fedora-meeting
LOGS

16:00:07 <jlaska> #startmeeting Fedora QA Meeting
16:00:24 <jlaska> #topic roll call
16:00:39 * Oxf13 
16:00:40 * kparal greets everyone
16:00:55 * dpravec rolls
16:02:04 <botox> fabulous
16:02:29 * tk009 
16:02:58 <jlaska_> sorry, having some connectivity issues on my end
16:03:56 <jlaska> dfdfdfdf
16:04:13 <jlaska> #chair jlaska_
16:04:45 <jlaska> adamw: is around also ... and I suspect wwoods is lurking
16:05:04 * wwoods lurk lurk
16:05:42 <adamw> back
16:05:42 <jlaska> okay, so unless any suggested changes, we can walk through the proposed agenda (https://www.redhat.com/archives/fedora-test-list/2009-August/msg00255.html)
16:05:54 <jlaska> #topic Follow-up from last week
16:06:08 <adamw> quick question so i (hopefully) don't miss it - when's the alpha go/no-go meeting? i never saw a time
16:06:27 <jlaska> I'm told it's at the rel-eng meeting @ 18:00 UTC
16:06:38 <adamw> ok thanks
16:06:53 <jlaska> so follow-up from last week is quick'n'easy ... there were no recorded action items
16:07:08 <jlaska> was there anything from last week folks would like to point out?
16:08:12 <jlaska> okay, silence is golden :)
16:08:15 <jlaska> let's dive in ...
16:08:32 <jlaska> #topic F-12-Alpha Test Compose
16:08:54 <jlaska> Just a quick thank you to all who pitched in, or tried to pitch in ... and filed bugs nonetheless
16:09:15 <jlaska> the F12Alpha blocker bug was fairly busy last week, and so was the test results wiki
16:10:03 <jlaska> Once we have more information on an Alpha candidate, look for some mail from Liam soliciting testing for that compose
16:10:11 <Oxf13> it still feels like we're doing the testing too late and nearly guaranteeing a slip due to finally finding things after we've frozen.
16:10:22 <jlaska> hrmm, I don't feel that way
16:10:27 <Oxf13> but in this case, probably becuase we were late in delivering the test compose
16:10:32 <Oxf13> because of other blockers.
16:10:33 <jlaska> yeah
16:10:45 <jlaska> we had early testing before the test compose
16:12:00 <jlaska> so a few issues I wanted to ask for feedback from the team
16:12:05 <jlaska> # alt.fedoraproject.org download speeds?
16:12:14 <kparal> i have a comment on that
16:12:20 <adamw> it was fine for me, but i grabbed pretty early
16:12:24 <adamw> went through at 500-600k/sec
16:12:29 <jlaska> infrastructure has some data, but is curious if we haev any painpoints to share
16:12:45 <kparal> i have complained that the speeds are low last week, but i have found out that we have only 20Mbit/s for the whole RH branch in Brno
16:13:07 <kparal> therefore it is most certainly caused by our slow line, not fedora infrastructure
16:13:09 <jlaska> aah, so that was the limiting factorf for you?
16:13:26 <jlaska> kparal: Interesting, I didn't know that
16:13:30 <kparal> yes, 200 people for 20Mbit = 20kB/s for each? :)
16:13:40 <jlaska> use it sparingly :)
16:14:10 <jlaska> I'll ping Liam and He Rui as well to see if they experienced issues, but I haven't heard anything so far
16:14:27 <kparal> therefore i take back my complaints on alt.fedoraproject.org site :)
16:14:48 <jlaska> kparal: hehe, okay ... you can take your comment card out of the box :)
16:15:14 <jlaska> okay ... so we'll keep tabs on this for future downloads
16:15:31 <jlaska> mmcgrath noted they have some data available and can scrub the logs for more if needed
16:15:54 <jlaska> but so far, if it's failing or slow ... I haven't heard so yet
16:16:05 <jlaska> okay next ...
16:16:06 <jlaska> # Rawhide still contains anaconda-12.7
16:16:20 <jlaska> Oxf13 mentioned this in #anaconda just before the meeting
16:16:22 <Oxf13> this is because no anaconda beyond 12.7 actually worked
16:16:27 <Oxf13> not tag worthy
16:16:36 <jlaska> but I was curious why rawhide still had the older test compose anaconda
16:16:39 <Oxf13> (not that 12.7 worked all that great)
16:17:15 <Oxf13> jlaska: anaconda is critical path, and none of the proposed anaconda updates passed my critical path checking.
16:17:18 <jlaska> Oxf13: it seems like we could at least knock out verification of some of the MODIFIED bugs though?
16:17:25 <Oxf13> not really
16:17:30 <Oxf13> couldn't get through an install
16:18:15 <Oxf13> things beyond 12.7 were definitely /worse/ than 12.7 itself
16:18:50 <jlaska> Oxf13: thanks for the info, I think that answers it
16:19:06 <dpravec> hooah, now i see you again :)
16:19:14 <jlaska> Oxf13: so you're doing some basic sanity of anaconda before tagging it for rawhide
16:19:19 <Oxf13> yeah
16:19:27 <Oxf13> I've played this game before
16:19:28 <dpravec> there were some connection problem with some freenode server
16:19:35 <jlaska> dpravec: I see that :)
16:19:37 <adamw> sorry to ask a messy question, but - why does this process seem so...difficult?
16:19:50 <adamw> what's going on in anaconda to cause these huge fluctuations?
16:20:15 <dpravec> i will try to inquire on this
16:20:23 <dpravec> anaconda team is located here in brq
16:20:33 <dpravec> they mostly told me that they have lots of dependencies
16:20:46 <Oxf13> adamw: there is a lot of code that can go wrong, and it is difficult to test anaconda fully with each and every patch
16:20:46 <dpravec> and its almost impossible to maintain useable anaconda between releases
16:20:53 <Oxf13> adamw: so the possibility of regressions is high.
16:20:57 <dpravec> so they would like just prepare working anakonda when release is near
16:21:03 <Oxf13> adamw: rawhide is basically their change test bed
16:21:07 <dpravec> which is a dangerous style i do not like
16:21:37 <dpravec> i would love to have working and tested anaconda each week at least
16:21:49 <dpravec> but it will be hard to get
16:21:53 * Viking-Ice sneaks in..
16:21:58 <jlaska> Viking-Ice: welcome
16:22:12 <adamw> at least during the alpha freeze, though, this shouldn't be happening
16:22:19 <Oxf13> dpravec: we're moving toward what you don't like, only making working anaconda when we're ready to branch for a release
16:22:40 <adamw> we were in the alpha freeze when 12.7 came out, right? so none of the dependencies should have changed, so builds past 12.7 _really_ shouldn't be getting more broken
16:22:52 <dpravec> Oxf13: i know... :)
16:23:07 * adamw just figured out who oxf13 is. /me needs more sleep
16:23:10 <Oxf13> adamw: yes, and a fix for a blocker introduced another blocker.
16:23:24 <adamw> ah.
16:23:26 <Oxf13> subtle code has a way of breaking, subtly
16:23:35 <Oxf13> actually, that may not be fully true
16:23:39 <dpravec> that needs more unittesting :)
16:23:42 * adamw wonders if it's therefore too subtle for its own good :)
16:24:04 <Oxf13> I think one of the current blockers was in fact introduced after we froze for alpha, and wasn't related to an alpha blocker bug
16:24:14 <jlaska> that's true
16:24:20 <Oxf13> however, just because Fedora freezes for alpha, doesn't mean that Anaconda upstream freezes as well
16:24:39 <Oxf13> we could get pissy here and demand that the only changes brought in from upstream during freeze are for blocker bugs
16:24:49 <Oxf13> not sure how well that will go over
16:25:28 <jlaska> we can certainly look for some ideas around this if this continues to be the pain point
16:25:30 <adamw> well i see two paths there in your subtle use of the word 'upstream' ;)
16:25:49 <dpravec> for me it looks like anaconda is a big problem for regular daily testing of rawhide
16:25:52 <Oxf13> adamw: the anaconda upstream is outside of Fedora's "control"
16:25:58 <Oxf13> just like rpm upstream is
16:26:01 <Oxf13> and kernel upstream
16:26:08 <adamw> if we treat anaconda as an 'upstream project' in that way, we should have an onion layer at the level of anaconda packaging for fedora which would act as the firebreak here
16:26:23 <Oxf13> adamw: yes, I agree
16:26:37 <dpravec> agreed
16:26:43 <adamw> because if we say anaconda's upstream and we can't control it but we just throw whatever they do straight into the distro, that's a problem.
16:27:00 <Oxf13> adamw: I don't think anybody is disagreeing with you
16:27:01 <adamw> perhaps at freeze point we should go from taking anaconda version upgrades wholesale to backporting specific patches
16:27:11 <wwoods> who's this "we"
16:27:17 <adamw> yeah, sorry, i was half way into that sentence before everyone agreed with me =). that doesn't often happen :P
16:27:21 <adamw> wwoods: fedora, sorry
16:27:28 <wwoods> I need names
16:27:34 <adamw> whoever is the packaging layer
16:27:48 <wwoods> ah.
16:28:02 <Oxf13> wwoods: Anaconda upstream, namely Jeremy, has long stated that anaconda as an upstream should be treated like any other upstream
16:28:13 <Oxf13> Fedora rules/policies should not and cannot apply to their upstream development
16:28:14 <adamw> whoever does the packaging of anaconda into fedora. even if they're anaconda devs, they should be wearing a different hat to do the fedora stuff, if anaconda is so separate from fedora and un-controllable by its freeze points as is claimed.
16:28:24 <Southern_Gentlem> adamw,  since fedora release engineering is also most part of the anconda team good luck
16:28:26 <Oxf13> ergo the layer to apply such rules and pressure is at the packaging layer.
16:28:55 <adamw> Southern_Gentlem: as i'm saying, they can't have their cake and eat it. :) if they don't want to treat fedora packaging as a separate job,the position that anaconda is like 'any other upstream' is not sustainable.
16:28:59 <Oxf13> Southern_Gentlem: actually I agree with adamw, and was very pissed when an unrelated change broke anaconda for the alpha.
16:29:10 <jlaska> I think coming to a conclusion is outside the scope of this meeting, but if there are ideas people want to develop and bring to anaconda for consideration ...we can note it
16:29:56 <adamw> ok, i put mine on the record then: anaconda packaging should be treated like packaging 'any other upstream', and when we hit development freezes, only necessary changes - not full unstable anaconda point upgrades - should be committed
16:30:38 <jlaska> thanks adamw
16:30:38 <Oxf13> adamw: while I agree, we don't actually make that claim of other packages.
16:30:45 <Oxf13> or rather we don't actually enforce it
16:31:03 <Oxf13> we want builds that fix bugs, but we don't go deep into that build to see if anything else has changed
16:31:06 <adamw> Oxf13: yeah, i should un-bundle those two points; the second on the basis that, as discussed here, anaconda is so vital and 'subtle'
16:31:33 <Oxf13> maybe subtle is the wrong word to use, 'difficult to test individual changes' is more appropriate
16:31:45 <adamw> anyway, as jlaska says, we're going to take over the meeting with this, so let's take it to a different context
16:32:16 <Southern_Gentlem> adamw,  i agree with you totally
16:32:29 <dpravec> if you want me to go to anaconda team in brq and ask them something, tell me
16:32:50 <adamw> i think we'll come up with a more polished proposal / explanation of it first :)
16:33:00 <dpravec> ok
16:33:16 <jlaska> alright, the last point on the F-12-Alpha-TC was around finishing what we started
16:33:40 <jlaska> it's dependenct on getting a new anaconda build tagged ... but just a general call for verification of MODIFIED bugs
16:34:01 <jlaska> I've you filed bugs, please do your best to continue triaging them
16:34:24 <jlaska> if you have bugs with fixes available (MODIFIED), please circle back and help confirm the reported problem is resolved (see https://www.redhat.com/archives/fedora-test-list/2009-August/msg00241.html)
16:34:26 <adamw> i think the above-discussed problems may be affecting that
16:34:33 <jlaska> yes they are :)
16:35:19 <jlaska> I'd like to make sure we do get some visibility on those MODIFIED bugs to make sure what we proposed as a blocker last week is resolved
16:36:20 <jlaska> that's all I had on the follow-up from F-12-Alpha-TC test run ... any other comments before moving on?
16:37:21 <jlaska> alrighty ...
16:37:27 <jlaska> #topic F-12-Alpha readiness
16:37:37 <Oxf13> not ready.
16:37:41 <jlaska> heh
16:37:54 <adamw> i think _rawhide_ itself is in fairly decent shape, but anaconda is clearly a pile of pain
16:37:55 <jlaska> there is an F-12-Alpha readiness meeting scheduled for this afternoon during the rel-eng meeting timeslot
16:38:15 <jlaska> adamw: yeah, you raised some points on Friday which prompted me to add this to the agenda
16:38:20 <jlaska> I think we have a good sense where the installer is now
16:38:50 <jlaska> but I don't have a lot of insight into other components  that you guys are more familiar with
16:38:54 <jlaska> e.g. xorg-x11
16:39:17 <jlaska> adamw: looks like that nouveau build was tagged ... are there other issues that need to be on the radar for xorg-x11-server and related drv's?
16:39:21 <adamw> xorg is, afaik, basically fine as of today
16:39:39 <adamw> with the updated server from the weekend and the updated nouveau from today, all the major brokenness i'm aware of is gone
16:39:58 <jlaska> I've got gdm again with those builds ... so good to hear!
16:40:15 <Oxf13> "from the weekend" ?
16:40:23 <Oxf13> I don't think anything got tagged on teh weekend
16:40:26 <adamw> or whenever it actually got into a compose
16:40:30 <Oxf13> aside from your nouvau
16:40:32 <adamw> i forget, may've been friday
16:40:37 <Oxf13> k
16:40:56 <dpravec> 2 weeks ago when i tried rawhide on my laptop, it failed to detect my intel card (which works in fedora)
16:40:56 <adamw> -28 went in saturday morning.
16:41:11 <adamw> dpravec: two weeks ago? that's _decades_ in rawhide time :D
16:41:23 <dpravec> :)
16:41:28 <jlaska> dpravec: are you still having that problem?
16:41:52 <adamw> seriously, if you could test with today's compose that'd be best
16:41:53 <dpravec> well i do have only virtualized rawhide now
16:42:05 <adamw> ah
16:42:14 <dpravec> this is bugging me in current fedora 11 https://bugzilla.redhat.com/show_bug.cgi?id=516558
16:42:15 <buggbot> Bug 516558: medium, low, ---, krh, NEW, cannot switch alt-button2 and alt-button3 when using desktop effects (compiz)
16:42:22 <oldfart> jlaska:  you asked on the qa list if the three bz's I encountered today prevented me from installing fedora and the answer is yes they do.
16:42:32 <jlaska> dpravec: let's try to keep things focused on the Alpha for now
16:42:36 <dpravec> ok
16:42:38 <Oxf13> adamw: ah, it was /tagged/ on Friday, but then showed up in the Saturday compose.  gotcha.
16:43:44 <adamw> i'm not really aware of any other major gotchas in current rawhide, my system's running fine at least.
16:44:12 <jlaska> okay ... dpravec if you've got a rawhide live image (fit'n'finish event should have one soon) that might be a good candidate to confirm current status of intel drv
16:44:40 <dpravec> i will try to run it on the bare metal
16:44:49 <dpravec> yes
16:44:54 <jlaska> the other concern I have for F-12-Alpha is around the FeatureList
16:45:44 <jlaska> this might be more of a question for the feature wrangler since I know he's typically on top of the F 12 features
16:46:12 <Oxf13> who is that masked feature wrangler?
16:46:25 <jlaska> and I don't see it as QA's job to report on development status of upcoming features ... but it's a still a worry
16:46:34 <jlaska> Oxf13: poelcat is the feature wrangler
16:46:45 <Oxf13> jlaska: I was riffing on the old radio shows
16:47:07 <Oxf13> jlaska: QA should concern themselves with what features are or are not "testable" and whether they have yet created a test plan
16:47:12 <jlaska> Oxf13: I think I left my sense of humor at the door
16:47:33 <jlaska> Oxf13: hrmm, maybe
16:47:37 <poelcat> jlaska: there is nothing funny about the feature wrangler ;-)
16:47:41 <Oxf13> my sense of humor slipped a week.
16:47:46 <poelcat> lol
16:47:53 <jlaska> Oxf13: forshadowing? :)
16:48:10 <Oxf13> muttergrumble
16:48:55 <jlaska> Oxf13: yeah I agree in a perfect world we would have QA resources lined up with each feature, but that's not going to happen any time soon
16:49:44 <jlaska> so, I don't have anything more to add unfortunately than just a grumble
16:49:48 <Oxf13> well, even with a lack of resources to do the /testing/, we should try to make sure that things are lined up for volunteers should they want to test
16:50:03 <jlaska> at what cost?
16:50:09 <yoyoned_> what is the state of test plans for the feature list
16:50:24 <jlaska> yoyoned_: exactly, I don't think we have that measurement
16:51:10 <jlaska> Oxf13: where we draw the line now on things that Fedora QA can actively engage on now is test days
16:51:19 <Oxf13> well, if we're not going to review the test plan, nor put any resources to testing it, why even bother with that section blocking feature submissions in the first place?
16:51:30 <adamw> didn't we review the test plan status for f11?
16:51:35 <adamw> to a resounding chorus of shrugs?
16:51:39 <jlaska> yes
16:52:21 <jlaska> so the short-term take away from this is ... "Hey look there are a lot of features coming"
16:52:32 <jlaska> and if you want any focused coordinated testing on this stuff ... think test day
16:53:08 * poelcat wonders what topic we are discussing/deciding right now?
16:53:17 <jlaska> poelcat: this was a jlaska ramble topic
16:53:21 <jlaska> moving on now ... nothing to see here
16:53:35 <jlaska> a bit concerned that we are feature freeze and have a lot of features that don't seem frozen
16:53:57 * poelcat could start another round of drama by sending some email ;-)
16:53:59 <Oxf13> feature freeze just means 'testable"
16:54:11 <Oxf13> for whatever measure of 'testable' feature owners set
16:54:23 <adamw> right, i think one thing that came out of the -devel-list discussion is that that was apparently not at all clear
16:54:31 <jlaska> yeah
16:55:39 <adamw> i'm looking at the 'feature process' page and not seeing any explanation of what the 'completion' percentage should actually be measuring
16:55:59 * poelcat suggests covering feature process in separate meeting... starting with a proposal
16:56:21 <adamw> yeah, fair point, we're off-topic here, probably
16:56:29 <jlaska> yeah, I started this topic, so I'll be happy to come back with something actionable
16:56:40 <jlaska> okay ... next up ... just a few upcoming QA events
16:56:46 <jlaska> #topic Alpha Compose
16:57:05 <jlaska> not a lot on this except that as some date in the near future a compose will land in the woods
16:57:38 <jlaska> there are still issues on the list, but stay tuned to the new @ 5 for the latest
16:58:15 <jlaska> once the compose lands, we'll be asked to help validate the alpha candidate as well as verify defects
16:58:19 <jlaska> also coming this week ...
16:58:22 <jlaska> #topic Fit'n'Finish Peripheral Test day
16:58:52 <jlaska> you may have seen mclasen's note to several fedora lists
16:58:53 <jlaska> https://www.redhat.com/archives/fedora-test-list/2009-August/msg00192.html
16:59:26 <jlaska> there are some very basic test cases defined, so I encourage folks to lend an hour to provide feedback using a live image
16:59:36 <jlaska> and last up for the week ...
16:59:47 <jlaska> #topic NetworkManager Test day
17:00:01 <jlaska> We'll have our first main track event kicking off this week with NM
17:01:07 <adamw> yaay
17:01:11 <jlaska> adamw: we've got a few requests to assist with the planning of the NM event this week
17:01:15 <jlaska> ;)
17:01:22 <adamw> i talked to vbenes already
17:01:34 <jlaska> oh good
17:02:29 <jlaska> I stubbed out a crummy page just to quell any feers that the event wasn't happening
17:02:41 <jlaska> but that page will clearly need the adamw touch
17:03:12 <jlaska> each week now we'll be having a main track test day event
17:03:13 <adamw> i'm working on that today, around all these useless^^^^^^^highly productive meetings
17:03:14 <adamw> ;)
17:03:23 <jlaska> ;)
17:04:09 <jlaska> should be a good example for dpravec and kparal who will be hosting their first next week ... ABRT
17:04:39 <jlaska> #topic <your topic here>
17:04:52 <jlaska> okay folks ... that's enough of me talking
17:05:24 <jlaska> any open topics people would like to discuss?
17:06:17 <jlaska> oldfart: Lemme circle back with you on those 3 bugs you raised earlier
17:06:24 <wwoods> I'll do a super-fast autoqa update
17:06:33 <jlaska> wwoods: go for it!
17:06:40 <jlaska> #topic Open discussion - AutoQA update
17:06:55 <wwoods> okay so last week I got the install test pretty well finalized
17:07:12 <wwoods> and then did a major refactoring of the current autoqa code base (in a git branch named wwoods-autotest)
17:07:37 <wwoods> autotest-ified the existing tests, rewrote the autoqa test launcher and updated the watcher scripts
17:07:44 <wwoods> updated the docs to reflect the changes
17:08:07 <wwoods> as of right now some of the tests are starting to send emails to the autoqa-results list
17:08:35 <wwoods> we should have them running automatically every day (and sending test results) Very Soon
17:08:38 <wwoods> probably before the week is out
17:08:53 <jlaska> wwoods: do I need to bump anything to get those in w/o moderation?
17:09:08 <wwoods> this is running on some internal test systems, it's an experimental trial
17:09:38 <wwoods> a public trial will be forthcoming, along with docs on writing tests/watchers and other fun things.
17:09:38 <wwoods> jlaska: no idea
17:10:20 <jlaska> wwoods: sounds like a ton of progress last week.  Nice work integrating the existing watchers+autoqa driver into autotest
17:11:08 <wwoods> yeah autotest is just the test harness/scheduler - we should be keeping the tests and the watchers no matter what else changes
17:11:34 <wwoods> anyway we'll work out the autoqa-results moderation stuff ASAP.
17:11:36 <wwoods> that's all.
17:11:43 <jlaska> cool, thanks for the update!
17:11:48 <jlaska> #topic Open discussion
17:12:03 <jlaska> any other topics to discuss in todays meeting?
17:12:12 <jlaska> closing out in 2 minutes ..
17:13:06 <Oxf13> *doooom*
17:13:40 <jlaska> heh, just about :)
17:13:44 <adamw> i'm going to sing the doom song now!
17:14:06 <jlaska> btw ... the movie is no good
17:14:10 <jlaska> any with that ... let's close it out
17:14:15 <Oxf13> people are just noticing that the dracut images are /huge/ and we don't have room for them on Live images
17:14:19 * adamw is the sixth best doom player in the world
17:14:21 <Oxf13> so that's going to be a fun conversation.
17:14:25 <jlaska> #endmeeting