez.
16:19:45 <adamw> btw, i forgot to ask: tflink, kparal, everyone else who was at devconf and so on: does my blog post roughly line up with your understanding, or did i mess anything up?
16:19:57 <kparal> I believe it lines up
16:20:04 <tflink> I didn't read it in detail but I didn't see anything that didn't seem right
16:20:10 <adamw> OK, cool.
16:21:04 <kparal> tflink: nice use of negatives :)
16:21:23 <roshi> lol
16:21:29 <adamw> it certainly wasn't not that!
16:21:41 <tflink> proof that being overly negative can be positive?
16:22:09 <kparal> yeah, you just need to be negative even times
16:22:18 <adamw> roshi: so if you're still interested in working the atomic angle, this is definitely going to be an interesting and important place to be over the next little while
16:22:27 <adamw> and it'd be super to have someone really plugged into it
16:22:38 <roshi> yeah, I'm pretty dug in at this point
16:23:19 <adamw> awesome
16:23:30 <roshi> if I go all in, you'll see less of the normal testing (validation and updates) from me though
16:23:35 <roshi> if people are fine with that
16:23:55 <adamw> one thing that concerns me with that whole area is the "how do we make something 'important' to fedora besides calling it a 'release-blocking image'?" question
16:24:24 <adamw> i think it's fine to start out that way, and if we really need more bodies for manual testing we can yank you back in :)
16:24:33 <roshi> for sure :)
16:24:49 <kparal> we have this concept or PrioritizedBug now, but I'm not sure it fits the purpose
16:24:53 <kparal> *of
16:24:56 <adamw> kparal: not really, that's for one-bug-at-a-time stuff
16:24:58 * roshi is always up for some hero runs when the time comes :p
16:25:34 <roshi> well, for Atomic, I think the rest of Fedora just kinda rolls on, as Atomic will only care about critpath things being broken
16:25:54 <roshi> I don't think there will be a FXX prefix on atomic releases in the future
16:26:03 <adamw> sure, but that can happen :) and it's stuff like making sure it composes all the time and so on
16:26:05 <adamw> yeah, that seems to be happening
16:26:06 <roshi> at least, I don't know what good it does with the 2wk releases
16:26:11 <adamw> roshi: do you talk to cwalters much?
16:26:43 <roshi> but then again, what are they going to do when something like the dnf change comes along? (meaning scope, where it breaks compat with stuff, not package managment)
16:26:49 <roshi> nope, why?
16:26:51 <roshi> should I be?
16:27:14 <adamw> it'd probably be handy, as he's ultimately driving a lot of the atomic stuff
16:27:23 <cmurf> i've been working on atomic cloud for a while; and also briefly worked on the atomic workstation as well.
16:27:25 <adamw> and he tends to get restless easy :P it'd be nice to be plugged in to what he wants
16:27:33 <roshi> I've been mostly working with dusty and adam miller
16:27:38 <adamw> cmurf: i don't think Atomic Workstation is as big of a deal at this point
16:27:47 <adamw> but definitely the more people we have following Atomic host stuff the better
16:28:30 <adamw> roshi: cmurf: it'd just be nice to get folks together to draw/update a clear, on-paper definition of what the goals/requirements for Atomic host actually *are*
16:28:49 <cmurf> that's supposed to be the Atomig WG
16:28:50 <adamw> since afaics we don't even have that; the Cloud WG is more or less the Atomic WG now, but i don't think its foundational documents have ever really been updated
16:29:05 <cmurf> but the transition fro Cloud WG to Atomic WG is beyond confusing to me
16:29:14 <roshi> as far as I know they haven't been updated
16:29:19 <adamw> we can't *do* all this
16:29:28 <adamw> but it would be good to at least go to the important people and say, look, it needs to happen
16:29:35 <roshi> yeah
16:29:38 <adamw> it helps all the other groups understand and get behind the effort
16:29:54 <roshi> all that happened while I was on hiatus, so I wasn't even aware of the change until recently :p
16:29:54 <adamw> anyhow, that's my cut on it :)
16:30:00 <adamw> roshi: for sure, i understand that :)
16:30:10 <cmurf> Right so the last email I got the impression Cloud WG becomes a Cloud SIG, all the Cloud images are made from the SIG and aren't release blocking, and Atomic WG produces only rpm-ostree based images.
16:30:38 <cmurf> And then Atomic WG folks need a PRD and all that jazz.
16:30:53 <roshi> yeah
16:31:33 <roshi> I'll bring up the PRD at the atomic/cloud meeting this week
16:31:41 <roshi> anything else lacking?
16:31:48 * roshi will look into it later
16:33:44 <roshi> from my reading, atomic needs a PRD and the Cloud SIG needs a clearer definition/plan on how/who is going to keep the images building
16:34:11 <adamw> yeah, that seems about right. the atomic thing more important strategically than the cloud thing (though we have a direct interest in the cloud images on a tactical level.)
16:34:50 <adamw> updated prd is a big thing to have. from there we can look at it and think of ways to build from there
16:34:52 <roshi> my concern with the cloud thing, is that it falls through the cracks and then someone else has to come along and fix it since we rely on it
16:35:00 <roshi> and those people would be us QA people :p
16:35:07 <adamw> our classical approach has been 'write some release criteria then some tests from the release criteria', and that's what we have for Cloud
16:35:12 <roshi> I'll bring it up at the meeting
16:35:16 <adamw> but we may need to rethink that a bit for what atomic is really trying to do
16:36:15 <roshi> yeah
16:36:38 <roshi> cmurf: what's the unicorn you were referring to? Atomic?
16:37:09 <cmurf> An actual unicorn. Although once word is out, chances are it gets kidnapped.
16:37:45 <kparal> and kidneys stolen
16:37:55 <adamw> chaaaaaaaarlie
16:38:05 <roshi> come on charlie!
16:38:06 <cmurf> What do you want?! I'm right here!
16:38:08 <kparal> adamw is now fully educated
16:38:20 <adamw> allllrighty
16:38:42 <adamw> so did anyone have other questions or thoughts? on the other areas maybe - factory 2.0 stuff, modularity stuff, qa team existentialism? :P
16:40:03 <roshi> I'm less worried about existentialism, since the timeframes are so far out, and our backlog of software bits that need or could use love is so long
16:40:34 <roshi> but when we hit the modularity bits, how do we dtermine which "modules" get our focus? at least in the short term
16:41:01 <roshi> because it could be that QA ends up being more closely tied to infra and writing the code to make all this work
16:41:14 <roshi> QA becomes the ones testing the testing, if that makes sense
16:41:41 <adamw> it's a good question
16:41:45 <adamw> i'm not sure i have an answer :)
16:41:57 <roshi> I don't expect anyone to have one at this point :p
16:42:01 <cmurf> Yes in some ways this sounds at least as significant as Fedora next, when was that gap, 20 to 21? Is there going to be another gap?
16:42:33 <adamw> i guess my instinct would be that it'd be more the responsibility of the groups responsible for each module to ensure at least CI level testing for the module was in place, not QA's job
16:42:34 <roshi> I'd guess there will be between F26 and F27
16:42:42 <adamw> cmurf: there was a proposal to make F27 a 12 month cycle iirc
16:42:46 <adamw> not sure where that is
16:42:50 <adamw> er, where it stands
16:42:56 <cmurf> gotcha
16:42:59 <roshi> right, but we'd be the ones providing help to the groups to get their stuff into CI, right?
16:43:10 <adamw> could be, yeah.
16:43:29 <roshi> so the impetus would be for us to be really tuned in to how the system works and getting things into it
16:43:29 <adamw> i think to some extent it's gonna be a case of figuring it out as we all go along, like it was with .next
16:43:37 <roshi> and debugging it, handling edge cases, etc
16:43:49 <adamw> but it's important to try and be out ahead of things thinking about it
16:45:07 <adamw> of course this isn't the only chance we have to discuss it :)
16:45:21 <adamw> sounds like we're mostly talked out for now, but we can always come back to it in future, on the lists etc.
16:45:27 <roshi> if it has to be done by last thursday, we're running out of time
16:45:47 <tflink> roshi: don't be silly ... it has to be done by last friday
16:45:51 <tflink> there's more time than that
16:45:53 <adamw> #action roshi and cmurf to work with Atomic WG to try and get the PRD and wiki space updated, so we can then think about how to derive requirements and test processes from it
16:45:54 <roshi> #action roshi to bring up the PRD at the next Atomic meeting
16:46:02 <roshi> lol
16:46:05 <adamw> i don't think i chaired you ;P
16:46:10 <adamw> #chair roshi tflink cmurf
16:46:10 <zodbot> Current chairs: adamw cmurf roshi tflink
16:46:56 <cmurf> the brilliance of it being due by last thursday is the long it takes the more time we have!
16:47:07 <cmurf> s/long/longer
16:47:24 <tflink> we're never farther behind than a couple of days
16:47:26 <tflink> I like it!
16:48:00 <adamw> #info there's lots of Interesting Times coming up, we have our finger on the pulse and we're skating to where the puck's going to be and are otherwise up to date on our cliches
16:48:35 <adamw> #topic Fedora 26 Changes and status
16:48:44 <adamw> soo, i wasn't necessarily expecting to get here and am not prepared. :P
16:48:56 <adamw> but so far as Fedora 26 status goes, here is the status: it's busted.
16:49:21 <roshi> so far so good then!
16:50:06 <coremodule> QE to the rescue!
16:50:12 <cmurf> Perfect. Do it again.
16:50:20 <adamw> at present the install root creation tasks all fail with some kind of error between RPM and NSS, so we haven't even had a rawhide compose for like four days
16:50:49 <adamw> the one we had right before that was nerfed by https://bugzilla.redhat.com/show_bug.cgi?id=1419946 , which we hope should be fixed in the next compose that actually works...
16:51:00 <adamw> so, we're definitely still on the blocker treadmill :P
16:51:15 <adamw> has anyone been keeping up with the latest F26 Changes? i have to admit i haven't followed them all
16:51:59 <adamw> #info Rawhide compose is currently broken due to some kind of issue between RPM and NSS
16:52:07 <adamw> #action adamw to follow up with releng on compose failure
16:53:31 <nirik> we think it might be nss. ;)
16:54:03 <adamw> seems plausible ;)
16:54:08 <adamw> it got a version bump, right?
16:54:12 <adamw> anyhoo
16:55:20 * adamw hastily eyeballing the Change list
16:55:22 <nirik> "Disable TLS 1.3, following the upstream change"
16:55:39 <adamw> nirik: that seems to have been pingponged in recent builds if you look back a bit, too
16:55:47 <nirik> yeah.
16:55:54 <adamw> cmurf: have you been following the Anaconda LVM RAID change? that seems significant
16:56:21 <cmurf> yes
16:57:12 <adamw> ok. are we just going to have to run a bunch of RAID installs to check it out?
16:57:12 <cmurf> it's not yet approved pending questions about whether GNOME shell notifications still happen when there's a faulty device
16:57:15 <adamw> ah k
16:57:31 <cmurf> if approved we'd see UI changes in the installer
16:57:42 <cmurf> (that's a guess, just because I'm not sure how it's avoidable)
16:58:16 <cmurf> I sent an email to test@
16:58:26 <cmurf> It's not urgent
16:58:28 <adamw> okay
16:58:33 <adamw> sorry, i must've missed it :)
16:58:49 <adamw> #info Anaconda LVM RAID Change seems significant, if approved we'll have to run some install tests to check it out
16:59:23 <cmurf> yeah the bot tests will probably break if there are ui changes
16:59:32 <brunowolff> Would it be bad to do dnf upgrades to f26 right now? I was looking at going to f26 when the rebuild packages got tagged into rawhide.
16:59:55 <adamw> brunowolff: my rawhide is mostly working, but i haven't updated to post-rebuild packages yet.
17:00:13 <adamw> the agetty thing doesn't make it unusable and you can get around it with enforcing=0 of course.
17:00:45 <brunowolff> Thanks, that's good enough for me. You guys were scaring me.
17:04:04 <adamw> there's the Container minimal image Change too, i've just sent a mail reply about that one
17:04:21 <adamw> not sure if the Kerberos credential cache one would be a big thing
17:04:33 <adamw> #action adamw to look into Kerberos credential cache Change and see if it's significant
17:04:51 <adamw> anything else on f26 status?
17:05:31 * sumantrom awaits to conduct test days with the early branch of f26
17:06:03 * coremodule is standing by to help sumantrom on test days.
17:06:42 <adamw> aaaaalllllllrighty
17:06:49 <adamw> #topic Open floor
17:07:06 <adamw> so, any other business? we're a bit over time, but i don't think anyone else is taking this slot and i scheduled the blocker meeting for top of the next hour
17:07:12 <adamw> (please yell if you're waiting for us to finish!)
17:07:38 <roshi> I don't have anything really, but I'm sure things will pop up as we get deeper into this rabbit hole
17:07:51 <adamw> ooh, i missed a cliche. good job.
17:08:02 <adamw> ah, one thing: kparal set up a QA subplanet for us
17:08:02 <cmurf> I'm so glad we have a gap, I need coffee, pushups, and food.
17:08:14 <adamw> http://fedoraplanet.org/quality/
17:08:16 * roshi is with cmurf on that one
17:08:25 <sumantrom> kparal++
17:08:25 <zodbot> sumantrom: Karma for kparal changed to 3 (for the f25 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
17:08:26 <adamw> it'd be great if everyone could subscribe their blog to it and remember to blog now and again
17:08:29 <cmurf> Oh and I need to kidnap Trudeau...
17:08:48 <adamw> instructions at https://fedoraproject.org/wiki/Planet#Joining_a_sub-planet
17:09:06 * roshi should really blog sometime
17:09:10 <adamw> #info we now have a QA subplanet: http://fedoraplanet.org/quality/ . please enrol your blog if you have one: https://fedoraproject.org/wiki/Planet#Joining_a_sub-planet
17:09:12 <adamw> yay blogging!
17:09:16 <roshi> it's on my list
17:09:25 <sumantrom> same here!
17:11:27 <adamw> sounds like that's about all we've got
17:13:01 * adamw sets the fuse
17:13:07 <adamw> thanks for coming out, everyone!
17:14:10 <marc84> adamw: thanks
17:14:33 <adamw> see you in 45 mins for the blocker review meeting in #fedora-blocker-review
17:14:35 <adamw> #endmeeting