--- Log opened Wed Aug 06 11:04:56 2008
-!- vwbusguy [n=scott@99-153-230-117.lightspeed.iplsin.sbcglobal.net] has quit [Nick collision from services.]11:05
-!- hanthana [n=hanthana@] has joined #fedora-meeting11:05
wwoodsjlaska, jds2001, f13, poelcat: ping11:05
* jlaska 11:05
-!- vwbusguy- [n=scott@99-153-230-117.lightspeed.iplsin.sbcglobal.net] has joined #Fedora-meeting11:05
wwoodsqa meeting starting, say "hi" for the logs11:05
f13wwoods: pong11:06
jds2001sorry i've been less than reliable lately - $DAYJOB has insane deadlines, and add to that my surgeon likes to schedle appts on Wednesday at 10:1511:06
f13"hi" for the logs11:06
jds2001"hi" for the logs11:06
* jds2001 just feels like a big copycat11:06
-!- No5251 [n=No5251@p50990d02.dip0.t-ipconnect.de] has quit ["Wenn du den Fnord nicht siehst, kann er dich auch nicht essen."]11:07
wwoodsit's OK11:07
wwoodsso, as usual, I don't have a big agenda11:08
-!- John5342 [n=John5342@user-54470a1c.wfd82a.dsl.pol.co.uk] has joined #fedora-meeting11:08
wwoodsfirst things first: Alpha's out. yay!11:08
jlaskajds2001: no worries ... "it" happens ;)11:08
-!- hanthana_ [n=hanthana@] has joined #fedora-meeting11:08
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA Meeting | F10Alpha release/review11:08
wwoodswe need to think about whether we want to do a non-blocking freeze for F11Alpha11:08
wwoods(we're not going to decide that right now, though)11:09
f13f10 alpha wasn't our best of releases :/11:09
wwoodsone lesson learned, though: we need to start composing and testing images *before* the "freeze"11:09
-!- J5 [n=quintice@] has joined #fedora-meeting11:09
wwoodsso we have some advance warning of Big Bads11:09
jds2001or do a blocking freeze, which folks would probably cry out about :/11:10
wwoodseither way11:10
-!- AndreasR [n=zeus@urz-ar224.urz.unibas.ch] has quit ["Verlassend"]11:10
jlaskathe confusion for me is the process by which we free the bits for folks to play leading up to a milestone ... f13 summed it up rather nicely during the releng meeting11:10
wwoodswe still need to do some early composes and get the word out to devs to get bugs fixed *BEFORE* the freeze11:10
f13I did a number of composes leading up to the freeze, but they were just on my workstation, not a lot of attention was paid to them (although andy poked at them whenever they were there)11:11
jds2001what I think is that it'd be useful to move reducto into Fedora infra. (but that'ss just me - there may be reasons i dont know that it's internal)11:12
wwoodsnot possible11:12
jlaskait's not clear what needs to be done on those internal composes in order for them to be considered worthy of the community11:12
wwoodsreducto isn't magical11:12
* jds2001 knows that :)11:12
wwoodsit's just a machine that happens to be on jesse's desk that has a bunch of disk space11:12
f13jds2001: I now have a few machines in the fedora infrastructure to do composes on11:13
f13but the problem with that is it greatly increases the lag time between compose and test11:13
jds2001how do you mean?11:13
f13with reducto in BOS, I can compose, test it in kvm/xen on reducto, get it into the hands of the anaconda folks for immediate use, as well as andy for immediate qa work11:13
jds2001network issues?11:13
f13if I compose in PHX, we're talking a good chunk of time to sync down the compose bits to test them.11:14
wwoodsthe reason we build the images in the office is because we then have instant access to the images to smoketest 'em11:14
f13700~K/s is the transfer rate out of PHX to BOS11:14
wwoodsbuilding them in the fedora infrastructure means *nobody* gets instant smoketest ability11:14
jds2001yeah that sucks11:14
wwoodsthat's not helpful11:14
f13however one can do multile composes11:14
f13multiple that is11:14
f13and nothing stopped people from composing on their own against rawhide11:15
wwoodscomposing in the office *and* in the fedora infrastructure simultaneously would solve the problem, but11:15
wwoodsyeah, exactly. anyone with sufficient disk can compose just as well on their own11:15
jlaskaI don't have a problem doing one compose ... but I Don't feel like we have visibility on what is expected to happen on that compose before we provide community access11:16
wwoodsdefine "community"11:16
jlaskafor example, how does f13 know that the reducto compose is "good" enough to push to mirrors?11:16
wwoodsfull Fedora community access to the pre-release compose attempts is not useful11:16
wwoodsthe question you're asking is: "when is it ready for release?"11:17
jlaskawwoods: I'm not asking about that right now ...11:17
jlaskaright ... how does f13 get confidence in the bits so that he can go through the mirror process11:17
wwoodsQA team testing of the candidate bits11:17
f13jlaska: the bits I compose gets rsynced to various locations for other QA folks to look at11:18
jlaskaf13: internal or external?11:18
f13we verify needed fixes, do cursory smoke testing for regressions11:18
f13jlaska: could be both if we had more active QA people outside RH11:18
f13jlaska: this is where composing in both locatiosn could help11:18
f13compose in phx and anybody can get to it, compose in BOS and we get to it immediately11:19
* jds2001 just does his own composes.11:19
f13and there is that too (:11:19
-!- kital [n=Joerg_Si@fedora/kital] has joined #fedora-meeting11:19
* jlaska imagining a wiki page where we outline a summary of the milestone release process11:19
jds2001since i mirror locally, it is faster for me to do a compose than download some other bits.11:19
jlaskaand we can then stub out links to "here's the link to roll your own tree" (as some folks do), and "here's the list of tests we perform prior to releasing"11:20
jds2001we have the release tests11:20
jds2001and i've blogged how to do a compose.11:20
jds2001(porbably needs to move someplace more foraml)11:20
wwoodsthe tests we perform prior to release.. that's the installation test plan11:20
jlaskadefinitely, the informatoin is out there ... just think it would be worthwhile to organize11:20
f13jds2001: the trick is to coordinate when I toss extra packages into my compose11:20
f13jds2001: freshly built packages after teh latest rawhide11:21
jlaskawwoods: we don't perform the installation test plan prior to a milestone release, do we?11:21
wwoodsyes. we do.11:21
jds2001f13: yeah, and that's what sucks :)11:21
jlaskawwoods: https://fedoraproject.org/wiki/QA/TestResults/Fedora10Install/Alpha ?11:21
zodbot<http://tinyurl.com/6bjv7g> (at fedoraproject.org)11:21
f13jds2001: yeah, those on IRC when that happens knows what's going on11:21
wwoodsjlaska: yep.11:21
jlaskawwoods: I'm thinking milestone in the sense of Alpha, Beta, PR, RC11:21
jlaskanot just GA11:21
jlaskaif we executed the full plan prior, that page would be red||yellow||green11:22
jlaska(not white)11:22
jds2001jlaska: you can compose reawhide and run through those wehenver11:22
wwoodswe never have time to execute the full plan as a prerequisite to release.11:22
f13jlaska: there isn't enough time/people to execute teh entire test fully11:22
* jds2001 cant type11:22
wwoodswe do as much as we can and releng/qa make a judgement call based on the current results.11:22
f13jlaska: we do the best we can with what we got, and make decisions based on coverage, milestone, and schedule.11:22
wwoodsthere's always more testing that can be done.11:22
jlaskaright, so should we bother specifying all this stuff if no one runs it?11:22
f13yes, we should11:22
wwoods"not finished" != "no one runs it"11:22
f13because if we don't, it'll definitely never get done.11:23
f13jlaska: also, many of these tests get ran after the release too, to fill in coverate11:23
wwoodstesting is never, and will never, be Complete11:23
jlaskathat seems like something we'd want to address11:23
f13unlike RHEL, we can't stand in the road and hold up traffic until we've searched every car11:23
wwoodssomething we'd want to address? you mean.. "we should test more"?11:23
wwoodsI fully agree.11:23
f13we have a set time to do as much testing as we possibly can, and eventually make a judgement call as to ship or delay11:24
wwoodsI think we could rewrite the test plan to better match the actual priorities11:24
wwoodse.g. we do not care if ext2 is tested for non-final releases11:25
jlaskahow about extending our FUDCon rawhide acceptance discussion into the milestone acceptance realm?11:25
-!- hanthana [n=hanthana@] has quit [Read error: 113 (No route to host)]11:25
f13jlaska: that's essentially what we've been trying to do since fudcon11:26
-!- mether [n=sundaram@nat/redhat-in/x-44d38762873e3721] has joined #fedora-meeting11:26
f13at fudcon we sort of handwaved at "alpha acceptance" and "beta acceptance"11:26
f13since then we've been trying to define what those mean11:26
wwoodsthat's what we were trying to do with testopia, before it went dark11:26
jlaskavery true11:26
jds2001btw, extjs 2.2 was released on monday11:26
f13also, it's hard to state what we care/don't care about in a static way, since it changes based on deadline.11:26
jds2001with the FLOSS exception11:26
jds2001need spot to opine on that :)11:26
spotdo not want11:27
f13at teh beginning of the freeze, we certainly care about every test, but 5 minutes before the go/nogo deadline, we suddenly care a lot less about a lot of the tests11:27
f13and which we care less about also depends on what other testing had been done.11:27
jlaskaf13: right ... we start with a baseline ... and test change from that point on11:27
spotRH Legal says stay far far far away from ExtJS. Exception, or no exception.11:27
* jlaska trying to find FUDCon wiki pages ...11:28
jds2001spot: understood :(11:28
f13jlaska: https://fedoraproject.org/wiki/JohnPoelstra/ImproveRawhideF1011:29
zodbot<http://tinyurl.com/5eem6f> (at fedoraproject.org)11:29
wwoodsso basically: no testopia ever11:29
jlaskaf13: that's it ... thanks!11:29
* wwoods wonders what dmalcolm is up to 11:29
-!- mbacovsk [n=mbacovsk@nat/redhat/x-0d5e8063327bce09] has quit [Read error: 110 (Connection timed out)]11:29
f13wwoods: not unless you port testopia to some other js api11:29
sticksterf13: dmalcolm tells me that's not an unattainable goal11:30
f13too many negatives!11:30
f13"it's possible" yes or no?11:30
* stickster being positive!11:30
sticksterf13: dmalcolm tells me that's an attainable goal! Yay!11:31
sticksterf13: Proabably a couple months for some contractor... now where did I leave my spares lying around?11:31
-!- sdziallas [n=sebastia@p57A2D752.dip.t-dialin.net] has quit ["Ex-Chat"]11:31
jlaskaf13: perfect ... https://fedoraproject.org/wiki/JohnPoelstra/ImproveRawhideF10#rawhide_as_a_MAJOR_MILESTONE11:31
zodbot<http://tinyurl.com/5eem6f> (at fedoraproject.org)11:31
f13jlaska: it's that step 11 that's the big one (:11:32
wwoodsand we were working on a testopia plan11:32
wwoodswith priorities etc.11:33
wwoodsthere's some tests that we just cannot release without results11:33
f13although we do have code to do steps 1,5,6, but not many of the others.11:33
jlaskayup :)11:33
wwoodsand there's some tests that we can take or leave11:33
-!- giallu [n=giallu@fedora/giallu] has quit [Read error: 110 (Connection timed out)]11:33
jlaskawe can certainly work up the testing milestone criteria w/o testopia11:33
wwoodsppc netboot? ext2 raid 1+0 on LVM? yeah, okay, if we've got time11:33
jlaskajust in terms of outlining test areas, or even discrete cases11:34
wwoodsright, I'm just saying we've got a start at it already11:34
jlaskawwoods: certainly ... so replacing testopia with ${mgmt_tool_formerly_known_as_testopia}11:35
-!- No5251 [n=No5251@p50990d02.dip0.t-ipconnect.de] has joined #fedora-meeting11:35
f13yeah, we probably need to come up with a couple different classifications11:36
wwoodswe need to pull some data outta the testopia db but yeah11:36
f13mandatory, default, optional (for lack of a better naming scheme)11:36
f13and then tag the tests with these labels11:36
wwoodswe typically call those "tier1 / 2 / 3" 11:36
f13ok, sure11:36
wwoodsexcept it gets funkier when you're talking about different milestones11:36
wwoodsso, yeah11:36
f13we also have different arches to deal with11:36
f13so if DVD install (tier1) was tested on i386, its' not manditory for x86_64 or ppc11:37
wwoods(DVD install being separate from DVD boot)11:37
f13I think that we definitely won't ship without tier1 tests being complete (pass) on at least one of the arches11:38
jlaskadoes it matter whether MILESTONE is alpha, beta, PreviewRelease, or ReleaseCandidate?11:38
f13that moves the rest of the arches to tier211:38
f13jlaska: at this time, I don't think so.11:38
wwoodstechnically RC is not a milestone11:38
wwoodsRCs are the test composes we do before GA11:38
jlaskawwoods: so should a different process apply for RC?11:38
f13jlaska: if we magically got a bunch more QA resources and know ahead of time we have the man power to blow through all the tests on all the arches, I'd change my mind.11:38
jlaskaf13: nod11:39
wwoodsno different than the process we apply to the test composes for alpha/beta/pr11:39
f13jlaska: in my mind, RC's are just attempts at getting to the milestone, not a milestone itself11:39
wwoodsexcept that we make a much stronger effort to get community access to the GA RCs11:39
-!- mdomsch [n=Matt_Dom@cpe-70-124-62-55.austin.res.rr.com] has joined #fedora-meeting11:39
wwoodsf13: exactly11:39
f13which I think can easily be done by using the dual-compose method11:39
jlaskaso in Fedora ... a release candidate won't become the GA?  We will always re-compose a GA tree?11:39
wwoodshuh? release candidates become GA. where did you get that idea?11:40
jlaskasoftware quality assurance ;)11:40
f13the only thing we do to a RC to make it GA is gpg sign the sha1sums11:40
f13and filter in teh Live images11:41
f13although jigdo metadata may not show up for the first couple RCs11:41
jlaskaokay so no respinning or anything, you take an existing spin and sign it?11:41
-!- sdziallas [n=sebastia@p57A2D752.dip.t-dialin.net] has joined #fedora-meeting11:41
f13it's a REAL release candidate11:41
jlaskawwoods: sorry, was thinking about how I'm used to release candidates and how they relate to the final product11:41
wwoodsas in it is a candidate to become the release11:41
wwoodsyeah that method is total horseshit11:42
wwoodsif you call something a "release candidate" without being fairly confident that you could end up releasing those exact bits11:42
wwoodsyou are lying to yourself and your users11:42
jlaskaI don't disagree11:42
f13I laugh at schedules that have marked "RC1, RC2, RC3"11:42
wwoodsthis is why we have a Preview Release11:43
jlaskaright ... planning RC[2-9] is an interesting discussion between QA and PM11:43
wwoodsthat fills the role that other people call the "Release candidate"11:43
-!- Frankly3D [n=frank@] has quit [Remote closed the connection]11:43
jlaskawwoods: I see ... so you were distinquishing between the public RC (called PreviewRelease), and the internal trees f13 builds leading up to the official GA, is that accurate?11:44
f13we also don't re-compose because software/hardware is fickle, something could go wrong in the compose process, so I want to sign/publish the bits we actually tested11:44
wwoodsjlaska: yes.11:44
wwoodsthe internal trees that get built before each milestone are for QA-team consumption only11:45
wwoodswe're resolved to make an effort to compose in public *and* in the RH offices simultaneously11:45
-!- Frankly3D [n=frank@] has joined #fedora-meeting11:45
wwoodsto allow more of the QA team to work on testing the RCs11:45
wwoodsand we actually put some of the F9GA RCs up as torrents, if memory serves11:46
jlaskaokay, thanks ... this was very helpful to hilight how the bits are freed.  And the different needs each milestone (internal and public) has11:47
f13at teh end of the day though, it's the bits that we've tested most that will be the ones that go out11:47
f13those typically are the ones done on reducto11:47
f13we'll just use the bits in PHX as a pre-sync content11:47
-!- John5342 [n=John5342@user-54470a1c.wfd82a.dsl.pol.co.uk] has left #fedora-meeting []11:49
wwoodsso, yeah - F10Alpha, Not The Best Release Process Ever (but not that bad)11:50
wwoodswhen we talked in the releng meeting monday it seemed like we were inclined to make the Alpha a blocking freeze 11:51
wwoodseither way we should compose images a week or two before the freeze to get early warning of potential issues11:51
f13that gets more people paying attention to the bits we're actually composing11:51
f13wwoods: i think we have scheduled weekly snapshots after alpha11:52
wwoodsreally? you willing to do that?11:52
f13beta freeze isn't that far away though, given the length of alpha11:52
f13wwoods: snapshots are 'best attempt'11:52
-!- cmpahar [n=cmpahar@] has quit ["Leaving"]11:53
wwoodsso yeah, if you'll make images we'll be better for it, I think11:54
wwoodsokay, anything else relating to the alpha?11:57
* wwoods getting distracted11:57
f13we could still get people to fill in test cases after the fact11:57
f13which will help us to focus on things in the snapshots and beta pre-testing11:57
wwoodsf13: very true11:58
f13and by 'could' I mean 'should'.11:58
f13and by 'we' I mean 'you'  (:11:58
wwoodsfair 'nuff11:58
wwoodsand I have been doing some testing there11:58
wwoodsso I'll continue with that and try to encourage others.11:58
wwoodsokay, so11:59
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA Meeting | New Bugzilla!11:59
wwoodswoo, the Great Grand Bugzilla Upgrade is upon us11:59
wwoodsthings are a little shaky but coming together11:59
wwoodsI'm updating python-bugzilla to understand the BZ 3.2 methods11:59
wwoodssending reports to the bugzilla team where applicable11:59
wwoodsjds2001: I'll be interested to hear how it's affecting you and the other triagers12:00
f13and I'm constantly trying to find where the UI moved things12:00
f13summary box is in a /really/ bad place12:00
wwoodsyeah, it's kinda weird12:00
f13if we thought bugzilla was scary to users before, it's down right pants filling now12:00
wwoodswhich is why it's important to update python-bugzilla12:01
wwoodsand try to make bug filing automatic and easy when possible12:01
-!- rnorwood_ [n=rnorwood@adsl-144-134-115.rmo.bellsouth.net] has quit [Read error: 110 (Connection timed out)]12:02
-!- rnorwood_ [n=rnorwood@nat/redhat/x-d2ef9bf71088a4b4] has joined #fedora-meeting12:02
wwoodsbut yeah, it's kind of intimidating 12:03
wwoodsit'd be nice if the whole big form block at the top was hideable12:03
f13and the yukon is kind of cold12:03
-!- fbijlsma [n=fbijlsma@] has quit [Read error: 110 (Connection timed out)]12:04
f13I think a number of the canned queries for triage don't work anymore12:04
wwoodsoh, yeah, someone mentioned that my canned rawhide/f8 queries are broken12:04
wwoodsgotta look into that12:05
f13 The search named rawhide-thisweek  does not exist.12:05
f13 The search named F8-thisweek  does not exist.12:05
f13thankfully bugz.fedoraproject.org didn't break, I use that a ton12:05
wwoodsWORKSFORME - must be a perms thing12:05
wwoodsthat's cool - that's based on python-bugzilla12:05
f13odd that you'd have perms for it and I wouldn't12:05
wwoodsso it looks like I did *something* right12:05
wwoodsf13: I created those queries12:05
wwoodsso they work for me.12:06
wwoodsI'll figure that out as soon as I get a chance12:06
wwoodsso yeah - looks like I'm spending most of this week on bugzilla stuff12:06
wwoodswhich is fine. bugzilla is pretty dang important.12:06
-!- sdziallas [n=sebastia@p57A2D752.dip.t-dialin.net] has quit ["Ex-Chat"]12:07
-!- mdomsch [n=Matt_Dom@cpe-70-124-62-55.austin.res.rr.com] has quit ["Leaving"]12:08
wwoodsSoon I'll bug people about stuff they'd like to see in python-bugzilla12:08
wwoodsas I start working toward a python-bugzilla 1.0 API12:08
wwoodsI know clumens wanted Bug.attach(file), for instance12:08
wwoodsso think about that, if you use python-bugzilla at all12:08
wwoodslunch looms large overhead12:09
f13I haven't used it since I did the massive bugfiling exercise a while back12:09
wwoodsso I'll try to wrap up quick here12:09
-!- wwoods changed the topic of #fedora-meeting to: Fedora QA Meeting | Feature testing12:09
wwoodsfeatures! woo! 12:09
wwoodswe need to make sure their test plans don't suck.12:09
wwoodsanyone here a feature owner?12:09
f13not I12:09
* wwoods is listed on a couple. doh.12:09
f13I'm listed, but I don't own it12:10
wwoodsjust call me Mr. Bit Off More Than He Can Chew12:10
wwoodsanyway, if you happen to be looking at a Feature page, see if you can figure out how to test that feature in Rawhide12:10
wwoodsif you can't.. ask a bunch of questions in the Talk page12:10
wwoodsif the owner(s) won't/can't answer them, their feature can't be tested. if we can't test it, we can't really use it12:11
-!- sdziallas [n=sebastia@p57A2D752.dip.t-dialin.net] has joined #fedora-meeting12:12
wwoodsif we can't test it, it should get dropped.12:12
abadger1999f13: It did but maploin pointed out it was broken and I fixed it.12:12
-!- giallu [n=giallu@fedora/giallu] has joined #fedora-meeting12:13
wwoodsSo yeah. I don't want it to sound like a threat, but if feature owners won't help with the test plans we will have to drop their features.12:13
wwoodsso there that is.12:14
wwoodsyer on notice, feature owners.12:14
wwoodsthat's about all I can think of right now. anyone else got anything?12:14
f13a hungry tummy12:14
f13I may be moving in the middle of beta freeze cycle12:15
jds2001wwoods: i've been looking at them and reacting accordingly at FESCo meetings12:15
f13although I may also be camping at somebody's house until beta is out the door too12:15
jds2001f13: moving not allowed :)12:15
f13jds2001: moving across the country even.12:15
jds2001back to Seattle?12:15
wwoodsWEST SIIIIIDE12:15
-!- Sonar_Gal [n=Andrea@fedora/SonarGal] has quit ["Leaving"]12:16
abadger1999You going to move to PHX? That's one way to cut the network latency for composes :-)12:16
wwoodsahem. yes. noted.12:16
f13jds2001: seattleish12:16
wwoodsman I should find somewhere cool to move to and work remotely.12:16
wwoodstoo bad I'm so lazy and we're just getting the house the way we like it12:16
wwoodsalso Leah refuses to move any further north. we're already in *North* Carolina after all. 12:17
skvidalwwoods: where is she from?12:17
f13for the lolz12:17
wwoodsskvidal: gulf coast mississippi12:17
skvidalwwoods: move to asheville, nc12:17
skvidalthat's no further north but it is less hot than raleigh12:18
* jds2001 has a friend for asheville12:18
jds2001ahh, hot is good12:18
jds2001cold == bad12:18
wwoodsyeah she already complains about the winters being too cold and dry here in Raleigh. heh.12:18
wwoodsanyway, I'm going wildly off-topic, I assume the meeting is thus concluded12:18
f13ok, I require foods12:18
jds2001heh what are tehy like in Raleigh?12:19
wwoodsI need lunch to be able to focus12:19
-!- spoleeba [n=one@fedora/Jef] has joined #fedora-meeting12:19
f13I'm going to likely be AFK the rest of today, and PTO thurs/fri12:19
f13(house hunting)12:19
wwoodsjds2001: I'd say there's one week per year of *real* winter - sub-freezing temps, an inch or two of snow12:19
wwoodsthe rest of it is, like, low 40s and grayish12:20
wwoodscome March 1 it's 70 degrees and sunny, like someone flipped a switch12:20
skvidalwwoods: this year we had a proper spring here12:20
skvidalit gradually warmed up - march was 50s and 60s as it should be12:20
skvidaland we got good rains then12:21
skvidalthe farmer's at the market here were talking about it being a proper spring12:21
skvidaland not having to irrigate much12:21
wwoodsso that's what it's supposed to be like? good to know12:22
wwoodsokay. lunch. thanks for your time, folks12:22
--- Log closed Wed Aug 06 12:22:45 2008

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