QA/Meetings/20100315

From FedoraProject

Jump to: navigation, search

Contents

Attendees

People present (lines said)

  • jlaska (109)
  • adamw (45)
  • kparal (22)
  • daumas (5)
  • zodbot (4)
  • Viking-Ice (4)

Regrets:

Agenda

Previous meeting follow-up

  1. maxamillion seeking input from the Mentors group for guidance on mentor responsibilities
  2. continued testing of bug#567346 to further isolate the failure conditions
  3. jlaska - update AutoQA_PatchProcess to include fedorapeople.org git repo

Fedora 13 test status

Upcoming test milestones:

  • Pre-beta acceptance test - awaiting new anaconda build before sending through AutoQA - https://fedorahosted.org/rel-eng/ticket/3495
  • 2010-03-18 - Beta test compose
  • 2010-03-18 - Palimpsest & udisk improvements Test Day
  • 2010-03-19 - F13Beta blocker review #2

proventesters

Jlaska asked the group what tasks remain in order to move forward on defining the proventesters group and workflow.

  • Adam noted that Jkeating may be working several tasks on this front.
  • Jlaska directed those interested in defining providetesters, to provide feedback to maxamillion's email.

Privilege Escalation test

Adamw reminded the group that we've not yet defined test cases to validate the recently approved Privilege_escalation_policy.

  • Adamw felt the first step is to come up with the script to identify privesc-relevant packages.
  • Kparal asked whether these would be automated tests, or tests more like the desktop validation plan. Discussion followed on the appropriate place to automate these
  • The group discussed candidate packages to inspect for polkit changes, including polkit configs, setuid binaries, certain d-bus config files, consolehelper files and pam configs.
  • Adam took action to discuss a privileged package detection script with wwoods

Open discussion - <Your topic here>

Update Policy Proposals

Kparal clarified that current package update proposals require that deps/conflicts tests occur whenever a package enters updates and updates-testing. Kparal also discussed concerns around AutoQA being a bottleneck for packages.

AutoQA packaging

Adamw asked how AutoQA packaging was progressing.

  • Jlaska noted that packaging wasn't moving, and that a FAD is being planned to help work down the list of java %buildrequires. More information will be sent to [1] later in the week.

Upcoming QA events

Action items

  1. adamw to check-in with wwoods on tooling needs for priv esc. test

IRC transcript

jlaska #startmeeting Fedora QA meeting 16:00
zodbot Meeting started Mon Mar 15 16:00:40 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 ... 16:00
* adamw gathers 16:01
* kparal is here and on eletricity power again 16:01
daumas there can be only one 16:01
jlaska kparal: hey, alright 16:01
jlaska wwoods: is out today 16:02
jlaska and I'm not expecting lili or rhe to be awake right now 16:03
jlaska kparal: is jskladan out for the day? 16:03
kparal jlaska: jskladan was hoping to be online, but it seems he's not here 16:04
jlaska okay 16:04
jlaska maxamillion had a conflict too iirc 16:05
jlaska #topic Previous meeting follow-up 16:05
jlaska #info maxamillion seeking input from the Mentors group for guidance on mentor responsibilities 16:05
jlaska not sure of anything new here, anyone else? 16:05
adamw not heere 16:05
jlaska I can catch up with maxamillion after 16:05
jlaska some of these tasks are a bit old ... next up ... 16:06
jlaska #info continued testing of bug#567346 to further isolate the failure conditions 16:06
jlaska .bug 567346 16:06
zodbot jlaska: Bug 567346 gpk-update-viewer does not install updates if there is any dependency issue, and does not correctly report this - https://bugzilla.redhat.com/show_bug.cgi?id=567346 16:06
adamw that's the update fail bug, right? 16:06
jlaska yeah, I believe this is the one you updated after Fridays blocker mtg? 16:06
adamw yeah. 16:06
jlaska Any objections to dropping this from the list, I think we've got it on the proper radar elsewhere 16:07
adamw sure 16:07
jlaska alrighty, next .. 16:08
jlaska #info jlaska to update AutoQA_PatchProcess to include fedorapeople.org git repo 16:08
jlaska nothing crazy here, I made a minor update to the AutoQA_PatchProcess wiki page pointing to the instructions for setting up a git repo on fedorapeople.org 16:08
jlaska https://fedoraproject.org/wiki/AutoQA_PatchProcess#Private_Branch 16:09
jlaska that's all I have from previous meetings 16:09
jlaska before we move on, anything else not covered? 16:09
jlaska alright ... diving into the agenda ... 16:10
jlaska #topic Fedora 13 test status 16:10
jlaska this first topic is intended as a check-in on f13 test activities 16:10
jlaska we've got quite a bit going on at the moment, and the beta is on the horizon 16:10
jlaska I'm just going to rattle off a few milestones that I'm tracking ... 16:11
jlaska #info Pre-beta acceptance test - awaiting new anaconda build before sending through AutoQA - https://fedorahosted.org/rel-eng/ticket/3495 16:11
jlaska next up, we have ... 16:12
jlaska #info 2010-03-18 - Beta test compose 16:12
jlaska I'm not seeing tickets yet for the beta composes, I'll catch up with Oxf13 after to make sure I don't fill dups 16:12
jlaska we also have ... 16:13
jlaska #info 2010-03-18 - Palimpsest & udisks improvements test day 16:13
jlaska so quite a bit going on this week, on top of the proposal storm 16:13
* Viking-Ice sneaks inn 16:14
jlaska I don't have a ticket for the palimpsest test day ... who is coordinating that event? 16:14
jlaska Viking-Ice: welcome :) 16:14
adamw i was about to ask the same 16:14
adamw ot 16:14
adamw it's on the schedule but i'm not sure who's in charge of that 16:14
adamw 18:02, 16 February 2010 Davidz (Talk | contribs) (3,046 bytes) (Add udisks test day) 16:14
adamw oh look, it's me! 16:15
adamw yeah, i have some emails from david about this, from back in february. i'll look after it. 16:15
jlaska adamw: want to divide and conquer on this one? 16:15
adamw nah, it's fine, i've got it. 16:15
jlaska since you had all of webcam last week? 16:15
jlaska we had a previous palimpsest test day, but I can't find it at the moment ... hopefully that has some useful tests 16:16
adamw there's stuff on the feature page too. 16:16
jlaska ah, nm ... I was thinking about DeviceKit (https://fedoraproject.org/wiki/Category:DeviceKit_Test_Cases) 16:17
jlaska I'll check-in with Hurry to see if she needs anything for the test compose milestone this week 16:17
adamw DeviceKit would be right 16:17
adamw DeviceKit-disks turned into udisks 16:18
jlaska I thought we had a focus on this already, but then the *Kit name threw me 16:18
adamw we had HAL, then DeviceKit, then DeviceKit-disks, then udisks 16:18
adamw easy! =) 16:18
jlaska kparal: I think we could make a fun flow chart for that progression :) 16:18
jlaska oh, and last thing on my radar for F13 testing this week 16:19
jlaska #info 2010-03-19 - F13 Beta Blocker bug review #2 16:19
jlaska Cranes Maska has quite a few bugs to knock out before that meeting 16:19
jlaska anything else on the F13 front folks want to share? 16:19
adamw i wouldn't expect much from that guy 16:19
jlaska adamw: I set the bar low 16:19
jlaska alrighty ... moving on ... 16:20
jlaska #topic proventesters 16:20
jlaska There has been some discussion on the list in previous weeks regarding the creation of a proventesters group, responsible for providing critpath bodhi feedback for QA 16:21
jlaska I've seen a few email from Oxf13, adamw and maxamillion, but wanted to give someone a chance to talk about where we are, what's next etc... 16:21
jlaska is there general agreement on the FAS group name, 'proventesters'? 16:22
adamw sure 16:22
adamw brb call of nature 16:22
kparal just call back :) 16:22
jlaska heh 16:22
jlaska so I don't see the group created yet, https://admin.fedoraproject.org/accounts/group/view/proventester ... so that seems like a reasonable starting point 16:23
daumas jlaska: what will a "proventester" acl give to a person who receives it? 16:24
jlaska daumas: great question, I'm not sure this process has been fully fleshed out yet 16:24
jlaska for today, I just wanted to check-in and see where things stood. I know maxamillion has a few threads out to help move things along on this front 16:25
jlaska perhaps people can lend their thoughts .... /me grabs links ... 16:25
jlaska http://lists.fedoraproject.org/pipermail/test/2010-March/088980.html 16:25
adamw basically we're working on maxa's proposal i thought? 16:25
Viking-Ice what happen with using the already existing QA group? 16:25
adamw Viking-Ice: http://lists.fedoraproject.org/pipermail/test/2010-March/089065.html 16:26
Viking-Ice Ah ok good point mentioned there.. 16:27
jlaska adamw: I thought so as well, but I know it's getting a lot of focus and I haven't seen a lot of feedback to maxa's RFC 16:27
jlaska if nothing else, I'll checking with maxamillion after the meeting and see what he needs to move forward 16:27
adamw be nice to hear where oxf13 is on it 16:27
jlaska #Help additional feedback needed, see http://lists.fedoraproject.org/pipermail/test/2010-March/088980.html 16:28
jlaska #help additional feedback needed, see http://lists.fedoraproject.org/pipermail/test/2010-March/088980.html 16:28
jlaska adamw: what aspect is Oxf13 working on? 16:29
adamw i'm not exactly sure, but he was definitely talking about the whole area 16:30
jlaska alright ... sounds like some data gathering is needed 16:30
jlaska I'll bug those two later on 16:30
jlaska okay ... next topic ... back by popular demand 16:30
jlaska #topic Privilege Escalation test 16:30
jlaska adamw: I know you've asked about drafting test cases to support the recent priv escalation policy, I wanted to give you some time to talk through it 16:31
adamw right 16:31
adamw basically just a reminder that we didn't write the policy for the fun of writing policies =) it was to support privesc testing 16:32
adamw i believe the next step was to come up with the script to identify privesc-relevant packages 16:32
kparal should this be automated test for autoqa or manual test, similar to current desktop validation test plan? 16:32
jlaska kparal: I wonder if these eventually might make good _instrospection_ tests? 16:32
adamw i'm trying to remember who it was said they would volunteer to do that back at the start, could be wwoods 16:32
adamw i think there's going to be automated and manual elements to it 16:32
adamw we can use autoqa to identify packages which have added critpath-relevant stuff, or packages where it changes 16:33
jlaska wouldn't hurt to start with manual definition 16:33
adamw and there may be common errors we can catch with autoqa (or may not) 16:33
adamw but there will also be manual tests we'll want to run on the identified packages 16:33
jlaska adamw: the best candidate I knew of was for any packages who change/add/remove their polkit definitions? 16:34
adamw that's part of it, but there are other things to check for too 16:34
adamw setuid binaries, certain d-bus config files I believe, and consolehelper files are the obvious 16:34
adamw there's also some lovely outliers out there, like the app I came across the other day which inexplicably chose to use some random thing called 'beesu' which wraps consolehelper 16:35
jlaska kparal: these seem like good candidates for rpmguard tickets once we have test defined ... since there is a comparison factor here? 16:35
kparal that's possible 16:35
jlaska kparal: as you mentioned earlier, I don't want to get too far ahead with tasking out autoqa stuff until we have the current mandatory list (and a place to store/view the results) complete 16:36
jlaska adamw: is this something you'd like to see in place for Final, or sooner? 16:36
adamw well the rpmguard stuff will be pretty 'obvious' once we have the tool to generate an overall list in place 16:36
adamw well it'd be nice to at least have a list of packages for us to eyeball, if nothing else, before final 16:37
jlaska who can provide that? 16:37
jlaska adamw: that seems like a reasonable goal ... given the 200 other things everyone has on their plates 16:38
adamw as I said, i think wwoods said when we started discussing this that he'd be able to come up with a tool 16:39
jlaska #info interested in a list of packages to apply priv esc tests against for final 16:39
adamw maybe take an action item for me to talk to him about it? anyone else want to be involved? 16:39
jlaska #info wwoods previously mentioned producing a tool to provide the list of packages 16:39
jlaska #action adamw to check-in with wwoods on tooling needs for priv esc. test 16:40
jlaska adamw: we have the policy in place now to help guide and severity questions around incoming bugs, so that does put us in better shape for F-13 testing than we were in for F-12 16:41
adamw sure, if something comes up we have a document to point to to say what it should do. 16:41
adamw but it would definitely be better to have an active testing plan in place rather than just hope for things to float to the surface =) 16:42
jlaska indeed 16:42
jlaska alright, anything else on that front ... or just the first step, follow-up w/ wwoods 16:42
adamw first step is fine for now 16:43
jlaska eggsellent, thanks adamw 16:43
jlaska okay, open discussion time ... 16:43
jlaska #topic Open discussion - <Your topic here> 16:43
jlaska anything not covered that folks would like to discuss? 16:43
* Viking-Ice nothing from me... 16:44
kparal well maybe I have a question about the proposals. but we can also discuss it afterwards 16:44
kparal about the update policy proposals, to be more exact 16:45
jlaska #topic Update Policy proposals 16:45
jlaska kparal: take it away 16:45
kparal alright, just something to be clear on. from the current discussions and proposals, it seems probably that we will have two autoqa checking rounds 16:46
kparal the first one when accepting packages to updates-testing, the second one when accepting packages to stable updates 16:46
kparal is that correct? 16:46
jlaska kparal: that's my understanding 16:47
kparal because for example I think we want to check dependencies sanity before pushing to updates-testing, not to break our repo 16:47
kparal OTOH the upgrade path should be checked just when before push to stable updates 16:47
jlaska right, and then the same to ensure that things are pushed int he right order to 'updates' 16:47
kparal otherwise the situation may change in between 16:48
jlaska exactly, for the tests that look for closure in deps and conflicts, it seems like those would need to run during any transition to 'updates-testing' or 'updates' 16:48
kparal ok, I'm just modifying my proposal again, so I will include these two rounds of autoqa checks. I'm just afraid it will sound even more complicated than the first proposal 16:48
kparal but the policies are meant to be complex, right? :) 16:49
daumas kparal: dep checking to stable pushes is loooong overdue 16:49
kparal daumas: what do you mean? 16:49
jlaska kparal: given the limited scope of the tests for package acceptance, I don't think we'll get a lot of push back from ensuring that both 'updates' and 'updates-testing' repos are dep solvable 16:50
jlaska kparal: I'm obviously not an english major, but I'll be happy to look at the wording and propose changes if there is something more concise that can convey the same meaning 16:50
daumas kparal: pushes get made, users encounter "unable to resolve deps" and complain to the lists. it's been commonplace for a while. it's more rare these days, but most recently happened with nss and nspr 16:50
kparal alright. I was thinking if we could make the AutoQA process run during the time the package spends in updates-testing. so AutoQA would not be a bottleneck (it may take a few hours) 16:51
kparal but as I understand it, we can't assume that something that was ok 2 days ago is still ok now 16:51
kparal e.g. the dependencies 16:51
kparal so AutoQA will slow it down a little at both sides 16:52
kparal if I see it wrong please correct me 16:52
jlaska kparal: one aspect with both policies is that they require autoqa be inproduction 16:53
daumas kparal: slowing down a few minutes will save days of recoup time. 16:53
jlaska so the very small test environment we are piloting now won't be representative of the resources available once deployed 16:53
kparal ok, that's a good answer for me 16:53
jlaska but honestly, I'd love to have the problem of autoqa being too slow 16:54
jlaska that would mean all these java packages are behind us :) 16:54
jlaska #topic Open Discussion - <your topic here> 16:54
jlaska alright ... anything else for today? 16:54
jlaska otherwise, I'll call #endmeeting in a few minutes 16:55
adamw where are you with the autoqa packaging stuff? 16:56
adamw drowning? need a rope? 16:56
jlaska adamw: several ropes are being shot out as we speak 16:56
adamw ah great 16:56
jlaska adamw: I'll have an email out to java-devel@ and probably devel@ later this week looking for FAD participants 16:56
jlaska #info adam asked how autoqa packaging was going, jlaska noted that a FAD is in the works, expect more later this week 16:57
jlaska alright gang ... thanks for your time 16:57
jlaska as usual, I'll send minutes to the list 16:57
jlaska #endmeeting 16:57

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