fedoraqa-devel
LOGS
14:00:30 <tflink> #startmeeting
14:00:30 <zodbot> Meeting started Mon Jun 29 14:00:30 2015 UTC.  The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:30 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:00:30 <tflink> #meetingname fedoraqa-devel
14:00:30 <tflink> #topic roll call
14:00:31 <zodbot> The meeting name has been set to 'fedoraqa-devel'
14:00:35 * kparal is here
14:00:46 <tflink> #chair kparal mkrizek
14:00:46 <zodbot> Current chairs: kparal mkrizek tflink
14:00:55 * mkrizek is here
14:01:08 <kparal> mkrizek: do you know where Josef is?
14:01:15 <tflink> doesn't look like we have a josef today, though
14:01:17 <mkrizek> kparal: no
14:02:19 <kparal> he wasn't online today at all
14:02:42 <kparal> I can call his cell phone
14:04:21 * randomuser lurks
14:04:38 * tflink can't get into his calendar to look if jskladan took the day off - assumes one of you did
14:04:52 <kparal> I did
14:04:59 <kparal> he doesn't pick up the phone
14:05:16 <tflink> :-/
14:05:22 <kparal> not sure what's up with him, but it seems there's no point in waiting
14:05:29 <tflink> yep
14:05:39 <tflink> who wants to go first with status report?
14:05:48 * kparal will
14:05:58 <kparal> #topic kparal status report
14:06:03 <kparal> #info worked towards completing task artifacts feature. I think it should be ready for deployment now. There's still one futher task (T520), but it should hopefully not affect real deployment, just local testing.
14:06:03 <kparal> #link https://phab.qadevel.cloud.fedoraproject.org/T399
14:06:10 <kparal> #info reported some issues related to setting up your own services (resultsdb etc), some further patches and documentation changes will follow, because currently there are quite a few pitfalls, as I learned (mkrizek saved me mostly) :)
14:06:23 * roshi lurks
14:06:34 <kparal> that's all I'm afraid, then I received new laptop and started reinstalling things
14:06:49 <tflink> fortunately, those docs keep improving
14:08:05 <kparal> if one of you has time to deploy the changes, I think we can start testing the artifacts feature
14:08:12 <tflink> any other questions/comments?
14:08:26 <mkrizek> nope
14:08:49 <tflink> mkrizek: you want to go next or should I?
14:08:54 <mkrizek> I will
14:09:07 <mkrizek> #topic mkrizek status report
14:09:24 <mkrizek> #info finished the logging rework patch, awaiting reviews
14:09:24 <mkrizek> #link https://phab.qadevel.cloud.fedoraproject.org/D382
14:09:25 <mkrizek> #link https://phab.qadevel.cloud.fedoraproject.org/T495
14:09:25 <mkrizek> #info figuring out how to "guess" branch name in distgit
14:09:25 <mkrizek> #link https://phab.qadevel.cloud.fedoraproject.org/D383
14:09:27 <mkrizek> #link https://phab.qadevel.cloud.fedoraproject.org/T492
14:09:29 <mkrizek> #info started to look into "latest stable build" in the koji directive
14:09:32 <mkrizek> #link https://phab.qadevel.cloud.fedoraproject.org/T491
14:09:34 <mkrizek> #info code reviews
14:10:16 <kparal> the logging patch and the dist tag discussion are my next two items on my todo list, I should look at it shortly
14:10:20 <tflink> did I really forget to review D382 on friday?
14:10:24 <tflink> :(
14:11:10 <tflink> yep, I did
14:11:55 <tflink> mkrizek: was the submitted patch for T491 helpful?
14:12:14 <tflink> nvm, slow on reading the comments
14:12:31 <tflink> any other comments/questions?
14:12:53 <mkrizek> :)
14:12:53 <kparal> not here
14:13:22 <tflink> #topic tflink status update
14:13:22 <tflink> #info trying to help with testcloud packaging
14:13:22 <tflink> #info fedmenu fun with taskotron landing pages
14:13:23 <tflink> #info made some progress on better phabricator packaging
14:13:25 <tflink> #info did some code reviews
14:14:23 <tflink> roshi hit a bit of a snag with testcloud packaging and the same issue will likely affect libtaskotron to a lesser extent when we submit it for inclusion in the Fedora repos
14:15:03 <mkrizek> what's the issue?
14:15:08 <tflink> 777 dirs
14:15:26 <roshi> letting the user have access to a directory as well as libvirt
14:15:36 <kparal> I think you will need to download to user-local dirs and then run the image with qemu:///session instead of ///system
14:15:38 <tflink> for testcloud, it stores instance data and backing stores in /var/lib/testcloud w/ 777
14:15:48 <tflink> kparal: there are other problems with session, though
14:15:56 <kparal> or ask the user to add perms for qemu
14:15:59 <tflink> and other problems with user dirs
14:16:06 <roshi> networking is constrained for session
14:16:17 <kparal> have you found something not working?
14:16:24 <kparal> I'd be interested to know
14:16:32 <tflink> i suspect that we're going to take a group-membership approach unless roshi found something else on Friday that I'm forgetting
14:16:39 <tflink> kparal: something?
14:16:42 <roshi> well, you still have permissions issues with user dirs
14:16:44 <kparal> in networking
14:16:55 <roshi> host to guest doesn't work ootb
14:17:01 <kparal> I tried qemu session and everything worked except for pinging servers
14:17:04 <roshi> guest to internet, that works
14:17:12 <roshi> ICMP is blocked
14:17:25 <roshi> it's the same networking that the old testcloud had with just qemu
14:17:37 <kparal> so in session you can't ssh into the VM?
14:17:41 <kparal> I think that worked for me
14:17:45 <roshi> had to do port mapping at invocation to get ssh to work
14:17:45 <tflink> from anywhere but the host, no
14:17:50 <roshi> but out of the box, nope
14:18:04 <kparal> oh, anywhere but the host, ok. I tried from the host
14:18:16 <tflink> but the bigger problem is still file permissions
14:18:37 <tflink> qemu:///system vd qemu:///session wasn't the biggest issue, IMHO
14:19:14 <tflink> i think it's more of a choice in implementation and side-effects
14:19:27 <kparal> virt-managers offers to automatically set acl on affected dirs. not sure if it is part of libvirt or just virt-manager
14:19:43 <tflink> kparal: i think it's libvirt
14:20:02 <tflink> bah, read too quickly - not sure what you're talking about
14:20:08 <roshi> http://wiki.libvirt.org/page/FAQ#What_is_the_difference_between_qemu:.2F.2F.2Fsystem_and_qemu:.2F.2F.2Fsession.3F_Which_one_should_I_use.3F
14:20:48 <kparal> tflink: when I try to boot from .iso file that is not accessible for qemu process (wrong permissions), virt-managers offers me to "fix" the path - it adds acls for qemu for every directory in the path
14:21:08 <tflink> ah, I suspect that's virt-manager
14:21:19 <kparal> it has to be something running under root, in any case
14:21:48 <kparal> hmm, maybe not root, my own process is enough
14:22:16 <tflink> admin privs, I suspect
14:22:39 * kparal never really looked into details
14:23:15 <roshi> pretty much all the docs say "use ///system, it'll add years to your life"
14:23:22 <roshi> that I've read, anyways
14:23:39 <randomuser> fwiw, we have a guy writing a python libvirt guide right now, with a fair amount of experience with the subject
14:23:39 <tflink> #link https://bugzilla.redhat.com/show_bug.cgi?id=1234649
14:24:06 <roshi> randomuser: awesome, link?
14:25:45 <randomuser> https://git.fedorahosted.org/cgit/docs/libvirt_application_development_guide_using_python.git/
14:26:10 <tflink> randomuser: cool, thanks
14:26:11 <kparal> roshi: if you require the user to be in the libvirt group, then it will be able to run VMs with qemu:///system
14:26:44 <randomuser> it's a friendly fork of a C guide from upstream - I bring it up mainly to present the writer as a resource at the moment
14:26:54 <kparal> tflink: mkrizek: jskladan still has PTO, he just failed to mark it properly in the calendar
14:27:05 <tflink> kparal: there are other ways around that but IIRC, there are some permissions issues with just using libvirt regardless of the qemu session
14:27:11 <roshi> thanks randomuser
14:27:53 <kparal> ok, just suggesting :) that libvirt group is a new feature, very recent
14:28:17 * tflink is mostly wary of using mechanisms that we have little/no control over :)
14:28:43 <tflink> any other questions/comments?
14:28:59 <mkrizek> not here
14:29:41 <tflink> k, moving on
14:29:50 <tflink> #topic Artifacts Deployment
14:30:13 <tflink> from the ticket status, it looks like building and initial deployment can start
14:30:36 <tflink> https://phab.qadevel.cloud.fedoraproject.org/T399
14:30:39 <kparal> I have added some notes into T505
14:31:09 <tflink> I think we have commits pending in most of our projects
14:31:28 <tflink> kparal: i think that rpmlint was the last one not using develop
14:31:37 <tflink> mostly because it hasn't changed in a long time :)
14:31:42 <kparal> that's true
14:32:38 <tflink> anyhow, I'd like to get at least dev updated this week if everything should be done enough
14:34:24 * tflink takes that as "no objections"
14:34:33 <kparal> no objections
14:34:35 <kparal> :)
14:34:49 <tflink> may wait for josef to get back, though. IIRC, there are several pending reviews for resultsdb/resultsdb_frontend
14:35:10 <kparal> is some of that needed for artifacts?
14:35:20 <tflink> directly, not so much
14:35:41 <tflink> but we've been asked to add the "fedmenu" stuff to our apps like the rest of infra hs
14:35:42 <tflink> has
14:36:23 <tflink> some of the patches are to add fedmenu, another is to make bodhi's querying of resultsdb easier
14:37:06 <kparal> and then there's the execdb patch to add indexes
14:37:19 <tflink> yeah, that too
14:37:46 <kparal> if you think it's better to take it all in one go, let's wait a bit then
14:38:21 <tflink> i just want to get them deployed before too much longer :)
14:38:35 <tflink> f23 alpha freeze is going to be here before we know it
14:39:09 <tflink> branch is in 2 weeks
14:39:12 * kparal checks schedule
14:39:16 <tflink> well, 2 weeks from tomorrow
14:39:21 <tflink> https://fedoraproject.org/wiki/Releases/23/Schedule
14:39:32 <kparal> 2015-07-28  Alpha Freeze
14:39:52 <tflink> i still hold with "it'll be here before we know it" :)
14:40:27 * kparal concurs
14:40:45 <kparal> there will be a lot of PTOs in the next weeks
14:40:59 <roshi> and we have a short week in the US this week
14:41:16 <roshi> July 3rd is a holiday for RH in the US
14:41:24 <kparal> ok
14:41:51 <mkrizek> July 6th is a holiday in the cz :)
14:41:53 <kparal> and there seems to be a canada day on wednesday - that will probably affect just canada, though ;)
14:42:16 <tflink> in other words, it's a perfect time to make major changes :)
14:42:19 <kparal> mkrizek: right, I'll put it into PTO calendar
14:43:20 <tflink> anyhow, I'd like to get some releases done this week or next week
14:43:57 <tflink> apparent forward progress, not testing too many things at the same time etc.
14:44:32 <tflink> any other thoughts/comments/questions on the topic?
14:45:00 * kparal shakes his head
14:45:11 <tflink> kparal: in dismay or as a "no"?
14:45:20 <tflink> :-D
14:45:21 <kparal> as a no :)
14:45:22 <mkrizek> :D
14:45:28 <mkrizek> nothing here
14:45:30 <tflink> ok, moving on
14:45:45 <tflink> #topic Support Infrastructure and Tooling
14:46:03 <tflink> just wanted to remind folks of the background-ish stuff that's going on and will be happening soon
14:46:12 <tflink> 1) moving qadevel
14:46:29 <tflink> the old machine is going away whether we like it or not
14:47:00 <tflink> i have a new stg instance up and running, please poke at it - https://phab.qadevel.stg.fedoraproject.org/
14:47:17 <kparal> there are fixed scrollbars
14:47:19 * kparal cheers
14:47:44 * kparal votes for deploying right now
14:48:15 <tflink> since we're going to be changing hostname either way (dealing with redirects, etc), I've been toying with the idea of deploying as phab.fedoraproject.org and opening it up to other interested fedora groups
14:48:37 <kparal> that's something I mentioned in the email to qa-devel today
14:48:57 <tflink> yeah, I read that right before the meeting, haven't responded yet
14:49:14 <kparal> I'm not sure it's a good idea for us to be spending the time maintaining a full-fledged fedora service?
14:49:30 <tflink> I'm not sure it's going to make a lot of difference, to be honest
14:50:23 <tflink> unless we scale past the point where one machine can handle phabricator, I suspect that the level of effort on the infra side would be almost the same
14:51:06 <kparal> I'm a bit afraid of a lot of requests from people, and also the number of projects and repos exploding creating "chaos", something hard to navigate in
14:52:05 <kparal> do you think that a single instance for all projects is better than separating it e.g. by teams?
14:52:20 <tflink> I'll take another look at multi-tenant hosting to see how hard that would be
14:52:36 <tflink> kparal: depends on how it's hosted, honestly
14:52:57 <tflink> either way, I'd like to see at least arcanist in the Fedora repos
14:54:04 <kparal> sure
14:54:42 <tflink> also, there is the git hosting discussion on qadevel@
14:55:01 <tflink> bah, this topic took longer than I was intending
14:55:16 <tflink> trying to wrap up quickly with
14:55:19 <tflink> #topic tasks
14:55:30 <tflink> is anyone looking for tasks to do?
14:56:29 * kparal has reviews, deployment patches and then back to disposable clients integration
14:56:30 * mkrizek still has plenty to do
14:56:51 <tflink> ok, just making sure :)
14:57:30 <tflink> and I think this brings  us to ...
14:57:34 <tflink> #topic open floor
14:57:41 <tflink> anything else to cover in the last 3 minutes?
14:57:55 <kparal> nothing from me
14:58:56 * mkrizek has nothing
14:58:58 * tflink dusts off a patent-pending-non-deterministic-fedora-qa-fuse and lights it
14:59:20 <tflink> thanks for coming, everyone
14:59:26 * tflink will send out minutes shortly
14:59:48 <tflink> #endmeeting