From Fedora Project Wiki

< QA‎ | Meetings

(Add IRC transcript)
(Updated notes)
 
Line 2: Line 2:


People present (lines said)
People present (lines said)
# jlaska (152)
# adamw (41)
# jskladan (37)
# kparal (32)
# wwoods (26)
# fcami__ (25)
# brunowolff (5)
# mkrizek (2)
# fenrus02 (1)
# phuzion (1)
# jsmith (1)
# Alam (1)
# Viking-Ice (1)


Unable to attend:
Unable to attend:
Line 18: Line 31:
: Call for ideas/owners for Fedora 15 [[QA/Test_Days]]
: Call for ideas/owners for Fedora 15 [[QA/Test_Days]]
: Monitor [[Releases/15/FeatureList|list of approved features]] for ideas
: Monitor [[Releases/15/FeatureList|list of approved features]] for ideas
: [[User:Adamwill|AdamW]] initiated GNOME3 discussion on [http://lists.fedoraproject.org/pipermail/desktop/2010-November/006702.html desktop@ list]
: Several events scheduled so far (GNOME 3, Xfce 4.8 and X Test Week)
; Next steps ...
; Next steps ...
: Contribute ideas to [[Talk:QA/Fedora_15_test_days]]
: Contribute ideas to [[Talk:QA/Fedora_15_test_days]]
: Identify owners to organize and host test days
: Identify owners to organize and host test days
 
: Schedule another l10n/i18n test day
GNOME 3 (pre-Alpha)
Xfce 4.8
X Test Week


== Bodhi patch ==
== Bodhi patch ==
Line 41: Line 51:
: Tune existing python scripts to detect UNTESTED blocker bugs
: Tune existing python scripts to detect UNTESTED blocker bugs
: Bruno drafted a set of sample queries, see [[User:Bruno#Mockup_for_QA_-_Tracking_bug_queries]]
: Bruno drafted a set of sample queries, see [[User:Bruno#Mockup_for_QA_-_Tracking_bug_queries]]
: The team discussed different approaches for how to integrate this list into current process
; Next steps ...
; Next steps ...
: See {{ticket|fedora-qa|89}}
: The team tentatively agreed to review the list in each blocker review meeting
: jlaska going to merge the proposed queries into the Blocker meeting SOP


== Requirements review for Fedora test case management ==
== Requirements review for Fedora test case management ==
Line 49: Line 61:
: See {{ticket|fedora-qa|152}}
: See {{ticket|fedora-qa|152}}
: Requirements review for Fedora test case management
: Requirements review for Fedora test case management
: Drafted [[Rhe/tcms_requirements_proposal]]
: Hurry started drafting requirements at [[Rhe/tcms_requirements_proposal]]
; Next steps ...
; Next steps ...
: Gather feedback on requirements proposal
: Hurry will continue refining requirements
: Continue refining requirements
: All - please review and add any comments/concerns in Talk page


== Critical Path test case development ==
== Critical Path test case development ==


; Owner : [[User:adamwill|adamwill]]
; Owner : [[User:adamwill|adamwill]]
: Critical Path test case development
: See {{ticket|fedora-qa|154}}
: Still experimenting with some mock-ups/examples
; Next steps ...
; Next steps ...
: See {{ticket|fedora-qa|154}}
: AdamW planning to mail the list explaining proposed system and linking to examples


== AutoQA Update ==
== AutoQA Update ==
Line 65: Line 78:
; Owner : [[User:kparal|kparal]]
; Owner : [[User:kparal|kparal]]
; Summary
; Summary
: clumens' anaconda_storage test has been reviewed and ready to land in git ''master'' branch
: The patch representing second attempt for solving ticket #205 has landed into autoqa-devel.  That means we will be able to send our test results as comments into Bodhi.
: lmacken enabled us to use staging Bodhi instance for AutoQA development purposes (mainly he turned off email notifications).  Kudos to him for that
: mkrizek and I worked on ticket #241 [6] ("Add support for a staging server"). The patch is not posted yet (waiting for previous patch to land in master), but the code is available at mkrizek-staging branch
: jskladan works on a different kind of a koji watcher. His work is available in the 'new_koji_watcher' branch. His work should enable us to hook up depcheck properly into our framework.  Some discussion followed on how best to group -pending updates, the group agreed to discuss offline
: wwoods still working on a blog post and code to handle ticket #248 (simultaneous depcheck test runs)
; Next steps ...
# Wwoods posting blog article explaining the need for {{Ticket|autoqa|248}}
# Wwoods posting blog article explaining the need for {{Ticket|autoqa|248}}
# Jskladan rewriting bodhi watcher to work with depcheck (encountering depcheck issues)
# Determine appropriate strategy for watcher and depcheck integration, and revise patchset as needed
; Next steps ...


= Open discussion - <Your topic here> =
= Open discussion - <Your topic here> =


= Action items =
= Action items =
# jlaska to ping jens+rhe and Alam for thoughts on a F15 l10n/i18n test day
# jlaska to merge [[User:Bruno#Mockup_for_QA_-_Tracking_bug_queries]] into [[QA:SOP_Blocker_Bug_Meeting]]


= IRC Transcript =
= IRC Transcript =

Latest revision as of 16:30, 14 December 2010

Attendees

People present (lines said)

  1. jlaska (152)
  2. adamw (41)
  3. jskladan (37)
  4. kparal (32)
  5. wwoods (26)
  6. fcami__ (25)
  7. brunowolff (5)
  8. mkrizek (2)
  9. fenrus02 (1)
  10. phuzion (1)
  11. jsmith (1)
  12. Alam (1)
  13. Viking-Ice (1)

Unable to attend:

  1. Rhe (hopefully sleeping)

Agenda

Previous meeting follow-up

Call for Test Days

Owner
jlaska
Summary
Call for ideas/owners for Fedora 15 QA/Test_Days
Monitor list of approved features for ideas
Several events scheduled so far (GNOME 3, Xfce 4.8 and X Test Week)
Next steps ...
Contribute ideas to Talk:QA/Fedora_15_test_days
Identify owners to organize and host test days
Schedule another l10n/i18n test day

Bodhi patch

Owner
fcami
Summary
Posted a bodhi patch to improve the bugzilla comment bodhi posts when an update is available
See also fedora-infrastructure ticket#701
Next steps ...
Is there another bodhi-1.x build planned?

Python script communicating with bugzilla

Owner
Bruno
Tune existing python scripts to detect UNTESTED blocker bugs
Bruno drafted a set of sample queries, see User:Bruno#Mockup_for_QA_-_Tracking_bug_queries
The team discussed different approaches for how to integrate this list into current process
Next steps ...
The team tentatively agreed to review the list in each blocker review meeting
jlaska going to merge the proposed queries into the Blocker meeting SOP

Requirements review for Fedora test case management

Owner
rhe
See fedora-qa ticket#152
Requirements review for Fedora test case management
Hurry started drafting requirements at Rhe/tcms_requirements_proposal
Next steps ...
Hurry will continue refining requirements
All - please review and add any comments/concerns in Talk page

Critical Path test case development

Owner
adamwill
See fedora-qa ticket#154
Still experimenting with some mock-ups/examples
Next steps ...
AdamW planning to mail the list explaining proposed system and linking to examples

AutoQA Update

Owner
kparal
Summary
clumens' anaconda_storage test has been reviewed and ready to land in git master branch
The patch representing second attempt for solving ticket #205 has landed into autoqa-devel. That means we will be able to send our test results as comments into Bodhi.
lmacken enabled us to use staging Bodhi instance for AutoQA development purposes (mainly he turned off email notifications). Kudos to him for that
mkrizek and I worked on ticket #241 [6] ("Add support for a staging server"). The patch is not posted yet (waiting for previous patch to land in master), but the code is available at mkrizek-staging branch
jskladan works on a different kind of a koji watcher. His work is available in the 'new_koji_watcher' branch. His work should enable us to hook up depcheck properly into our framework. Some discussion followed on how best to group -pending updates, the group agreed to discuss offline
wwoods still working on a blog post and code to handle ticket #248 (simultaneous depcheck test runs)
Next steps ...
  1. Wwoods posting blog article explaining the need for autoqa ticket#248
  2. Determine appropriate strategy for watcher and depcheck integration, and revise patchset as needed

Open discussion - <Your topic here>

Action items

  1. jlaska to ping jens+rhe and Alam for thoughts on a F15 l10n/i18n test day
  2. jlaska to merge User:Bruno#Mockup_for_QA_-_Tracking_bug_queries into QA:SOP_Blocker_Bug_Meeting

IRC Transcript

jlaska #startmeeting Fedora QA Meeting 16:00
zodbot Meeting started Mon Dec 13 16:00:06 2010 UTC. The chair is jlaska. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00
zodbot Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00
jlaska #meetingname fedora-qa 16:00
zodbot The meeting name has been set to 'fedora-qa' 16:00
jlaska #topic Gathering life forms 16:00
jlaska Hi all ... show of hands for those joining the QA meeting today 16:00
jlaska or cilia, whichever is appropriate 16:00
* jskladan here 16:00
* mkrizek is here 16:00
* Alam here 16:01
adamw yo! 16:01
* brunowolff I am here in case someone has a question about the not VERIFIED bug queries 16:01
* kparal hellos everyone 16:01
* fenrus02 waves 16:02
brunowolff As an aside the fedora meeting page says the meeting is at 1500. I thought I missed it. 16:02
jlaska brunowolff: oh thanks ... lemme update that 16:02
* jlaska waits 60 seconds to start meeting 16:02
jlaska who are we missing? Viking-Ice wwoods robatino? 16:03
jlaska alright, let's get moving 16:03
jlaska #topic Previous Meeting follow-up 16:03
jlaska I have no action items from last week 16:04
jlaska we have some ongoing topics, but those will be covered in the agenda 16:04
jlaska I continue to delay the QA retrospective recommendations work due to other conflicts 16:05
jlaska thankfully ... that's not stopping some of you from moving forward on big topics for F15 16:05
jlaska So if you're feeling down about being behind on a task ... you can join me in #behind :) 16:05
jlaska alright, moving into the agenda 16:06
* Viking-Ice here 16:06
jlaska howdy! 16:06
jlaska #topic Call for Test Days 16:06
* phuzion is here 16:06
jlaska Just re-iterating last weeks call for topics 16:06
jlaska phuzion: hello :) 16:06
* adamw just sent out an email with that topic to devel-announce... 16:06
jlaska #link http://lists.fedoraproject.org/pipermail/devel-announce/2010-December/000733.html 16:07
jlaska adamw: that's great, thank you 16:07
jlaska adamw: you scheduled quite a few events last week already 16:07
adamw well, mostly just several recurrences of two or three :) 16:07
jlaska the xorg-x11-drv week ... our bread'n'butter ;) 16:07
jlaska Alam mentioned another l10n/i18n event 16:08
jlaska I'll ping jens, rhe and Alam about that for F15 16:08
jlaska iirc, jens and rhe hosted a similar (and successful) event for F14 16:08
adamw that'd be cool yep 16:08
adamw the one for f14 was great, it identified some bugs we got marked as NTH and fixed 16:09
adamw probably wouldn't have been caught otherwise 16:09
jlaska #action jlaska to ping jens+rhe and Alam for thoughts on a F15 l10n/i18n test day 16:09
jlaska The RetraceServer guys were interested in hosting something mid-march 16:10
jlaska but nothing final yet ... I've got an action item to ping them too 16:10
jlaska adamw: do you think systemd will likely land on the schedule somewhere? 16:11
adamw should do yep 16:11
adamw i'm waiting to hear from lennart and/or harald 16:11
jlaska adamw: sweet, I should have known you already reached out on that :) 16:11
jlaska Some links for those reading the logs ... 16:12
jlaska #link https://fedoraproject.org/wiki/Releases/15/FeatureList 16:12
jlaska #link http://fedoraproject.org/wiki/QA/Fedora_15_test_days 16:12
jlaska adamw: anything else to note on this topic? 16:12
adamw harald's on vacation, i'm hoping lennart will get back soon 16:12
adamw nope, i've mostly thrown in the things i had on my list 16:12
jlaska adamw: thanks for moving that process forward 16:13
adamw npnp 16:13
jlaska #topic Bodhi patch 16:13
* fcami__ waves 16:14
jlaska hey, there he is :) 16:14
fcami__ sorry, just got back from a meeting 16:14
* wwoods is lurking 16:14
jlaska wwoods: hey 16:14
jlaska so fcami posted a patch to update the comment that bodhi posts into bugzilla when an update is available for testing 16:14
fcami__ then I sent an updated patch to jlaska/adamw yesterday 16:14
jlaska #link http://fedorahosted.org/fedora-infrastructure/ticket/701 16:15
fcami__ the latest patch integrates all the comments done by bugzappers during the last meeting 16:15
jlaska fcami__: for next steps ... I think that patch needs review from the bodhi-devel team, and then figure out what to apply it against? 16:15
jlaska is lmacken lurking? 16:15
fcami__ yes, if you're all OK with that 16:15
adamw sure 16:16
adamw just looking at the patch... 16:16
adamw why are there two slightly different messages in different places? 16:16
fcami__ sorry, I should have posted it to some mailing list 16:16
jlaska is the patch available for lmacken to review as well? 16:16
fcami__ hmmm, they shouldn't be different 16:16
* adamw tries to fpaste it 16:17
adamw one seems to have just a bit more text 16:17
fcami__ the two places are 1/ what happens when you edit a bodhi action, adding bugs 2/ when you create a bodhi action with bugs 16:17
jlaska #link http://fpaste.org/F3Pv/ 16:17
jlaska ^^^ patch 16:17
adamw oh. no, actually, they are the same 16:17
adamw i just was missing where it wrapped a line :) 16:17
adamw so looks fine 16:17
fcami__ oy ok :) 16:17
jlaska is there a bodhi-devel list? How do we get this on the right radar? 16:17
adamw except i still thought it would be good to have it give a note to log in before leaving feedback 16:18
fcami__ I can add that, yes 16:18
fcami__ there is bodhi@fedorahosted IIRC 16:18
fcami__ I'll CC lmacken 16:18
jlaska fcami__: ah yeah ... https://fedorahosted.org/mailman/listinfo/bodhi 16:18
jlaska fcami__: also, is this patch against the current bodhi code base, or the bodhi-2.0 stuff? 16:19
fcami__ jlaska, whatever's in master in the git repo. maybe I should dig more. 16:19
jlaska fcami__: I think lmacken can probably add some data there ... since he'll likely be the one accepting the patch (I think) 16:20
* jlaska sees you already started the discussion https://fedorahosted.org/pipermail/bodhi/2010-December/000543.html 16:20
fcami__ yes, I did that, that's the previous version 16:21
jlaska what's the next step? 16:21
fcami__ nirik replied to let me know there was an infra ticket 16:21
fcami__ next step is to update the patch with the log in requirement 16:21
jlaska cool, and then we should track progress on the thread you started on bodhi@lists.fedorahosted.org ? 16:22
fcami__ yeah, I think so 16:22
fcami__ I'll CC test@lists.fedoraproject.org 16:22
fcami__ as I did last time 16:22
jlaska okay ... sounds like you've got all the ducks in a row :) 16:22
jlaska shall we check-in on this again next week? 16:23
fcami__ yeah, probably so 16:23
jlaska okay 16:23
fcami__ I should update the patch tonight 16:23
jlaska thanks fcami__ 16:23
fcami__ and then I'll try to reach lmacken directly 16:23
fcami__ np jlaska 16:23
fcami__ because I'd like to know if he's OK with the general idea 16:24
* fcami__ gives the mic back 16:24
jlaska alright ... bruno's next 16:24
jlaska #topic Improve tracking untested blocker bugs 16:24
jlaska This is a follow-up on ... 16:24
jlaska #link http://fedorahosted.org/fedora-qa/ticket/89 16:24
jlaska brunowolff created a draft wiki page with some stock queries we can use to keep track of blocker bugs that get CLOSED but are never VERIFIED 16:25
jlaska #link https://fedoraproject.org/wiki/User:Bruno#Mockup_for_QA_-_Tracking_bug_queries 16:25
jlaska brunowolff: that looks about what we needed. I don't know if we have a pressing need for tracking the NTH bugs that aren't VERIFIED 16:26
jlaska so if we were looking to slim up the links ... that might be one option 16:26
jlaska adamw: any thoughts? 16:27
brunowolff The one possibly significant limitation is that only bugs directly dependent on each tracker are checked. 16:27
adamw we don't need to track the NTH bugs, but hey, it doesn't hurt either. 16:27
jlaska The general idea here I think is that we want to incorporate review of these bugs into the blocker bug SOP somehow 16:27
brunowolff If there are normal bugs depending on other normal bugs that aren't verified, that won't get caught. 16:27
adamw yeah, that limitation is unfortunate 16:27
adamw especially when we use component or team blockers, like the anaconda and kde and virt trackers 16:28
jlaska brunowolff: yeah, I'm used to that limitation ... so that doesn't bother me too much 16:28
jlaska just as long as we call it out somewhere 16:28
jlaska so ... how would we respond to bugs on these lists? 16:28
adamw i kinda thought the aim of the tracker was to ensure that such bugs don't exist 16:29
jlaska my initial thought, is these lists would serve as extra data ... helping us decide whether some serious looking Blocker bugs never were tested 16:29
adamw so it's to be used to identify process fail more than some sort of ongoing thing 16:29
jlaska adamw: yeah, you got it 16:29
jlaska I don't know if we'll be able to knock out *all* bugs that show up on these lists 16:29
adamw although i suppose when we mark bugs as blockers early in the cycle and they just get fixed in the normal process of releasing we may skip verified 16:30
jlaska but that's a laudable goal 16:30
jlaska adamw: true true 16:30
jlaska howabout for F15 ... we periodically review the list during Blocker meetings, just like you do with NTH bugs? 16:30
adamw we could, sure 16:30
jlaska looking for any ... "Oh shoot!" moments? 16:30
adamw the only drawback with that is...longer blocker meetings 16:30
adamw :P 16:30
jlaska yeah, true true 16:30
jlaska hmm 16:30
jlaska the other option is firing the list of test@lists.fp.org each week ... calling out issues if needed? 16:31
adamw yeah 16:31
adamw so it's list spam vs. meeting time 16:31
adamw pick which you hate more 16:31
jlaska heh :) 16:31
jlaska we can move it to another meeting 16:32
jlaska this one 16:32
jlaska I feel like it's at least worth 5 minutes of a meeting to quickly scan the list 16:32
jlaska I'll be happy to do the scanning during the meetings in F15 16:32
adamw sure 16:32
jlaska if it's useless ... oh well 16:32
adamw let's see how it goes 16:32
jlaska yeah 16:32
jlaska should I integrate brunowolff's links into the blocker meeting SOP page? 16:33
jlaska otherwise ... I'll forget when we are in the thick of a release :) 16:33
adamw if we're going to do it in the blocker meeting, sure 16:34
adamw document the whole process is the goal 16:34
jlaska brunowolff: unless you have other recommendations, I can try to sanely merge your links into the blocker meeting SOP 16:34
jlaska #link http://fedoraproject.org/wiki/QA:SOP_Blocker_Bug_Meeting 16:34
jlaska #action jlaska to merge User:Bruno#Mockup_for_QA_-_Tracking_bug_queries into QA:SOP_Blocker_Bug_Meeting 16:34
brunowolff That's fine. If you have any further questions for me, you can bug me by email. 16:35
jlaska brunowolff: will do, thanks for the bz help :) 16:35
jlaska #topic Fedora test case mgmt requirements 16:35
jlaska Just a quick update on #topic 16:35
jlaska #link http://fedorahosted.org/fedora-qa/ticket/152 16:35
jlaska Hurry drafted an initial requirements page to start collecting/organizing data 16:35
jlaska #link https://fedoraproject.org/wiki/Rhe/tcms_requirements_proposal 16:35
jlaska Please do add comments/ideas/suggestions to the Talk: page 16:36
jlaska I plan to add some, hopefully constructive, feedback later this week 16:36
jlaska Alright, next topic ... 16:37
jlaska #topic CritPath test case development 16:37
jlaska #link http://fedorahosted.org/fedora-qa/ticket/154 16:37
jlaska adamw: how's it coming along on the critpath work? 16:37
jlaska any updates/blockers/concerns/question/jokes ... 16:38
adamw nope, just waiting on me to do something 16:38
adamw till's feedback was good, i'll go ahead incorporating that 16:39
jlaska cool. I don't know if it helped, but there's plenty of stuff we can get through the mediawiki API 16:39
adamw i'm going to write up one or two example (but useful...) test cases, in the appropriate name / category layout, then mail the list again explaining the proposed system and linking to the examples 16:39
jlaska lemme know if you want me to do any queries for you 16:39
jlaska adamw: sounds like a plan 16:39
jlaska #info still experimenting with some mock-ups/examples 16:40
jlaska #info Will mail the list explaining proposed system and linking to examples 16:40
jlaska alright ... kparal, are you ready to rock? 16:41
jlaska thanks adamw :) 16:41
kparal jlaska: ready to rock the boat :) 16:41
jlaska #topic AutoQA Update 16:41
adamw jlaska: i don't really need the queries, it's bodhi/f-e-k that need them :) 16:41
jlaska adamw: right on 16:41
jlaska adamw: I meant if you wanted feedback as to whether your examples are queryable as intended etc... 16:42
* kparal waiting a while 16:42
jlaska kparal: another busy week for AutoQA ... what's the latest? 16:42
kparal Hello gang, even though I didn't want to be as wordy as last time, I'm afraid I will be :) Here are the latest news: 16:42
kparal 1. clumens' anaconda_storage test has been reviewed propertly (regarding AutoQA part of the test) and it is ready to be merged into master, once clumens deems it's ready. Therefore it should be very probably part of the next release [1]. Enjoy automatic Anaconda tests in the near future, hooray! 16:43
kparal [1] https://fedoraproject.org/wiki/AutoQA_Roadmap 16:43
kparal (Hey, I have written it in advance, so it would be quicker, but feel free to stop me after any section if you have some comments to it) 16:43
kparal 2. The patch representing second attempt for solving ticket #205 [2] ("Provide a support for sending comments into Bodhi") has landed into autoqa-devel [3]. Big thanks to mkrizek. If no further concerns, it will be merged into master very soon. That means we will be able to send our test results as comments into Bodhi. Example is here: [4]. Currently this is planned for depcheck and for upgradepath test (not enabled yet). 16:44
kparal [2] https://fedorahosted.org/autoqa/ticket/205 16:44
kparal [3] https://fedorahosted.org/pipermail/autoqa-devel/2010-December/001435.html 16:44
kparal [4] https://admin.fedoraproject.org/updates/python-rpmfluff-0.3-6.fc14 16:44
jlaska hehe ... you kept dmalcolm on his toes in that update :) 16:45
kparal yes we did :) 16:45
kparal but he agreed. maybe he didn't know what to expect 16:45
jlaska nice workflow with that patch set guys ... proposing, incorporating feedback, providing examples etc... 16:46
wwoods it's lovely stuff 16:46
kparal 3. lmacken enabled us to use staging Bodhi instance for AutoQA development purposes (mainly he turned off email notifications). Kudos to him for that. There is currently also an invalid certificate problem, but that should be solved in some time. Meanwhile the process of (mis)using staging Bodhi server from AutoQA is described at [5]. 16:46
kparal [5] https://fedorahosted.org/pipermail/autoqa-devel/2010-December/001438.html 16:46
jlaska "insecure=True" ... that gives you a warm feeling :) 16:47
kparal :) 16:47
kparal alright, next on 16:48
kparal 4. mkrizek and I worked on ticket #241 [6] ("Add support for a staging server"). The patch is not posted yet (waiting for previous patch to land in master), but the code is available at mkrizek-staging branch. It should allow us have configurable Koji/Bodhi server URLs and configurable options whether to send emails (3 different kinds, actually) and Bodhi comments. 16:48
kparal [6] https://fedorahosted.org/autoqa/ticket/241 16:48
jlaska oh nice, that'll be a handy change 16:48
kparal (and expect much more documentation in autoqa.conf, as a bonus) 16:49
wwoods hooray documentation! 16:49
wwoods heh 16:49
jlaska btw ... who gets the credit for the killer __docstring__'s ? 16:49
kparal jlaska: which ones? 16:49
jlaska in the bodhi feedback patchset ... mkrizek has some really good docstrings for all the methods 16:50
kparal jlaska: the praise goes to mkrizek I believe 16:50
kparal I just asked him to document it well, he did the rest :) 16:50
jlaska nice! 16:51
kparal ok, and last but not least: 16:51
* mkrizek flattered 16:51
kparal 5. jskladan works on a different kind of a koji watcher. His work is available in the 'new_koji_watcher' branch. His work should enable us to hook up depcheck properly into our framework. It will also obsolete the post-bodhi-update watcher. I'll give him a word to say some key concepts of our new watcher and why we even need it. 16:51
* jskladan steps forward 16:52
jskladan OK, gang. We found ourselves in need of some updates on the watchers front. 16:52
* kparal hands out the mic to jskladan 16:52
jskladan The required changes were 16:52
jskladan 1) Make use of the -pending tags in Koji: https://fedorahosted.org/autoqa/ticket/228 16:52
jskladan 2) Create 'batch' scheduler: https://fedorahosted.org/autoqa/ticket/204 16:52
jskladan ad watching -pending tags) 16:52
jskladan At the moment koji watcher is querying Koji for the list of recently built packages (based on the time of build). 16:53
jskladan This is a bit unsatisfactory - we can (and we do) miss some spots in the testing chain: 16:53
jskladan 1) package Foo is built at date XYZ. It gains tag dist-f14-updates-candidate 16:53
jskladan 2) koji-watcher founds out "ha, new package, let's test it" 16:53
jskladan 3) tests are OK 16:53
jskladan 4) package Foo gains dist-f14-updates-testing-pending tag 16:53
jskladan 5) we'd like to run tests like depcheck on it, but because the 'built at' date, which the actual watcher checks is not altered, we miss the change 16:53
jskladan So we decided to use different querying model, based on the tagHistory() in Koji. 16:53
jskladan (kids, _do_ try this at home it's awesome :) koji list-tag-history --tag=dist-f14-updates-pending) 16:54
* jlaska tries 16:54
jskladan which effectively tells us which packages were 'pushed' to the tags we care about. 16:54
* jskladan hopes that there's no typo :) 16:54
jskladan ad "batch scheduling": 16:54
jlaska so this adds a new 'hook' name that tests would need to code to (for batch updates)? 16:55
jskladan yes, exactly 16:55
jskladan We also found us in need of 'batch' scheduling - e.g. we don't want to run depcheck for every package built, but we'd like just to inform autoqa "hey, there is new stuff in dist-f14-updates-pending tag, run tests". 16:55
jskladan So there is brand-spanking-new 'watch-koji-builds-batch' watcher. 16:56
jlaska jskladan, wwoods have you guys worked out how depcheck would integrate with this stuff? 16:56
jskladan which groups the updates according to the tag, and sends the whole 'batch' of updates at once 16:56
jskladan jlaska: I have not yet spoken with Will 16:56
jskladan jlaska: but I've tried it out in my testing environment, and it seems to be working like charm 16:57
jskladan jlaska: even though I'm not really sure about the differences in post-bodhi-update & post-koji-build 16:57
jlaska how do you mean? 16:57
jskladan (e.g. if we can miss something by watching only koji-tags, and amending the post-bodhi hook as such) 16:57
jskladan because at the moment, it seemed to me, that watching the tags is more straightforward for depecheck 16:58
jskladan because we can easily determine the repo 16:58
jskladan (or the tag, if you want) 16:58
jlaska from my understanding, it does seem to better align with the use of hte koji -pending tag 16:59
jskladan yes, this is also my point of view - I just wanted to make sure, that by setting the post-bodhi aside (at least for depcheck), we're not gonna miss any updates 16:59
jlaska can run the old watcher, and your watcher side-by-side to compare? 17:00
jlaska s/can/can we/ 17:00
jskladan the koji & bodhi watcher? sure can, but I belive that it suffers from the same problem as the old koji watcher - e.g. it's reacting to 'builds' instead of 'tag changes' 17:01
jlaska jskladan: I'll have to take it offline ... I think I'm getting confused :) 17:01
jskladan eh, my bad 17:01
kparal this is a bit longer description than I expected :) 17:02
jlaska wwoods and jskladan ... can you guys sync up on the watcher and depcheck integration? 17:02
jskladan to put it straight: the plan is to effectively disable post-bodhi, and replace it with post-koji watcher over the 'pending' tags 17:02
wwoods oop sorry, got waylaid by something, catching up on scrollback 17:02
jskladan in the mean time - if you're interested in the new koji watcher, feel free to explore the git: <http://git.fedorahosted.org/git/?p=autoqa.git;a=shortlog;h=refs/heads/new_koji_watcher> 17:03
jskladan and post comments on autoqa-devel (or IRC) 17:03
wwoods using tag-history is definitely the right approach 17:03
jlaska jskladan: nice ... I've got plenty of catch-up work to do on autoqa-devel :) 17:03
wwoods but batching updates.. I don't think that's necesssary or desirable 17:04
wwoods since the plan is to move to a messagebus as soon as it's feasible 17:05
wwoods and the messagebus is going to send individual messages for each event 17:05
wwoods it's not going to batch the messages, so we don't want to build batching into the infrastructure and then drop it later 17:05
kparal let's consult it after the meeting 17:06
wwoods this is ticket #204 17:06
wwoods https://fedorahosted.org/autoqa/ticket/204 17:06
jlaska yeah, was going to suggest continuing this in #fedora-qa after the meeting 17:06
jlaska what else on the autoqa front? 17:07
wwoods right - this is just a quick summary for the meeting notes 17:07
jlaska wwoods: thank you, that makes my minute gathering life easier! 17:07
wwoods ticket #204 involves batching updates from the watcher 17:07
wwoods a related problem is ticket #208, which concerns simultaneous depcheck runs 17:07
wwoods err, sorry 17:07
wwoods that's ticket #248 17:08
wwoods https://fedorahosted.org/autoqa/ticket/248 17:08
wwoods I've tried to give some details on the proposed solutions there 17:08
jlaska there's been a ton of change recently on the AutoQA front ... so if we need a side-meeting to talk through some of that ... let's do it 17:08
jlaska let's discuss post-meeting 17:10
wwoods anyway there's some complex, thorny issues with timing and tagging for depcheck (and other tests in this path) 17:10
* jsmith likes that idea 17:10
jlaska wwoods: kparal: I think you both are talking from different POV's in ticket#248 17:10
wwoods we should definitely discuss it further 17:10
wwoods I'm also working on a blog post to try to explain what the problems are and how we can fix 'em 17:10
wwoods not finished yet, but here's a diagram I'm working on: 17:10
wwoods http://wwoods.fedorapeople.org/images/repos-draft.png 17:10
kparal Thanks everyone, but this was supposed to be a summary, erm :) 17:10
kparal wwoods: we will discuss if after the meeting 17:11
wwoods heh, ah well 17:11
wwoods but yes. Further Discussion Outside The Meeting Is Suggested 17:11
jlaska no worries ... sounds like it identified a topic we need to drill down on outside of the meeting 17:11
jlaska wwoods: nice diagram :) 17:11
kparal thanks jskladan for the problem description 17:11
kparal and this was also my last topic in the "AutoQA news" 17:12
* jskladan is shocked by the outcome :-D but looking forward to discussing this further :) 17:12
jlaska thanks for the autoqa updates all 17:12
jlaska alright ... open discussion time 17:12
jlaska #topic Open discussion - <your topic here> 17:13
jlaska we've gone a bit over today ... are there any other items we need to review here? 17:13
* jlaska waits 60 seconds 17:13
jlaska 20 seconds ... 17:14
adamw nothing from me 17:14
jlaska 5 seconds ... 17:14
jlaska Alright gang ... thanks for your time and updates today 17:14
jlaska I'll follow-up to the list later today with minutes 17:15
jlaska #endmeeting 17:15

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