13:00:29 <mclasen> #startmeeting Silverblue
13:00:29 <zodbot> Meeting started Mon Oct  1 13:00:29 2018 UTC.
13:00:29 <zodbot> This meeting is logged and archived in a public location.
13:00:29 <zodbot> The chair is mclasen. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:29 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
13:00:29 <zodbot> The meeting name has been set to 'silverblue'
13:00:37 <mclasen> #meetingname teamsilverblue
13:00:37 <zodbot> The meeting name has been set to 'teamsilverblue'
13:00:43 <mclasen> #topic Roll call
13:01:01 <mclasen> good morning
13:01:04 <mclasen> who do we have today ?
13:01:09 <walters> .hello2
13:01:10 <zodbot> walters: walters 'Colin Walters' <walters@redhat.com>
13:01:27 <aday> o/
13:01:34 <rishi> .hello2
13:01:35 <zodbot> rishi: rishi 'Debarshi Ray' <debarshir@redhat.com>
13:01:40 <kalev> .hello2
13:01:41 <zodbot> kalev: kalev 'Kalev Lember' <klember@redhat.com>
13:02:25 * juhp waves
13:02:32 <ksinny> .hello sinnykumari
13:02:33 <zodbot> ksinny: sinnykumari 'Sinny Kumari' <ksinny@gmail.com>
13:04:00 <mclasen> alright, lets see. what do we have to discuss ?
13:04:08 <mclasen> #topic agenda
13:04:37 <mclasen> if owen shows up, I'd like to talk about flatpak building
13:05:09 <rishi> I could give a brief status update on the toolbox.
13:05:25 <mclasen> sounds good
13:05:52 <mclasen> there are a few tickets in pagure I can bring up
13:06:37 <mclasen> kisinny: should we talk about the test day outcomes ? anything to report there ?
13:07:16 <ksinny> yeah, can talk about it
13:07:44 <mclasen> ok, if there are no other suggestions, maybe we can start with the toolbox
13:07:47 <mclasen> rishi ?
13:07:52 <mclasen> #topic toolbox status update
13:08:25 <rishi> Ok. So, as of last week, the fedora-toolbox images are now in the cadidate registry.
13:08:36 <rishi> ie., candidate-registry.fedoraproject.org.
13:08:52 <mclasen> what is the candidate registry ? is that like updates-testing ?
13:08:59 <rishi> Yeah, like updates-testing.
13:09:21 <rishi> Builds are supposed to automatically get moved to registry.fp.o every two weeks.
13:09:34 <mclasen> a good step forward!
13:09:43 <rishi> Once that happens it will remove the first step from this README.md: https://github.com/debarshiray/fedora-toolbox
13:10:21 <rishi> However, the client-side tools are still rocky.
13:10:35 <mclasen> I've experienced that (again) on friday
13:10:36 <rishi> https://github.com/containers/libpod/issues?utf8=%E2%9C%93&q=is%3Aissue+author%3Adebarshiray+
13:10:57 <rishi> Currently I am aware of those two open issues in that list.
13:11:22 <rishi> As for the fixed ones, we need to ensure that those make their way into the Silverblue image.
13:11:40 <mclasen> they will if they make it into fedora builds, right ?
13:12:09 <rishi> If those Fedora builds make their way into the stable updates repository, I guess.
13:12:38 <rishi> I suppose just a build sitting in Koji isn't enough for anything that's not Rawhide. Am I wrong?
13:12:45 <mclasen> true
13:12:52 <mclasen> it needs to get through bodhi and into updates
13:13:23 <mclasen> did podman gain regression tests for any of the things you've dealt with so far ?
13:13:26 <juhp> In the meantime one could workaround it locally?
13:13:27 <rishi> The reason I brought it up is that podman upstream is a bit fast-moving -- upstream releases every Friday, and Fedora builds even more frequently.
13:13:51 <rishi> Some of these builds might even be automated. I am not sure. Guessing from the commit messages. :)
13:14:09 <rishi> Which is all good, but sometimes this prevents an update from ever making it into stable.
13:14:18 <rishi> One update resets the next, and so on.
13:14:32 <juhp> ah
13:14:32 <rishi> Or, sometimes some builds don't get an update filed.
13:14:46 <rishi> So, maybe we should keep an eye on that.
13:15:01 <mclasen> I guess that is a problem with the current bodhi setup
13:15:31 <mclasen> sometimes you want to replace an update with a follow-up fix. but sometimes you just want the old update to go out and be able to stage the next one early
13:15:36 <juhp> Better to use rawhide then ;o)
13:15:45 <rishi> mclasen: Yeah.
13:15:46 <juhp> mclasen: right
13:15:49 <otaylor> Unfortunatley, rpm-ostree doesn't have module support, or I'd suggest that we could have a container-tools module pulled into the silverblue compose
13:16:19 <rishi> Plus the 'rpm-ostree override ...' dance makes it a bit more tricky to workaround this on Silverblue.
13:16:38 <rishi> Re: tests for regressions in rootless Podman ...
13:16:42 <mclasen> otaylor: is that something on the rpm-ostree roadmap ?
13:16:52 <rishi> There's some talk of enhancing the test suite: https://github.com/containers/libpod/issues/1522
13:17:16 <rishi> Plus some existing tests did get tweaked as part of the fixes that have gone out so far. So, that's encouraging.
13:17:18 <mclasen> I should ask walters, maybe
13:17:38 <rishi> Oh, I forgot! runc!
13:17:46 <mclasen> what about it ?
13:17:47 <walters> er...i don't understand the container-tools module, would we really ever have parallel versions?
13:17:51 <walters> and wouldn't that just exacerbate the bodhi karma problem for the "ursine" rpm?
13:17:54 <walters> but yes that's https://github.com/projectatomic/rpm-ostree/issues/1435
13:18:16 <rishi> So, we need to get some movement on https://github.com/opencontainers/runc/pull/1862
13:18:30 <rishi> That's needed for getting out UID mapping to work.
13:18:49 <otaylor> walters: OK, maybe not a a well-thought-through idea, but my idea was that Silverblue could "opt in" to a more aggresively updated set of container-tools versions than what we carry in f29. (Though why not carry the same things in f29....)
13:19:09 <rishi> I was told that getting runc reviews is difficult and we have limited leverage on that project.
13:19:32 <mclasen> docker all over again ? :(
13:19:44 <walters> i've very often thought about decoupling atomic delivery from all of the existing infrastructure, and this applies to silverblue too, but it opens up a huge number of questions
13:20:24 <otaylor> rishi: didn't Guiseppe say that we're carrying that locally for fedora?
13:20:40 <rishi> otaylor: That's another PR of his. Not this one.
13:20:44 <otaylor> rishi: Ah
13:21:13 <rishi> mclasen: I learnt from Giuseppe that mrunalp is the only pawn we have in this game. :)
13:21:36 <rishi> So, that's all I have.
13:21:49 <otaylor> walters: I think the karma problem is fixable in isolation - if we have a container tools bodhi update being filed every friday (or whatever), let's figure out how to make it go out reliably
13:21:56 <mclasen> ok, so it sounds like getting a working toolbox in f29 is mostly up to dependencies
13:22:01 <rishi> In short: we need to find a way forward with https://github.com/containers/libpod/issues/1522 and work our way through rootless podman issues.
13:22:42 <mclasen> thanks for the update, rishi
13:22:52 <mclasen> anything else on this topic, or should we move on ?
13:23:16 <mclasen> ok, next topic
13:23:23 <mclasen> #topic test day feedback
13:23:39 <mclasen> ksinny, do you want to give a quick update ?
13:24:23 <ksinny> Fedora Silverblue Test Day went well, we had good number of testers this time
13:25:06 <ksinny> Test results are availbale at http://testdays.fedorainfracloud.org/events/47 (with comments/bug if any)
13:25:20 <ksinny> Few common issues I noted were:
13:25:28 <ksinny> 1. Slow rebase for few people from F28 to F29 silverblue.mirror use mostly depending upon region?
13:25:28 <mclasen> #info test day results are here: http://testdays.fedorainfracloud.org/events/47
13:25:43 <ksinny> 2. Issues like  "Installing applications from gnome nightly, like Calendar or Contacts, I get "Unable to install Calendar as runtime sdk.gnome.org not available"
13:25:53 <ksinny> 3. Long time for the Flathub packages to appear in Gnome Software - not sure how long (they appeared the same day at least)
13:26:02 <ksinny> 4. I am saying this passes because it does notify you that there are updates (whether kernel or flatpak) however, if you try to use software to perform the actual update it fails, everytime.
13:26:17 <mclasen> that looks like a good amount of testers, indeed!
13:26:24 <ksinny> For the 1st one, I think it is realted to mirror depending upon where the person is
13:26:33 <ksinny> I too face this issue
13:26:48 <mclasen> sorry to hear :(
13:27:13 <ksinny> For 2nd issue, I think people may night to add some flatpak repo?
13:27:48 <mclasen> that might be a lack on RuntimeRepo in those flatpakref files
13:28:02 <mclasen> it has been pointed out as an issue on the flatpak side that we allow that
13:28:16 <ksinny> hmm, is that an issue or something needs to be done from user end?
13:28:17 <juhp> ah
13:28:23 <ksinny> ah ok
13:28:39 <mclasen> well, for now you'll need to make sure that you have a repo for the runtime enabled
13:28:49 <ksinny> For the 3rd issue I am not sure what could be the problem, i didn't see it myself
13:28:55 <mclasen> the RuntimeRepo key makes that automatic, by telling g-s/flatpak where the runtime can be found
13:29:37 <ksinny> mclasen: We may need to document this issue during F29 GA
13:29:40 <mclasen> maybe kalev has an idea. I'd be willing to just blame gnome-software on the third
13:29:47 <juhp> Is there an issue tracking that?
13:29:52 <mclasen> ksinny: yes, good point
13:30:05 <kalev> mclasen: I think it's just gnome-software failing. it used to work, but something has regressed
13:30:07 <juhp> (I mean 2)
13:30:25 <juhp> erm 3
13:30:35 <mclasen> this is the flatpak issue: https://github.com/flatpak/flatpak/issues/2120
13:30:36 <ksinny> juhp: I am not aware of
13:30:57 <ksinny> Mostly, folks added issues in comment instead of creating on pagure/bugzilla
13:31:09 * juhp confused 2 for 3 reading above...
13:31:11 <mclasen> #action mclasen will make sure we have  silverblue issues for the top test day outcomes
13:31:17 <juhp> Okay I try to retest later anyway
13:31:59 <mclasen> ksinny: anything more on the test day ?
13:32:18 <ksinny> For the fourth one, I too noticed that Gnome software tells about an update is available for OS with all cahnes, but when I click on update and reboot stuff
13:32:38 <ksinny> It just reboot system and doesn't upgrade system to latest update available
13:32:48 <ksinny> mclasen: That's all
13:32:57 <mclasen> thanks, ksinny
13:33:20 <ksinny> np!
13:33:30 <mclasen> kalev: ^ thats the same thing I hit this morning
13:34:23 * ksinny is used to of doing update from connandline though
13:34:23 <mclasen> anyway, moving on
13:34:49 <mclasen> otaylor: should we talk about flatpak building ?
13:34:57 <mclasen> #topic flatpak builds
13:35:25 <mclasen> I made two attempts at building a flatpak as a module
13:35:46 <mclasen> first I tried on silverblue, and got stuck trying to work out nested containers and whatnot
13:36:06 <mclasen> that probably needs some handholding instructions for contrainer-challenged people like me
13:36:16 <walters> if you hit nested container issues don't hesitate to ask, i'm happy to help
13:36:34 <mclasen> on the weekend I tried again, with a traditional f28 workstation in a vm
13:36:50 <walters> if you're using mock, the main thing is to disable nspawn
13:36:50 <otaylor> I'll work on getting a set of instructions for building on silverblue going - (probably with a podman --privileged big hammer)
13:37:08 <mclasen> in the vm, I got a little further
13:37:34 <otaylor> (if disabling nspawn can be made work, then I can try not using that big hammer)
13:37:45 <mclasen> I could get the module build even though it really brought my laptop to its knees
13:37:56 <mclasen> but the build-container step just failed in a very unclear manner
13:38:08 <otaylor> mclasen: Oh, good doc addition!
13:38:38 <mclasen> some long python backtrace ending in http.client.RemoteDisconnected: Remote end closed connection without response
13:38:48 <mclasen> at that point, I gave up
13:38:59 <otaylor> mclasen: Please send me all the logs
13:39:10 <otaylor> (or I'll just work on getting it going with a container and you can try then)
13:39:19 <mclasen> I also noticed that I could not get the build going as user
13:39:30 <mclasen> it just hangs, with stdout and stderr going to some log files
13:39:36 <otaylor> (    NUM_CONCURRENT_BUILDS = 1 in /etc/module-build-service/config.py needs to be in the docs)
13:39:39 <mclasen> stdout contains an explanation that it needs privileges
13:39:47 <mclasen> and stderr contains a password prompt
13:39:51 <otaylor> mclasen: that's in the docs
13:39:56 <mclasen> killing the process at that point leaves the console disabled
13:40:22 <mclasen> a thorny process...
13:40:22 <otaylor> (https://docs.fedoraproject.org/en-US/flatpak/tutorial/ - under setup)
13:40:39 <otaylor> mclasen: But it could certianly be made to fail better!)
13:40:47 <mclasen> ah, I'm not sure that was there when I looked
13:41:21 <otaylor> mclasen: it's been there for a week or  two
13:41:37 <otaylor> mclasen: but easy to miss
13:42:03 <walters> https://pagure.io/fm-orchestrator/issue/641 ?
13:42:23 <mclasen> did anybody else try and succeed ?
13:42:33 <otaylor> walters: there's a /etc/module-build-service/mock.cfg
13:42:34 <mclasen> I know david king did
13:42:49 <rishi> Not me. I haven't tried building a Flatpak out of Fedora RPMs, yet.
13:43:17 <otaylor> /me would encourage people to try, and will try to provide live help on #fedora-workstation
13:43:29 <mclasen> yes, please try
13:43:35 * mclasen will try again today
13:43:50 <rishi> Yeah, I have it on my list of things to try this week.
13:44:08 <mclasen> otaylor: do we have a way to install apps from the registry yet ?
13:44:11 <otaylor> I'm working on a flatpak-common module that will make things faster and make more applications easy. It's also shaking out some problems in spec files
13:44:53 <otaylor> mclasen: Not yet. First thing in this morning is to work on testing bodhi out in fedora staging infrastructure - it's supposed to be working there now for flatpaks
13:46:49 <mclasen> ok, some progress at least
13:47:34 <mclasen> anything else on flatpak building, or should we move on ?
13:48:18 <mclasen> ok, moving on to tickets
13:48:38 <mclasen> #topic https://pagure.io/teamsilverblue/issue/36 can't install docker
13:49:15 <mclasen> I guess this is just a case of "too many repos enabled" ?
13:49:52 <walters> it's just another instance of https://github.com/projectatomic/rpm-ostree/issues/415 i'm fairly sure
13:50:22 <mclasen> yep
13:50:35 <mclasen> one thing of interest in the ticket is that dusty mentions a testing ref
13:50:39 <mclasen> I hadn't seen that before
13:50:44 <walters> there's still work ongoing on rojig, though it's behind other PRs right now
13:50:52 <mclasen> fedora/29/x86_64/testing/silverblue
13:52:08 <mclasen> is this something new ?
13:52:12 <mclasen> did we have that all along ?
13:54:40 <mclasen> I guess nobody knows
13:54:59 <mclasen> maybe time for one more ticket
13:55:13 <mclasen> #topic https://pagure.io/teamsilverblue/issue/41 disable bootloader_menu_autohide
13:55:22 <mclasen> I've stated my opinion in the issue
13:56:28 <aday> hidden menu is what we're aiming for as the proper ux
13:57:08 <mclasen> I don't think there's a good justification to deviate from the workstation setup for it
13:58:22 <mclasen> any more opinions on this ?
13:58:24 <rishi> mclasen: aday: Does it prevent users from rolling back by booting into the non-default image?
13:58:42 <mclasen> you have to hold down the right key, I guess
13:58:43 <rishi> I haven't tried F29, so no idea how it actually looks in practice.
13:59:11 <mclasen> you can tell rpm-ostree to switch the default boot entry before rebooting, too
13:59:26 <otaylor> And we have failed boot logic, right?
14:00:36 <mclasen> do we ? I am not sure, tbh
14:01:11 <mclasen> #action mclasen to investigate the status of failed boot logic
14:01:56 <mclasen> we also have the bls and the no-grub changes coming, but I am not certain if either of those will be in f29 (or on by default)_
14:02:36 <mclasen> but now we're past the hour, so thats it for today
14:02:42 <mclasen> thanks, everybody
14:02:46 <mclasen> #endmeeting