Almost sounds like we could've prevented this in 40 at least with this?: ```bash RUN rpm-ostree ins

Almost sounds like we could've prevented this in 40 at least with this?:

RUN rpm-ostree install bootupd \
    &&\
    /usr/bin/bootupctl backend generate-update-metadata \
    && \
    ostree container commit

RUN rpm-ostree install bootupd \
    &&\
    /usr/bin/bootupctl backend generate-update-metadata \
    && \
    ostree container commit

RUN rpm-ostree install bootupd \
    &&\
    /usr/bin/bootupctl backend generate-update-metadata \
    && \
    ostree container commit

RUN rpm-ostree install bootupd \
    &&\
    /usr/bin/bootupctl backend generate-update-metadata \
    && \
    ostree container commit

antheas•6/19/24, 1:34 PM

the bootloader resides in /efi

antheas•6/19/24, 1:34 PM

the update has to be done live and can brick the system

EyeCantCUOP•6/19/24, 1:36 PM

Isn't that the exact use case of bootupd?

EyeCantCUOP•6/19/24, 1:38 PM

At least, I saw that was suggested by Timothee for the future to prevent this from occurring. I had assumed that's what it was doing

EyeCantCUOP•6/19/24, 1:40 PM

Hmm... Yeah, the solution is to outright do what bootupd does: https://github.com/fedora-silverblue/issue-tracker/issues/543#issuecomment-2048350047

antheas•6/19/24, 1:43 PM

Yeah

antheas•6/19/24, 1:43 PM

That's what bootupd does

Aantheas the update has to be done live and can brick the system

antheas•6/19/24, 1:44 PM

That's why bootupd doesn't right now

EEyeCantCU Almost sounds like we could've prevented this in 40 at least with this?: ```bas...

antheas•6/19/24, 1:45 PM

Not sure what that does but if it makes bootupd run automatically then yeah that would work

EyeCantCUOP•6/19/24, 2:03 PM

I wish I had a straightforward way of testing it but I've already updated everything and have our key enrolled everywhere

antheas•6/19/24, 2:04 PM

copy your efi

antheas•6/19/24, 2:04 PM

then run it and see if it updates stuff

Wiertelus•6/19/24, 2:57 PM

how can i add kernel arguments like mem_sleep_default=deepmem_sleep_default=deep to my image

Wiertelus•6/19/24, 2:57 PM

ive tried with pm-ostree kargs but since the image isnt booted i cant do it that way

Kyle Gospo•6/19/24, 2:58 PM

You have to make a service that applies them on the first boot

Wiertelus•6/19/24, 2:59 PM

ah alrighty thank you!

M2•6/19/24, 5:36 PM

TIL you can do heredocs in dockerfiles

M2•6/19/24, 5:44 PM

https://github.com/ublue-os/toolboxes/pull/107

GitHub

fix: make the docker-dbox work by m2Giles · Pull Request #107 · ubl...

M2•6/19/24, 5:46 PM

Need to use a more up to date podman for the here doc to work

j0rge•6/19/24, 5:46 PM

switch the builder to 24.04?

M2•6/19/24, 5:47 PM

Yeah or use the updated podman from obs

M2•6/19/24, 6:26 PM

Both appear to not be new enough

M2•6/19/24, 6:26 PM

Switching to docker

M2•6/19/24, 6:26 PM

Absolute fail Ubuntu

antheas•6/19/24, 6:26 PM

i use nested podman to build the new bazzite image now

antheas•6/19/24, 6:27 PM

nested buildah

antheas•6/19/24, 6:27 PM

needs access to dev/fuse though

M2•6/19/24, 7:18 PM

Switched to docker and it builds in seconds.

M2•6/19/24, 7:21 PM

Podman gained heredoc about 6 months ago. And gets rid of the whole && everywhere.... And it's not in the current Ubuntu LTS....

M2•6/19/24, 9:23 PM

For those interested in using docker and/or incus from inside a distrobox. We are building docker and incus distroboxes to be used with a rootful distrobox.

You can see them at ublue-os/toolboxes. There is a PR for fixing up the docker distrobox.

The idea is that you will volume mount in their dir (/var/lib/docker and /var/lib/incus)

For docker it will also use socat to redirect a socket back to the host.

For running the clients you can install them from brew. Example distrobox.ini are in the repo next to the containerfiles.

For those on Bazzite, this is a way to run docker/incus without having to layer

MM2 For those interested in using docker and/or incus from inside a distrobox. We ar...

Luke Skywunker•6/19/24, 9:39 PM

Oh this will be useful for fixing our BlueBuild distrobox

M2•6/19/24, 9:45 PM

It's an ubuntu box, so pretty easy to FROM. With the new changes, it auto exports a socket to the host. So if you want to do multiple ones you'll need to delete that service file (/usr/lib/systemd/system/docker-setup.service)

fiftydinar•6/20/24, 7:00 AM

Seems like xpad in missing in Fedora images since 6.9 kernel got in
https://www.reddit.com/r/Fedora/comments/1dk20p7/kernel_694_broke_8bitdo_ultimate_controller/

From the Fedora community on Reddit

Explore this post and more from the Fedora community

M2•6/20/24, 1:44 PM

@Noel

Bootc thoughts:

Since that command is necessary to have an F40 image install. And it has to be done after any initramfs change as well. Should we only have it on downstream issues until Anaconda is fixed up to not need that command?

MM2 @Noel Bootc thoughts: Since that command is necessary to have an F40 image in...

Noel•6/20/24, 1:46 PM

It was my understanding that Anaconda wants to ultimately move to using bootc to deploy images at some point.

Noel•6/20/24, 1:46 PM

when that is happening is up in the air currently.

M2•6/20/24, 1:47 PM

So until then, have bootc only on downstream images to avoid the blow up. Since we control those

Noel•6/20/24, 1:47 PM

Instead of the main images?

Noel•6/20/24, 1:47 PM

yeah, that might be safest.

M2•6/20/24, 1:47 PM

Yeah

M2•6/20/24, 1:49 PM

Like we probably should have documentation for it... But I can see it as a massive footgun for anyone making their own image. So they should opt into bootc

MM2 @Noel Bootc thoughts: Since that command is necessary to have an F40 image in...