16:01:05 #startmeeting Fedora QA meeting 16:01:05 Meeting started Mon Jan 16 16:01:05 2012 UTC. The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:05 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:01:11 #meetingname fedora-qa 16:01:11 The meeting name has been set to 'fedora-qa' 16:01:17 #topic roll call 16:01:23 morning folks, who's around? 16:01:24 * tflink is present 16:01:27 * mkrizek present 16:01:31 * pschindl is here 16:01:41 * j_dulaney is recovering from meeting adamw and tflink 16:01:45 * kparal 16:02:07 * tflink didn't realize that we were that overwhelming 16:03:00 i recommend liquor 16:03:38 * j_dulaney has class this afternoon 16:03:49 then it'll work for both! 16:03:54 #topic previous meeting follow-up 16:04:09 alrighty, we had just the one action item last week: 16:04:22 "kparal to follow up with spot on the legal questions around the marvin logo" 16:04:33 looks like you made a change there, kparal? 16:04:37 I'll cover that in the AutoQA update section 16:05:01 okay, fair enough 16:05:31 then let's move right along to... 16:05:41 #topic FUDCon report / follow-up 16:06:11 looks like three out of four who were at fudcon are here, so i just thought we could do a quick brain dump on how it went and any cool stuff we picked up there 16:06:17 I think j_dulaney wants to go first ;) 16:06:24 I do? 16:06:34 *dear in headlights look* 16:07:05 i can go if you like, I was just kidding 16:07:17 Actually, my most pertinent thing would go in the AutoQA section 16:07:35 However, we do need to start thinking about release criteria for ARM 16:08:05 And start working on a test plan; if it goes primary it would be good to have it all ready to go 16:08:13 yeah, ARM was a big focus of the event 16:08:22 it seems like there's a general movement towards making ARM a primary arch 16:08:55 Obviously, most of our criteria should transfer over, but we should research what needs to be changed/added/removed 16:09:06 so, jlaska was looking at secondary arch release criteria just before he was moved into the cloud, so we could go and pull what he had and start there 16:09:16 indeed 16:09:34 anyone interested in taking that, or should I? 16:09:47 * j_dulaney might could look into it 16:10:00 Since it looks like I'll be getting an ARM device to play with 16:10:51 yeah, that was the other neat thing: the arm mafia were giving away very low-cost arm devices at their talk, and j_dulaney got one 16:11:03 tflink and I should be be able to get them via RH channels, so we should have a decent base of devices for testing 16:11:18 * tflink also has a panda for armv7 testing 16:11:31 for me the obvious possible differences for arm will be a) we may care less about desktop functionality (though it sounds like that should actually work okay anyway) and b) deployment is not via anaconda 16:11:33 tflink is lucky 16:11:48 adamw: Indeed 16:11:49 so we'll need to figure how to adapt the validation process to arm deployment 16:11:51 For now, anyway 16:11:54 yeah 16:12:16 if that just means 'check images for platforms X, Y and Z exist and actually deploy correctly' then, hey, it's easier than anaconda testing! 16:12:49 Indeed 16:13:07 another question might end up being HW platforms - are we ok with just armv6 and armv7 coverage or do we want to go into more deatail (beagle, panda, rasberry, pogo etc.) 16:13:16 Although we may want to include deployment instructions in specific test cases for common paltforms 16:13:27 and at least make sure we have v5, v6 and v7 coverage (not sure how important that is, though) 16:14:06 i think that's a good q to ask the team, how do they look at it 16:14:11 what are their goals in that area 16:14:26 We might want to ask the ARM Sig what hardware is explicitly covered 16:14:38 there is a list of supported platforms somewhere 16:15:05 * j_dulaney will liaise with the ARM folks 16:15:36 I know that trimslice and a couple others aren't supported 16:15:37 okay 16:16:15 not yet, anyways 16:16:42 so yeah, as a summary: arm is clearly now a Big Important Thing, we will want to do some best-effort testing on it and try and work it into the validation process somehow for F17, and it may well be going primary arch at F18, the way things look, so we'll want to be prepared for that. 16:17:38 Indeed 16:17:45 j_dulaney: i tell you what, i'll file a ticket to set up a release criteria and validation test process for arm (somehow, no mechanism specified), and cc us all on it 16:18:03 Alright 16:18:07 #action adamw to file trac ticket(s) for getting arm into the criteria / validation process somehow 16:18:18 #info as a summary: arm is clearly now a Big Important Thing, we will want to do some best-effort testing on it and try and work it into the validation process somehow for F17, and it may well be going primary arch at F18, the way things look, so we'll want to be prepared for that. 16:18:25 okay, aside from arm, who has cool stuff from fudcon to report? 16:18:46 adamw: do we want to talk about anaconda plans? 16:19:00 yes we do! 16:19:05 that's a v. big one 16:19:13 * tflink has a huge todo list from fudcon 16:19:15 it is now official that the new anaconda UI will *not* go into F17 16:19:22 * adamw cracks champagne 16:19:38 you are all authorized to reduce your gin consumption by 50% 16:19:42 * j_dulaney is at school, and it's a dry campus 16:19:47 heh 16:19:52 adamw: not waiting until it's in fedora for the champagne? 16:20:01 tflink: no, it not being in f17 is cause enough :) 16:20:19 #info the new anaconda UI will not go into F17 16:20:28 the plans haven't been 100% finalized, but the idea will be to start testing the UI w/ rawhide 16:20:38 clumens says there simply is not enough time to get it done by f17 feature freeze 16:20:50 hopefully, we'll figure out a way to have a somewhat-stable rawhide tree to build test install images from 16:21:08 so the broad plan is for it to land in F18, and have it go into rawhide some time pretty early after the F17 branch - long before F17 is actually released 16:21:13 after F17 branch, I mean 16:21:24 thanks to no frozen rawhide, we will then have a good long time to test it before F18 goes out 16:21:46 before the point where the new UI is actually deployable to rawhide, anaconda team are planning to provide test images 16:22:06 we're a bit handwavy right now on where those will land and how they'll be announced, but they're going to exist. probably. 16:22:55 at that point the new UI will be very incomplete and may not even be capable of performing an actual install, but the anaconda team will give us guidance on what bits of it should actually be testable, and what testing would be valuable to them. 16:23:18 once it's vaguely capable of installing a system, it goes into rawhide, and we can test from rawhide just like we always can. 16:24:16 #info broad plan is for anaconda team to push new UI into Rawhide targeted for F18 as soon as it is capable of performing installs (can be prior to F17 release thanks to NFR). prior to that point, they will start providing test images as soon as external testing will actually be of use to them, with instructions. 16:24:53 does that about right, tflink / j_dulaney? did I miss anything? 16:24:58 sound about right* 16:25:12 that sounds about right to me 16:25:23 Looks good here 16:25:29 cool 16:25:42 any comments/questions/rotten vegetables, anyone else? 16:26:08 * j_dulaney throws a rotten tomato at adamw 16:26:20 j_dulaney and I spent some time trying to work out some details on testing AutoQA 16:26:20 There's your poker payment 16:26:42 tflink: I was going to bring that up in the AutoQA section 16:26:54 j_dulaney: k, I'll hold off 16:27:14 i meant specifically on the anaconda stuff 16:27:22 i still have a few other things to brain dump on fudcon in general 16:27:34 * adamw checks to see if the audience have dozed off 16:27:40 * j_dulaney is awake 16:27:59 we need to make a list of the bugs that were dismissed as "will be fixed in new UI" for F17 16:28:01 j_dulaney: oh, and I covered our gamblin' losses, never fear :) 16:28:15 make a list for F17 for the bugs in F16, rather 16:28:19 casey wound up ahead $18, ivan was up $2, and the rest of us were broke 16:28:40 tflink: yup, that's a good one 16:28:40 LOL 16:28:54 so we handwaved away a few moderately serious bugs in 16 with 'oh, new UI in 17 anyway' 16:28:55 the other anaconda stuff I have is details around how we might handle new UI testing 16:29:06 like bootloader UI stuff 16:29:29 anaconda team understands we will need to fix those in old UI now it's staying in 17, but we should probably co-ordinate on tracking them all down and making sure none are missed 16:29:47 i can probably do that since i was probably most involved in that stuff for 16 anyway 16:30:10 it would be nice to formalize the text mode stuff, too 16:30:17 * tflink realizes that it won't change for F17 16:30:21 #action adamw to co-ordinate with bcl on making sure all important old-UI bugs discovered in F16 validation get fixed in the old UI for F17 16:30:46 tflink: we do have text UI test cases. we just process-hacked them away from being blockers for f16. 16:31:20 adamw: I was thinking more along the lines of what should be fixed for F17 16:31:40 Def. the bootloader stuff 16:32:00 I forgot to write down the exact issue, but there was one in particular that sounded as if it would be fixed 16:32:10 would/should 16:33:30 Sorry guys 16:33:36 did you want to expand on 'how we might handle new UI testing'? 16:33:57 eventually, yes but I don't think we need to right now 16:34:16 okay 16:34:27 sorry, i just dropped for a minute there, flipped the airplane mode switch on my laptop :) 16:34:45 so, small things I had: lmacken tells us bodhi 2.0 is REALLY GOING TO GET DONE some time soon 16:34:56 again? 16:35:03 LOL 16:35:04 he's been pulled off it for a while, but he's pretty confident he and a new hire will be able to work on it more or less full-time in the near future 16:35:05 yup, again :) 16:35:26 do we have the name of the new hire? 16:35:31 he told me, but I forgot. 16:35:37 I need names so that I can bother them :) 16:35:45 not a complete greenhorn, but someone who was already active, apparently 16:35:50 * tflink makes note to contact lmacken re: potential changes to the bodhi interface that could affect AutoQA 16:35:50 great 16:36:04 tflink: good one 16:36:18 #info Bodhi 2.0 really should be coming Real Soon Now, according to lmacken 16:36:41 #action tflink to contact lmacken re co-ordination between bodhi 2.0 and autoqa 16:36:42 * kparal notes: usual 6 months ahead 16:36:50 i hope not :) 16:37:20 so vaguely related to that, i got notting to clarify the exact current practical/theoretical status of proventesters 16:37:49 the theoretical state is, as agreed back at a fesco meeting in november, pt karma should now count for no more than anyone else's: i.e updates require +1 from anyone pre-beta, +2 from anyone post-beta 16:38:05 the practical state is, pt karma is still needed, as no-one changed bodhi 16:38:23 fesco did file a ticket but it didn't actually get assigned to anyone, i suggested notting assign it directly to lmacken and then kick his ass 16:38:37 * nirik notes new hire is Ralph Bean / irc: threebean (unless I am totally mistaken... do confirm, etc, etc) 16:38:48 nirik: thanks! 16:39:05 adamw: lmacken said he would bang that change out in the next few days. ;) 16:39:13 if you want to keep an eye on when the change is actually implemented, see https://fedorahosted.org/bodhi/ticket/653 16:39:40 #info for proventesters, in theory pt karma should now be the same as anyone else's, but in practice it's still special: follow https://fedorahosted.org/bodhi/ticket/653 to see when it gets changed 16:39:51 nirik: that sounds correct 16:39:59 thanks! 16:40:34 i did ask notting (as a proxy for fesco) to try and make sure that any messaging is 'proventesters is temporarily on hiatus while we figure out a way to use the system more efficiently', not 'proventesters is dead 16:40:43 because i do think once we get bodhi 2.0, we can do something useful with pt status 16:40:57 or at least try :) 16:41:05 Indeed 16:41:33 let's see, what else... 16:41:39 oh yeah, i was hoping andre would be here 16:41:45 robatino: around? 16:41:51 yes 16:42:17 coolbeans 16:42:21 so, this one sounded up your alley 16:42:53 pjones has most of the bits available so we could write a test, as part of the whole 'release sanity' stuff, to check that built images are actually bootable 16:43:18 he has a tool which just looks at the el torito stuff from an image and makes sure it ought to be both bios and efi bootable 16:43:45 in a way it's almost not worth the trouble for TC/RC images, but it shouldn't hard to implement, and we could always use it on nightlies too 16:43:51 sound like something you'd be interested in? 16:44:15 not sure i'm qualified for that 16:44:22 okay 16:44:32 anyone else interested? 16:45:09 * j_dulaney has enough 16:45:16 To keep me busy, that is 16:45:22 yep 16:45:22 would it be worth putting in with rats_sanity or mediakit_sanity? 16:45:27 in autoqa? 16:45:37 yeah 16:45:39 possibly? you'd maybe be more qualified to answer that than I :) 16:46:04 okay, to keep the meeting shortish, let's start a big messy mail thread about it instead 16:46:07 * tflink passes the buck to kparal or hongqing 16:46:07 if you have some links, i could take a look 16:46:21 i'll start up a thread with everyone in it 16:46:40 let's discuss in autoqa-devel 16:46:45 #info pjones would like us to use a tool he wrote to automatically check if images are actually bootable 16:47:19 #action adamw to get a discussion going with pjones/kparal/hongqing/robatino to decide how we want to do that 16:47:37 it doesn't have to be autoqa necessarily - the sanity stuff we do on TCs/RCs at present isn't, right? robatino just fires off a script 16:47:55 no, i do that manually, actually 16:48:34 oh, okay. 16:48:36 shows what I know! 16:48:39 anyway, we'll figure it out. 16:49:46 the other little thing I had, cwickert and red_alert noted it was kind of hard to find and navigate between validation results pages, so i filed a ticket for myself to look at improving that 16:50:30 ticket is https://fedorahosted.org/fedora-qa/ticket/273 if anyone else wants to contribute ideas / fixes 16:50:44 adamw: thanks, I'll try to provide some feedback in the ticket later 16:51:07 ooh yeah, and bcl has a new tool for building live images 16:51:14 which i'd vaguely heard about but hadn't looked at very hard 16:51:25 it's on my list of stuff to look into 16:51:29 yeah, me too 16:51:45 it isn't clear whether the official livecd build process will switch to the new tool or not, though 16:51:50 s/if/when 16:52:00 sounds like it will require some small changes to the kickstarts, but a potentially bigger thing is it builds the images in virt, not mock 16:52:08 virt or native 16:52:08 so you need a virt-capable host, and the koji integration may change 16:52:11 right 16:52:19 you don't need a virt-capable host 16:52:29 well, okay. my impression is he kinda preferred the virt route, though 16:52:41 true, the native implementation was a compromise 16:53:04 but my understanding was that if the official liveimages were built with that tool, they would be native instead of virt 16:53:07 it sounded to me like he wants people who actually build live images to start trying to use it, and that'll help him decide how soon it ought to be used as the default 16:53:24 so he just wants us to poke it and see what falls off, really 16:54:22 ahh, now i know why i kinda missed the initial announce and discussion of this 16:54:27 it got filtered to anaconda-devel which i don't always read... 16:55:02 I have been meaning to test it as well but haven't gotten to it. It is also getting very later to commit to this switch for F17. 16:55:14 yeah, it kinda is. 16:55:20 so you can find the initial thread about the tool at https://www.redhat.com/archives/anaconda-devel-list/2011-December/msg00117.html 16:55:46 if you use livecd-creator at the moment to build live images, go read it and the linked info, and give livemedia-creator a shot, and poke bcl about how it's broken 16:56:15 #info bcl has a new potential livecd-creator replacement in development, called livemedia-creator. announcement at https://www.redhat.com/archives/anaconda-devel-list/2011-December/msg00117.html . go use it and tell him how it's broken 16:56:40 adamw: the new way is going to need major rewrite to use in koji, so is not likely for a few releases 16:56:46 dgilmore: ok. 16:57:07 adamw: and not all platforms really support virt 16:57:20 so i hope that they come up with a non virt way to do it 16:57:29 #info koji integration will be a major change so livemedia-creator will not be used to create the official images for a while 16:57:29 as ive asked and pointed out to them many times 16:57:32 heh 16:57:50 well, as tflink noted, seems like native build is now possible or will be 16:57:58 anyway, not necessary to discuss it now, i just wanted to flag up the tool 16:58:15 anything else I'm missing? man, fudcon braindumps are hughe 16:58:17 also huge 16:58:58 oh, we gave a few talks, the 'get into qa' hackfest was kind of a bust probably as it was the morning after fudpub, and the anaconda / qa hackfest resulted in the outcomes discussed above 16:59:27 I don't think you're missing anything huge but I'm still going through all my notes 16:59:31 okay 16:59:46 let's move on before we all drop dead, anything big we can always discuss next week / on the list 16:59:52 splitsville! 17:00:50 #topic autoqa update 17:01:02 how're we doing on autoqa front? kparal, tflink, take it way 17:01:04 adamw: i just tagged the mass rebuild into the buildroot 17:01:06 #chair kparal tflink 17:01:06 Current chairs: adamw kparal tflink 17:01:09 dgilmore: yay 17:01:14 I have a few notes 17:01:25 1. New AutoQA logo (again). The Marvin logo had copyright issues, changed to Sherlock Holmes: 17:01:31 #link https://fedoraproject.org/wiki/File:Autoqa_logo.png 17:01:40 nice. 17:01:58 spot said we really couldn't use Marvin because of copyright 17:02:05 yeah, that's what i figured :) 17:02:26 #info autoqa logo changed to a sherlock holmes-y image, to avoid legal issues with the first image proposed 17:02:26 also citing spot 'as sad as it makes me to say it (I'm also a big HHGTTG fan), but Marvin would certainly agree that it was inevitable' 17:02:33 I asked him about it in person; his response was that it was a very bad idea 17:02:57 current logo shouldn't pose a problem, holmes copyright expired in 2000 17:03:02 says wikipedia 17:03:28 anyway, if you have something better, just make yourself heard 17:03:42 alright, more news 17:03:45 2. Reported host of Bodhi/Autotest issues due to which some of our tests currently crashes. I am devoted to pestering relevant people until they fix it (you can't hide from me, lmacken!) 17:04:03 unfortunately currently a lot of our tests crashes 17:04:11 due to incorrectly tagged updates, etc 17:04:38 We sort of discussed this a bit at FUDCon 17:04:39 3. autoqa-results (again) contains only staging results. I will re-enable production results after we deploy 0.8. 17:04:49 tflink more than me 17:05:14 j_dulaney: tflink: some more details you want to share? 17:05:31 about 2) 17:05:43 tflink: The topic is incorrect bodhi incorrect tagging discussed at FUDCon 17:05:51 You were more involved than I 17:06:02 * j_dulaney must go AFK for a minute 17:06:14 I'll go on and we can come back to it 17:06:23 4. mkrizek found a probable cause why our koji/bodhi watchers get stuck endlessly from time to time. The patch should be on its way. 17:06:48 koji library apparently uses infinite timeout on network sockets 17:07:02 and sometimes the server just doesn't respond 17:07:04 it seems 17:07:20 5. depcheck is no longer scheduled as noarch test in some circumstances. This will be fixed on production only after we deploy 0.8. 17:07:27 j_dulaney: incorrect tagging? I don't remember that discussion 17:07:40 that was just a minor bugfix, but you can see that sometimes in Bodhi in the comments section 17:07:49 6. hongqing finished rats_install test and asked for review. I'm currently trying to get an extra bare metal machine to test it out. 17:08:00 that's all from me 17:08:10 did I forget something? 17:08:22 or do you have comments to some of that? 17:09:21 * kparal supposes the loud silence means 'no' 17:09:35 tflink: I thought you were talking to lmacken about it 17:09:50 Maybe it was me, and my memory is exceptionally fuzzy 17:10:14 * tflink is still organizing thoughts and action items from FUDCon, haven't been thinking as much about AutoQA explicitly the last several days 17:10:37 j_dulaney: I talked to him a little about some other stuff but not tagging 17:11:01 kparal: sounds like good fixin' 17:11:09 tflink: Maybe it was *me* and I can't remember? Anyway, I can't remember what was said 17:11:14 so, we were gonna put fudcon autoqa stuff in here too? 17:11:24 adamw: Indeed 17:11:46 tflink, can you kick this off, since I am now not sure of my sanity 17:12:26 j_dulaney and I spent some time to figure out what would be needed for an initial self-test setup for AutoQA 17:13:00 j_dulaney: welcome to where the rest of us have been for months ;) 17:13:00 we started looking at what kinds of interfaces and data would be needed to effectively isolate AutoQA for repeatable testing 17:13:34 most of it was pretty high level but we have a start 17:13:42 * j_dulaney has a list that will be pushed to the ml later this day, once it's sorted out and made presentable 17:14:00 I think that the next step will be to go through and implement something 17:14:12 * j_dulaney also has audio of the whole discussion, including tflink's chalkboard scrapings 17:14:23 mmm....chalk 17:14:32 iirc, we were going to start with the scheduler and add more once we get that hammered out 17:14:57 the idea will be to generate pseudo-rpms with rpmfluff and metadata to fill out the bodhi and koji interfaces 17:15:32 we want to minimize the amount of data we need to configure, so part of this will be determining some sort of configuration format and test specification 17:16:19 oh, there was a request to change the name of mock_fedorainfra and I'll be doing that before too long 17:16:30 it needs some refactoring and bugfixes, anywyas 17:17:09 Essentially, we've got an exceptionally rough idea where to start 17:17:16 so it's almost done! 17:17:40 adamw: you have an odd definition of "almost done" 17:18:03 We've figured out what the hardest bit will be, and sort of figured out how to do it 17:18:12 tflink: :P 17:18:42 Anyway, that about covers it 17:19:07 coolbeans. 17:19:12 anything else on autoqa? 17:19:30 nothing concrete 17:20:03 had some interesting conversations with lmr about autotest features and openqa but that would likely be better to summarize elsewhere 17:20:10 alright 17:21:39 #topic open floor 17:21:55 anything for open floor? anything come up while we were away playing poker^H^H^H^H^H^H^H^H^H^H^H working hard? 17:23:38 Suger testing 17:24:15 what about it? 17:24:20 Somehow, I missed the automated email trac gave out, but probinson poked me about a bunch of changes coming up in Sugar 17:24:45 A test day has been requested; I haven't had the chance to drop it in the schedule 17:24:50 ah, okay 17:24:58 yeah, you should have quite a bundle of test day requests pending 17:25:06 https://fedorahosted.org/fedora-qa/ticket/270 17:25:11 But, also: 17:25:18 http://wiki.sugarlabs.org/go/0.96/Roadmap 17:25:19 #info there will be some significant sugar changes this cycle 17:25:29 They're migrating to GTK3 17:25:57 #info sugar is migrating to GTK+ 3, details and roadmap at http://wiki.sugarlabs.org/go/0.96/Roadmap 17:26:17 As a result, there is significant rewrites of much of the activities for this cycle; core rewrites will wait for next cycle 17:26:39 cool. 17:27:01 That's it for now; I'll ping the ML once I have more info 17:27:05 thanks 17:27:19 anything else before we all escape meeting hell? :) 17:28:11 don't think so 17:28:12 quick, propose something 17:28:27 * j_dulaney proposes we adjourn 17:28:30 * adamw sets fuse quickly 17:28:36 seconded with feeling! 17:28:49 thanks for coming to the Great FUDCon Braindump, all 17:29:11 #endmeeting