fedora-bootc-meeting
LOGS
<@jbrooks:matrix.org>
15:00:24
!startmeeting fedora_bootc_meeting
<@meetbot:fedora.im>
15:00:27
Meeting started at 2025-03-11 15:00:24 UTC
<@meetbot:fedora.im>
15:00:27
The Meeting name is 'fedora_bootc_meeting'
<@jbrooks:matrix.org>
15:00:36
!topic roll call
<@rsturla:fedora.im>
15:00:44
!hi
<@zodbot:fedora.im>
15:00:46
None (rsturla)
<@hricky:fedora.im>
15:01:00
!hi
<@zodbot:fedora.im>
15:01:02
Hristo Marinov (hricky) - he / him / his
<@walters:fedora.im>
15:01:04
!hi
<@zodbot:fedora.im>
15:01:06
Colin Walters (walters)
<@jeckersb:fedora.im>
15:01:09
!hi
<@zodbot:fedora.im>
15:01:12
John Eckersberg (jeckersb)
<@mmartinv:matrix.org>
15:01:30
!hi
<@zodbot:fedora.im>
15:01:34
Miguel Martin (mmartinv)
<@sean:thrailkill.cloud>
15:01:40
!hi
<@zodbot:fedora.im>
15:01:42
Sean Thrailkill (snthrailkill)
<@miabbott:fedora.im>
15:01:49
!hi
<@zodbot:fedora.im>
15:01:50
Micah Abbott (miabbott)
<@pwhalen:fedora.im>
15:02:14
!hi
<@zodbot:fedora.im>
15:02:16
Paul Whalen (pwhalen)
<@jlebon:fedora.im>
15:02:26
!hi
<@zodbot:fedora.im>
15:02:28
None (jlebon)
<@jbrooks:matrix.org>
15:02:47
How's everyone doing today?
<@walters:fedora.im>
15:03:20
The agenda is https://etherpad.opensuse.org/p/bootc-initiative-meetings and it looks like we have something from mmartinv ?
<@siosm:matrix.org>
15:04:12
!hi
<@zodbot:fedora.im>
15:04:28
Timothée Ravier (siosm) - he / him / his
<@jbrooks:matrix.org>
15:04:30
For topics today, should we continue talking about the local layering issue? Or are there other topics on our minds?
<@jbrooks:matrix.org>
15:05:28
!topic Update from Miguel Martin on building Fedora bootc in the community Konflux cluster
<@walters:fedora.im>
15:06:51
cc gursewak re ⬆️
<@mmartinv:matrix.org>
15:06:53
I have updated the https://gitlab.com/fedora/bootc/base-images/-/merge_requests/70 with the latest proposal
<@mmartinv:matrix.org>
15:07:55
Waiting on https://github.com/konflux-ci/build-definitions/pull/1998 so we can refer the upstream pipeline instead of the patched one
<@mmartinv:matrix.org>
15:08:26
Not sure if it will get accepted though, maybe we need to add the pipeline to the repo
<@walters:fedora.im>
15:08:42
Thanks so much for doing this! A lot to build on top of this, including especially things like keeping this pipeline in sync with what we have for centos stream, etc.
<@walters:fedora.im>
15:08:50
Can we get in sync on the subthread from https://gitlab.com/fedora/bootc/base-images/-/merge_requests/70#note_2385156856 ?
<@walters:fedora.im>
15:09:49
This whole effort started before I did work on the customizable base images; I think my preference here would be to start with the konflux build of what is now `standard` and also `minimal`
<@walters:fedora.im>
15:11:24
This also relates t ohttps://github.com/coreos/fedora-coreos-tracker/issues/1861
<@walters:fedora.im>
15:11:27
*relates to
<@siosm:matrix.org>
15:11:34
It would be good if we could figure out a way to make the build pipeline logs public (I don't have access as far as I can see). But that can happen later.
<@walters:fedora.im>
15:11:37
This also relates to https://github.com/coreos/fedora-coreos-tracker/issues/1861
<@jlebon:fedora.im>
15:12:44
Colin Walters: do you perceive a high cost from building three instead of two?
<@jlebon:fedora.im>
15:12:53
at least for the time being until we have a clearer story around CI
<@walters:fedora.im>
15:12:58
Yeah, the FCOS Jenkins has the same issue still right?; but by "public" you mean "unauthenticated" specifically right? They're available to anyone with a FAS account AFAIK
<@siosm:matrix.org>
15:13:51
I logged in with my FAS and could not access the logs
<@walters:fedora.im>
15:13:53
No, I wouldn't call it a high cost. But it's more a question of priorities and ordering I think
<@siosm:matrix.org>
15:14:10
If I go to koji right now in the Fedora Infra, everything is public
<@jlebon:fedora.im>
15:15:14
it's also not just about packages I would emphasize. any glue code/workarounds/etc... related to the packages in that list can be more easily shared if it's a separate target and not intermingled with e.g. `standard`
<@siosm:matrix.org>
15:15:16
FAS gated would be OK for me, but ideally we should not re-create the Jenkins situation that we have in FCOS, which is mainly because we don't want to expose Jenkins which has a terrible security track record
<@walters:fedora.im>
15:15:24
But I guess to ask directly, since it's not explained in the commit message but I can infer it: was the idea here that Fedora IoT would be `FROM quay.io/fedora/fedora-bootc:tierx` or so? Is there a tracker for that?
<@jlebon:fedora.im>
15:15:30
so it's about maintenance of that package set
<@mmartinv:matrix.org>
15:16:04
travier: you need to create an MR to get permissions granted, see https://gitlab.com/fedora/infrastructure/konflux/tenants-config/-/merge_requests/9/diffs for reference
<@jlebon:fedora.im>
15:16:05
Colin Walters: that was what i expected (and similar for FCOS and Atomic Workstations eventually)
<@walters:fedora.im>
15:16:15
But it is the case currently AFAIK that there are no workarounds for anything not in minimal, right?
<@siosm:matrix.org>
15:16:19
mmartinv: Yes, the idea is that I shouldn't have to do that
<@jlebon:fedora.im>
15:17:02
Colin Walters: currently, yes. but tier-x/minimal-plus ships openssh, NM, rpm-ostree, polkit, and a few other biggies
<@rsturla:fedora.im>
15:17:06
Unless you can use a wildcard the user field to provide all Konflux users access?
<@jlebon:fedora.im>
15:17:09
one can easily imagine things needed for those
<@miabbott:fedora.im>
15:17:27
travier: i think we'd have to ask the Konflux folks if that restriction could be relaxed
<@rsturla:fedora.im>
15:17:31
Unless you can use a wildcard the user field to provide all Konflux users access to a particular ro role?
<@miabbott:fedora.im>
15:18:15
maybe an issue on https://gitlab.com/fedora/infrastructure/konflux
<@jbrooks:matrix.org>
15:18:30
there's https://matrix.to/#/#konflux:fedora.im as well
<@rsturla:fedora.im>
15:18:35
Unless you can use a wildcard the user field to provide all Konflux users access to a particular ro role?
<@rsturla:fedora.im>
15:18:35
Oh, it's Kubernetes native permissions, not something Konflux specific
<@jlebon:fedora.im>
15:18:38
Colin Walters: anyway, not strongly against not having it to be clear, but it just seems a much cleaner story to me to have an actual image variants build on top
<@siosm:matrix.org>
15:18:51
https://gitlab.com/fedora/infrastructure/konflux/tenants-config/-/issues/3
<@walters:fedora.im>
15:19:00
I put this somewhere but one possible middle ground here is to ship `quay.io/fedora-ci/fedora-bootc-minimal-plus:rawhide` e.g. (note `fedora-ci` i.e. it's "fedora internal")
<@mmartinv:matrix.org>
15:19:10
Yeah, that was the idea, not sure if we have a tracker though, Micah Abbott ?
<@jlebon:fedora.im>
15:19:23
that WFM
<@jlebon:fedora.im>
15:19:44
it seems fine to me if this is only for variants use
<@walters:fedora.im>
15:19:53
I still lean towards trimming down `standard` in the medium term
<@walters:fedora.im>
15:20:25
(Although it's quite complex because some things may need to be added there eventually...like `firewalld`)
<@miabbott:fedora.im>
15:20:33
It probably needs to be updated to reflect reality
<@miabbott:fedora.im>
15:20:33
I think the closest we have is https://github.com/fedora-iot/iot-distro/issues/53
<@miabbott:fedora.im>
15:20:33
<@pwhalen:fedora.im>
15:21:28
@micah I'll try to update that this week.
<@jbrooks:matrix.org>
15:21:52
Let's log some action items on these
<@miabbott:fedora.im>
15:24:17
<@miabbott:fedora.im>
15:24:17
?
<@miabbott:fedora.im>
15:24:17
For actions, i see something like "building quay.io/fedora-ci/fedora-bootc-minimal-plus:rawhide" and "updating https://github.com/fedora-iot/iot-distro/issues/53 with reality"
<@miabbott:fedora.im>
15:24:50
<@miabbott:fedora.im>
15:24:50
For actions, i see something like "building quay.io/fedora-ci/fedora-bootc-minimal-plus:rawhide" and "updating https://github.com/fedora-iot/iot-distro/issues/53 with reality"
<@miabbott:fedora.im>
15:24:50
anything else?
<@jbrooks:matrix.org>
15:24:58
!action pwhalen to update https://github.com/fedora-iot/iot-distro/issues/53 to reflect reality
<@jlebon:fedora.im>
15:25:15
quay.io/fedora-ci is not a thing currently. i guess it could be under quay.io/fedora but just not in the main fedora-bootc repo
<@walters:fedora.im>
15:25:29
I edited the description of https://gitlab.com/fedora/bootc/base-images/-/issues/25
<@walters:fedora.im>
15:25:51
Ah sorry I meant https://quay.io/organization/fedoraci which is where ELN is
<@walters:fedora.im>
15:26:01
https://quay.io/repository/fedoraci/fedora?tab=tags
<@jlebon:fedora.im>
15:26:16
gotcha
<@walters:fedora.im>
15:26:40
Oh or at least it *was*...looks like it moved https://docs.fedoraproject.org/en-US/eln/deliverables/#_container_image
<@walters:fedora.im>
15:28:23
OK so we're hopefully closing in on different projects deriving from bootc base images finally
<@jbrooks:matrix.org>
15:28:50
OK, is there more on this topic?
<@jbrooks:matrix.org>
15:29:44
Do we have other items to discuss today?
<@jlebon:fedora.im>
15:30:18
not sure if i surfaced this in this meeting yet, but i did have a rough PR in https://github.com/coreos/fedora-coreos-config/pull/3348 of rebasing FCOS on top of minimal-plus
<@jlebon:fedora.im>
15:30:18
there's some work needed on the tooling side to enable this
<@jbrooks:matrix.org>
15:30:39
!topic open floor
<@jbrooks:matrix.org>
15:31:24
Is minimal plus the same as tier-x?
<@jlebon:fedora.im>
15:31:48
what you'll notice in that PR that's neat is that there's not much actually going on in the `Containerfile`. we're reusing all the manifests we already have, just applied at container build time
<@jlebon:fedora.im>
15:32:02
Jason Brooks: yeah, that's the proposed new name
<@jbrooks:matrix.org>
15:32:34
That's cool w/ the reused manifests
<@jbrooks:matrix.org>
15:33:56
All right, anything else? Should we draw this episode to a close?
<@jbrooks:matrix.org>
15:34:37
OK, we'll follow up w/ anything else in the bootc room
<@jbrooks:matrix.org>
15:34:41
!endmeeting