#meeting-1:fedoraproject.org log

<@jbrooks:matrix.org>

15:00:32

!startmeeting fedora_bootc_initiative

<@meetbot:fedora.im>

15:00:35

Meeting started at 2025-04-08 15:00:32 UTC

<@meetbot:fedora.im>

15:00:35

The Meeting name is 'fedora_bootc_initiative'

<@jbrooks:matrix.org>

15:01:09

!topic roll call

<@rsturla:fedora.im>

15:01:15

!hi

<@zodbot:fedora.im>

15:01:16

None (rsturla)

<@hricky:fedora.im>

15:01:45

!hi

<@zodbot:fedora.im>

15:01:46

Hristo Marinov (hricky) - he / him / his

<@walters:fedora.im>

15:01:53

!hi

<@zodbot:fedora.im>

15:01:54

Colin Walters (walters)

<@miabbott:fedora.im>

15:02:09

!hi

<@zodbot:fedora.im>

15:02:10

Micah Abbott (miabbott)

<@jbrooks:matrix.org>

15:02:24

meetbot is laggy...

<@dustymabe:matrix.org>

15:02:47

!hi

<@zodbot:fedora.im>

15:02:50

Dusty Mabe (dustymabe) - he / him / his

<@jeckersb:fedora.im>

15:03:33

!hi

<@zodbot:fedora.im>

15:03:34

John Eckersberg (jeckersb)

<@jmarrero:matrix.org>

15:04:53

!hi

<@zodbot:fedora.im>

15:05:05

Joseph Marrero (jmarrero)

<@jbrooks:matrix.org>

15:05:05

All right, Robert Sturla would like to talk about what's involved in using bootc in other distros, and Colin Walters said we might talk about it at a high level.

<@jbrooks:matrix.org>

15:05:27

Sean Thrailkill: Wants to talk about composefs status

<@jbrooks:matrix.org>

15:05:32

Other topics for today?

<@snthrailkill:matrix.org>

15:05:44

!hi

<@zodbot:fedora.im>

15:05:47

No Fedora Accounts users have the @snthrailkill:matrix.org Matrix Account defined

<@jbrooks:matrix.org>

15:07:06

!topic what's involved in using bootc in other distros

<@jbrooks:matrix.org>

15:07:44

Robert Sturla: , would you summarize what you're looking to learn?

<@walters:fedora.im>

15:07:58

I'll comment in some of these but lately my highest level though here is to enable systemd-boot, which sidesteps bootupd entirely

<@walters:fedora.im>

15:08:40

https://github.com/ostreedev/ostree/pull/3359 is in progress...slowly but I am trying to help

<@rsturla:fedora.im>

15:09:09

<@rsturla:fedora.im>

15:09:09

So there's already an issue in bootupd for removing a dependency on RPM - https://github.com/coreos/bootupd/issues/468

<@rsturla:fedora.im>

15:09:09

There have been quite a few conversations across bootc and bootupd issues where people are trying to get a bootc machine running on other distros (mainly Arch), so I would like it if we could get some decisions or additional guidance on how this should be implemented before community members go off down the wrong direction

<@rsturla:fedora.im>

15:11:11

For example, specifically for the bootupd issue above, Colin mentioned each distro shipping their own scripts which can extract version information from the kernels so we don't need to hard-code everything within bootupd itself

<@rsturla:fedora.im>

15:11:49

Oh, integrating with systemd-boot would solve a lot of the issues people have been facing!

<@conan_kudo:matrix.org>

15:12:31

as long as UEFI is the only target, yes

<@conan_kudo:matrix.org>

15:12:41

!hi

<@zodbot:fedora.im>

15:12:43

Neal Gompa (ngompa) - he / him / his

<@conan_kudo:matrix.org>

15:13:15

and that you're working with x86, other arches get "special"

<@rsturla:fedora.im>

15:13:59

I guess my questions mainly revolve around requirements in bootupd, but if there's going to be an option to bypass this entirely, I'm more than happy with that.

<@conan_kudo:matrix.org>

15:14:36

even with sd-boot, bootupd is still relevant because the UEFI environment is not necessarily "real" and offered by hardware, those pieces still need to be handled

<@dustymabe:matrix.org>

15:14:54

Does that mean bootupd wouldn't exist and systemd-boot would take care of everything in the long term future?

<@walters:fedora.im>

15:15:45

I don't think we can drop grub

<@walters:fedora.im>

15:15:50

The larger "we" here

<@walters:fedora.im>

15:16:03

As Neil said, there's BIOS to consider, not to mention upgrade scenarios

<@walters:fedora.im>

15:16:28

Also https://github.com/coreos/bootupd/issues/766 touches on the fact that there's a whole lot of other non-UEFI special things

<@conan_kudo:matrix.org>

15:16:31

and at least generally speaking as a maintainer of a number of Fedora deliverables, I don't have a compelling reason to use sd-boot when I need to use grub for everything else anyway

<@conan_kudo:matrix.org>

15:17:22

I think we previously discussed the situation with Fedora ARM being uniquely terrible for bootloader updates

<@conan_kudo:matrix.org>

15:17:52

I want us to leverage bootupd to make the experience for Fedora ARM the same as Fedora Asahi Remix... that is, completely transparent and managed properly

<@conan_kudo:matrix.org>

15:18:21

I also expect this problem coming in with Fedora RISC-V, which inherits nearly the entirety of the same limitations Fedora ARM has

<@dustymabe:matrix.org>

15:20:13

So moving to sd-boot wouldn’t really enable us to drop anything?

<@conan_kudo:matrix.org>

15:20:22

no, not really

<@dustymabe:matrix.org>

15:20:27

But I guess would make supporting other distros easier?

<@walters:fedora.im>

15:20:30

So...it's hard for me to promise anything in this space but I will try to continue to at least comment on issues like this and I can probably get some time to mentor in a dedicated video call if someone was working on it

<@conan_kudo:matrix.org>

15:20:40

specialized builds and derivatives might want to use it

<@walters:fedora.im>

15:21:09

dustymabe: Yes, for UEFI it all just works and is well designed and already knows how to update itself and report its status etc

<@conan_kudo:matrix.org>

15:21:24

You should reach out to Dan Čermák and discuss it with him, he was interested in bringing it to openSUSE and I think that would help a lot in genericizing the stack for broader adoption.

<@walters:fedora.im>

15:21:37

the rpm dependency in bootupd is a hacky workaround for the grub binaries not being introspectable generically in this way

<@walters:fedora.im>

15:21:59

So far no response in https://github.com/openSUSE/perl-bootloader/issues/186 ...

<@conan_kudo:matrix.org>

15:22:15

he's more active in matrix than github :)

<@conan_kudo:matrix.org>

15:22:26

and he's not the perl-Bootloader maintainer either

<@dustymabe:matrix.org>

15:24:23

Conan Kudo: maybe an email intro would be worthwhile?

<@conan_kudo:matrix.org>

15:24:41

perhaps, I can do that if you want

<@walters:fedora.im>

15:25:02

Anyways overall I promise to go through and respond to bootupd discussions a bit more frequently and hopefully enable anyone who wants to work on this

<@conan_kudo:matrix.org>

15:25:09

we're interested in it from a kiwi-upstream perspective because we'd love a standard interface for bootloader stuff

<@conan_kudo:matrix.org>

15:25:22

everyone is a little weird and abstracting that way would be awesome :)

<@conan_kudo:matrix.org>

15:25:31

everyone is a little weird and abstracting that away would be awesome :)

<@siosm:matrix.org>

15:25:42

!hi

<@zodbot:fedora.im>

15:25:44

Timothée Ravier (siosm) - he / him / his

<@siosm:matrix.org>

15:26:15

The switch to systemd-boot is interesting for Confidential Computing & UKI support. The simpler update is a side benefit

<@rsturla:fedora.im>

15:26:44

I saw your reply in bootupd, and even just the link to the bootc issue helps understand the implementation a little more, even if said implementation is very simple.

<@rsturla:fedora.im>

15:26:44

Makes it easier for myself and other contributors to feel more confident with the route needed to go down.

<@rsturla:fedora.im>

15:26:52

Makes it easier for myself and other contributors to feel more confident with the route needed to go down.

<@rsturla:fedora.im>

15:26:52

I saw your reply in the bootupd issue, and even just the link to the bootc issue helps understand the implementation a little more, even if said implementation is very simple.

<@jbrooks:matrix.org>

15:28:41

OK, this has been a long and winding topic 🙂 Shall we move on to a bit of composefs status?

<@jbrooks:matrix.org>

15:29:26

!topic composefs integration status

<@jbrooks:matrix.org>

15:30:15

I saw a couple of replies in the bootc room about Sean Thrailkill 's query, is there more to discuss here?

<@walters:fedora.im>

15:30:32

I can say that for some of the people at Red Hat working on bootc this area is likely to be high priority and I'm personally excited

<@snthrailkill:matrix.org>

15:30:41

I think I have enough context between the issues Collin linked

<@snthrailkill:matrix.org>

15:30:45

Let me go through that first

<@snthrailkill:matrix.org>

15:30:49

I'll follow up in channel

<@jbrooks:matrix.org>

15:31:38

OK, any open floor topics?

<@jbrooks:matrix.org>

15:31:47

!topic open floor

<@walters:fedora.im>

15:32:16

The rework of the Fedora base image builds is ongoing, and there's testing images being pushed https://gitlab.com/fedora/bootc/base-images/-/merge_requests/156#note_2428860164

<@walters:fedora.im>

15:32:50

Among other things the [from scratch](https://docs.fedoraproject.org/en-US/bootc/building-custom-base/) image flow works in those

<@walters:fedora.im>

15:33:03

But we'll write this up more widely once it's farther along

<@jbrooks:matrix.org>

15:35:17

Awesome, I look forward to it.

<@jbrooks:matrix.org>

15:35:43

Oh, I got a lightning talk about bootc accepted to he upcoming KCD SF Bay Area

<@dustymabe:matrix.org>

15:36:02

Nice work Jason Brooks

<@dustymabe:matrix.org>

15:36:07

are you going to be at Flock this year?

<@jbrooks:matrix.org>

15:36:18

https://community.cncf.io/events/details/cncf-kcd-sf-bay-area-presents-kcd-san-francisco-bay-area/cohost-kcd-sf-bay-area

<@jbrooks:matrix.org>

15:36:53

I am not currently planning on attending Flock, but that may change.

<@walters:fedora.im>

15:37:21

I am going to devconf.cz, which is right after Flock...not yet sure if I'm doing both

<@jbrooks:matrix.org>

15:37:31

All right, let's adjourn this one, let's talk in the bootc room about topics for next week

<@dustymabe:matrix.org>

15:38:24

I got accepted to Flock so I'll be going to that one if I get travel approved

<@dustymabe:matrix.org>

15:38:33

Was definitely hoping to catch up with you Colin Walters

<@snthrailkill:matrix.org>

15:38:34

Same

<@dustymabe:matrix.org>

15:38:48

Thanks for running the meeting Jason Brooks

<@jbrooks:matrix.org>

15:39:10

!endmeeting