fedora-qa-20090311

--- Log opened Wed Mar 11 13:10:49 2009
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA meeting | init13:10
wwoodsQA meeting time13:11
wwoodsf13, jlaska, poelcat, adamw, viking_ice: ping13:11
* jlaska 13:11
wwoods(and anyone else who's here, say hi for the log)13:11
adamwoh, bonus!13:11
f13oh damn, xchat-gnome can't do empty /me13:11
adamwhi13:12
-!- giallu [n=giallu@fedora/giallu] has quit [Read error: 110 (Connection timed out)]13:12
* viking_ice *13:12
* f13 <blank>13:12
wwoodsf13: try /me [ctrl-B]13:12
viking_ice /me [ctrl-B]13:12
wwoodsif you can emit the mIRC bold etc. control characters13:12
* wwoods 13:12
f13wwoods: no such luck.13:12
viking_icehahaha13:12
* f13 13:12
f13ah there we go, added a space.13:12
wwoodsthen doing the bold character - followed by no text - gets you an empty line13:12
wwoodsold, stupid IRC trick13:12
* wwoods is an IRC mute!13:12
wwoods13:12
wwoodsahem. anyway13:13
f13see, irssi just let me do empty /me s13:13
wwoodsso, first things first13:13
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA meeting | review last week's minutes13:13
wwoodshttp://wwoods.fedorapeople.org/fedora-qa/fedora-qa-20090304.log.html13:13
wwoods[jlaska+adamw] - update on semantic packaging13:14
jlaskaI haven't gotten far here ... just got mediawiki installed and configured this morning :(13:14
wwoodswell, that's a start13:15
wwoodskeep us posted13:15
adamwnothing from me - i am waiting on jlaska to get it working, then i will do the package13:15
jlaskastill need to wedge in the semantic extension13:15
wwoods[jlaska] - review upstream xorg test suite status (in addition to piglit) ... for use during upcoming Xorg driver specific test day13:15
jlaskawill do ... let's raise it again for next week13:15
-!- inode0 [n=inode0@fedora/inode0] has joined #fedora-meeting13:16
wwoodssure13:16
wwoods[wwoods] - discuss with f13 + jakub a method for identifying similar gcc4.4 optimization bugs13:16
jlaskaI'm hoping to have someone on the RHEL X testing team join on the test day to provide more information on the upstream test suite13:16
wwoodshaven't done anything on that - AFAIK we haven't seen other miscompilations, luckily13:16
f13I haven't talked with jakub about that yet13:16
f13I've been too focused on other beta emergencies13:17
wwoodsdidn't we get a fixed gcc for that issue?13:17
adamwi have a contribution there - i talked with ben skeggs a bit about rendercheck13:17
jlaskaadamw: ah right, good update (re: rendercheck)13:17
adamwsorry, that's for x test stuff. ben suggested providing it as a scratch build for the nouveau test day13:17
adamwhe didn't think it made sense as a regular package in the distro13:17
wwoodsadamw: oh, sure, that's fine13:17
adamwi thought maybe it did, so we could use it for other tests in future13:17
wwoodsscratch build or a private repo13:17
adamwi'm waiting for his reply on that13:18
wwoodseither would work. it's dead simple to set up a private repo on your fedora people page13:18
adamwbut we'll have a package for it in some form :) for the nouveau test day.13:18
wwoodswell, I mean, that's a *public* repo13:18
wwoodsjust not part of the official repos13:18
* adamw hates single-purpose third-party repos with a passion, but will stow it for a more appropriate forum13:18
wwoodsheh13:18
wwoodsoh, nobody likes them, but they're useful for things that are well and truly temporary13:19
adamwsorry, anyhow, that's the status there - continue!13:19
wwoodsif it's not temporary, get it in the dang repo13:19
wwoodserr. the dang distro13:19
-!- inode0 [n=inode0@fedora/inode0] has quit ["Leaving."]13:19
-!- inode0 [n=inode0@fedora/inode0] has joined #fedora-meeting13:20
wwoodsoffhand, does anyone have a bug ID for the example gcc miscompiles13:20
wwoodsso we have some reference for exactly *which* gcc bug we're asking jakub about13:20
wwoodsreading the gcc changelog there's.. a significant amount of stuff going on there13:21
wwoodshttp://koji.fedoraproject.org/koji/buildinfo?buildID=9364313:21
adamwi can pull out the reports, er, just a sec...13:21
adamwdamn, no, can't find 'em13:22
-!- mcepl [n=mcepl@49-117-207-85.strcechy.adsl-llu.static.bluetone.cz] has joined #fedora-meeting13:22
-!- drago01 [n=linux@chello062178124130.3.13.univie.teleweb.at] has joined #fedora-meeting13:23
adamwi note from the vte changelog, though, that the workaround was removed after an update to gcc13:23
adamwso presumably for that one, whatever was wrong in gcc has been fixed13:23
adamwhttp://bugzilla.redhat.com/show_bug.cgi?id=487844 is the nspr (firefox) bug13:24
buggbotBug 487844: medium, medium, ---, kengert, ASSIGNED, Aliasing and Optimizations cause numeric conversion errors13:24
nirikthe webkit one was I think similar or the same issue... bug 48811213:25
buggbotBug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=488112 medium, low, ---, peter, MODIFIED, WebKit crashes in midori and layout issues13:25
wwoodsalright, we'll try to follow up on that with jakub once there's some spare time13:25
-!- inode0 [n=inode0@fedora/inode0] has quit ["Leaving."]13:26
wwoodsadamw: you were going to talk to fcami about a radeon test day.. and intel too?13:26
-!- inode0 [n=inode0@fedora/inode0] has joined #fedora-meeting13:26
adamwok, we've been talking to the devs13:27
adamwwe're aiming to do both13:27
adamwintel test day we're hoping to do tomorrow13:27
wwoodswhee!13:27
adamwi have a bunch of awesome stuff from kristian on testing13:27
adamwso for the rest of today i'm going to get that into the wiki page and then jlaska and i are going to try and promote the day13:27
wwoodsoh that's excellent13:28
-!- petreu| [n=peter@p3EE3E7CF.dip.t-dialin.net] has quit ["( www.nnscript.de :: NoNameScript 4.02 :: www.XLhost.de )"]13:28
adamwwe're a bit concerned that we're on very short notice for what could be a really good test day, so if we don't get a good response tomorrow, we might do it again on another day.13:28
wwoodsoh, can you make sure that there's a link to the info (or the test day page itself) from the feature page13:28
viking_icelet the test span over the weekend13:28
wwoodsso that people who are checking out a feature will get that info13:28
viking_iceand just gather results13:28
adamwviking_ice: i'm always careful to have the pages say we can take test results at any time, now13:29
adamwviking_ice: but the main point of the 'day' is to have the relevant people in irc13:29
adamwviking_ice: and we can't really ask the intel devs to hang in -qa irc troubleshooting for a whole weekend :\13:29
adamwviking_ice: but yep, i'll definitely emphasize that results from the tests are useful at any time13:29
viking_icewhen doing hw testing we need as much audience as possible13:29
adamwwwoods: good point, i'll do that for intel then go back and do it for the other features we've done test days on13:29
viking_iceno they can be available for a single day + the result and related discussion can be performed on the test list...13:30
wwoodshad some red hat desktop folks poking at crash catcher (now ABRT) and having no idea how to test it13:30
viking_icethen when the devs have time again they can chip onto those threats13:30
wwoodsand there was no indication from the feature page that we'd done a test day or anything at all13:30
adamwviking_ice: mostly for async stuff we want to keep it in bugzilla so it's tracked, as usual13:31
adamwviking_ice: that's why i've tried to write all the plans for test days that i've organized so that negative results are filed to bugzilla13:32
adamwwwoods: yep, you're right, i'll definitely add those.13:32
-!- mcepl [n=mcepl@49-117-207-85.strcechy.adsl-llu.static.bluetone.cz] has left #fedora-meeting []13:33
wwoodsso, that's last week covered13:34
wwoodslet's talk about this week's bonanza of fun13:35
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA meeting | beta looming13:35
wwoodswe have frozen for beta, as I understand it13:35
wwoodswe're also trying to switch to larger keys for signing, and.. there's a squashfs problem in anaconda or something?13:35
adamwexecutive summary: shit's broken13:36
wwoodsf13: wanna help with an update here? I've been working on other stuff so I'm not on top of it13:36
wwoods"I can't see anything. Are we boned?"  "Yeah, we're boned."13:36
f13heh13:36
f13ok, so there is a serious squashfs bug in rawhide kernels13:36
f13this is particularly bad for installer, since stage2 is squashfs.13:37
f13this means that just about /anything/ can go wrong in your install due to corrupted filesystem13:37
jlaskabug#48969813:37
buggbotBug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=489698 high, low, ---, kernel-maint, NEW, squashfs zlib corruption13:37
f13the good news is that the problem has likely been found and a new kernel is forthcoming13:37
f13http://koji.fedoraproject.org/koji/buildinfo?buildID=9381613:37
f13lots of people can reproduce so once that kernel is done I'm going to create some images with it and pass them around to see if that fixes things13:38
-!- Nirmal [n=npathak@nat/redhat-in/x-cefe5e8f174c76e7] has quit ["Leaving"]13:38
f13other things going on13:38
wwoodswell that doesn't sound so bad13:39
wwoodsheh13:39
f13package signing, a big chunk per day13:39
f13I'm breaking up the signing into many chuncks so that we can soak up the sync hits across multiple days rather than landing 7K srpms worth of signed packages all at once13:39
f13so mirrors are going to be a bit slow picking up the signed versions of rpms, but there should be not so much new build churn due to freeze13:40
-!- j-rod [n=j-rod@static-72-93-233-3.bstnma.fios.verizon.net] has joined #fedora-meeting13:40
f13There is the anaconda storage rewrite as well13:40
f13that's getting fixes daily so we'll likely see a new anaconda every day13:40
adamwis that looking like it's going to be in reasonable shape for teh beta?13:41
f13hard to say13:42
f13its working well for a good chunk of functionality13:42
jlaskaif no other changes were happening in rawhide, I think it would need a week to shake out remaining bugs13:42
f13I think the work now is on the more ancillary support, like resizing, software raid, etc...13:42
f13they've got about a week13:43
jlaskaright ... I worry that issues like the zlib bug keep us from getting meaningful feedback to the anaconda-devel crew on their daily builds13:43
f13well.13:44
-!- gregdek [n=gdk@nat/redhat/x-2a0b12a7f03c1d5b] has joined #fedora-meeting13:44
f13that's why we're only taking fixes in the freeze13:44
f13tested fixes13:44
f13that's why the freeze exists13:44
jlaskaI know ... not criticizing ... just pointing out potential areas that might make the 1 week remaining take longer13:45
f13sure13:46
wwoodswell, we'll see how it goes13:47
wwoodswhat would we consider the line between "wonky" and "too broken to release"?13:47
-!- dkovalsk [n=dkovalsk@nat/redhat/x-15386df07300d0a6] has quit ["leaving"]13:47
adamw"help! it's on fire!"13:48
wwoodsheh13:48
viking_iceit has to be installable and or upgradeable and come with network connection13:48
wwoodsheh. sorry. specifically with respect to the storage rewrite13:48
adamwviking_ice: i think the question is specifically what level of installer breakage will we accept...13:48
jlaskathe default selections at the parttype screen can't fail13:49
adamwif it's broken for sixteen-disk RAID-10 arrays on LVM on iSCSI on tuesdays in March that's probably OK, for instance13:49
wwoodswhat if no custom layouts are possible? 13:49
f13fail13:49
wwoodscustom layouts OK, unless you use mdraid? 13:49
f13I think mdraid should work too13:49
wwoodscustom layouts OK, except no dmraid? 13:49
f13that's a pretty common setup13:49
f13dmraid is less critical, but still pretty important13:49
wwoodswhat mdraid levels do we require? what if mdraid is mostly OK, but fails if you try to set a system up with hot spares or something?13:50
wwoodswhat will we block the release for?13:50
jlaskaclumens suggested raising this issue as well ... we might want to pull in the devel folks for what their take is13:50
viking_iceyup13:50
f13wwoods: hrm, tough call.  I think we'll just have to make the decision when the time comes13:51
-!- notting [n=notting@redhat/notting] has joined #fedora-meeting13:51
f13raid0,1,5 should work, raid10,10,50,05 probably not as important13:51
jlaskawe do have release criteria established ... do these need to be updated? https://fedoraproject.org/wiki/QA/ReleaseCriteria13:51
wwoodsthose are intended to be general release guidelines - intentionally vague to allow interpretation and flexibility13:52
jlaskaor reflect additional milestones other than final?13:52
wwoodswe are now flexibly interpreting13:52
jlaskayeah, I've always been uncomfortable with that flexibility ... but I recognize it's difficult to think of every possibility ahead of time13:53
wwoodswell, feel free to write up a set of guidelines that are specific to the storage portion of anaconda13:53
wwoodswrite up a test plan and put that in the release guidelines13:53
jlaskaheh, I knew the patches accepted was coming  :)13:54
wwoodsI don't think anyone would argue with that, really13:54
wwoodsbut first we have to determine what things we want to require. so here we are.13:54
wwoodsshort answer: simple mdraid setups (0,1,5), LVM, LUKS, and.. what dmraid possibilities are there?13:55
f130,1,513:55
f13basically its hard to pontificate about what bugs we'd accept or reject when we don't know what bugs are there13:56
f13far easier on a bug by bug level to say Yes or No13:56
wwoodsand those can be set up in anaconda? or do they get set up in the BIOS and anaconda just needs to recognize/handle 'em?13:56
j-rodmdraid looked like it was going to work before zlib attacked13:56
f13once the bugs are known.13:56
adamwyeah, i'm with f13 on it, i've always been rather more on the reactive side13:56
jlaskapossibly, it makes it tough for others to discern whether their issue is a blocker or not13:56
f13wwoods: anaconda just needs to recognize them.13:56
viking_iceWell my opinion is the installation networking yum and upgrading should be rock solid ( in all possible way and form before FINAL if not release blocker )13:56
j-roddmraid gets set up in the bios, ana... yeah13:56
adamwit's quite hard to write a list of exactly what theoretical bugs would be critical and what wouldn't, and get it right13:57
jlaskaagreed13:57
wwoodsdisagree. that's the entire point of writing a spec.13:57
adamwviking_ice: we're talking beta here, not GA13:57
f13jlaska: thats why they should default to putting it on the F11Beta tracker.  We'll take care of Yes (leaving it) or No (punting it to F11<SomethingElse>)13:57
wwoodsyou write up a broad list of the things you expect to work13:57
jlaskaI don't think it's hard to however to publish the criteria we'd use to evaluate bugs13:57
jlaskaf13: is that new ... so everything goes to blocker and you guys are the gate?13:58
wwoodsI really don't know what we're willing to leave broken13:58
f13jlaska: I don't think its new per se, just not well communicated.13:58
* jlaska nods13:58
wwoodsI, for one, do not care if dmraid is broken13:58
f13wwoods: I don't either.  Its judgement calls.13:58
wwoodsI have a violent and irrational hatred of it13:58
f13Does it effect a large number of users?13:58
f13Is there a fix possible in the next day or so?13:58
jlaskaf13 bingo ... # of affected users13:58
wwoodsright, there's still going to be judgement calls involved13:58
f13Is there an easy work around?13:58
wwoodsbut I don't know if we actually care about, say, dmraid 10. I know we care about LVM.13:59
-!- Kick__ [n=karsten@p54A3F127.dip.t-dialin.net] has joined #fedora-meeting13:59
wwoodswell, whatever, we're spinning13:59
wwoodsstorage bugs go to F11Beta and we'll exercise judgement as they come in13:59
* nirik wonders if qa will be done so fesco could take the room for it's special session... or shall we go to #fedora-meeting-1 ?13:59
jlaskagood discussion ... yeah I don't see us concluding anything in this meeting13:59
-!- nphilipp [n=nils@nat/redhat/x-d49a6fefd1f2cab3] has joined #fedora-meeting13:59
* jlaska has a meeting conflict now 14:00
wwoodsthere's no obvious storage features that we just plain don't care about14:00
wwoodsnor any that we will not tolerate even the slightest problem14:00
wwoodsso all the shades of gray will be sorted later14:00
wwoodsnirik: didn't know that was going on - give us just a sec14:01
wwoodsI think we can adjourn to #fedora-qa in a moment14:01
nirikno worries... we can move too, it's not a big deal.14:01
wwoodsbut first: is there anything that we absolutely must discuss for the Public Record right now?14:01
-!- kasal [n=kasal@49-117-207-85.strcechy.adsl-llu.static.bluetone.cz] has joined #fedora-meeting14:01
f13I don't think so14:02
viking_iceme neither..14:02
jlaskaseveral test days coming .... one tomorrow, one next Tuesday ...14:02
wwoodsokay then - let's call it done14:02
jlaskacheck your local listings ... https://fedoraproject.org/wiki/QA/Test_Days/F1114:02
wwoodstuesday? nice. make sure to mail -test-list etc.14:02
-!- drago01 [n=linux@chello062178124130.3.13.univie.teleweb.at] has quit [Remote closed the connection]14:03
wwoodsone week 'til beta, right? whee!14:03
jlaskasure ... that's what I do for all the events14:03
wwoodsthanks all14:03
jlaskathanks wwoods14:03
--- Log closed Wed Mar 11 14:03:17 2009

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!