Universal Blue•8mo ago

do you know why we have this on bluefin?

j0rge•4/28/25, 9:05 PM

anything in blame?

tulipOP•4/28/25, 9:07 PM

@M2 m2 shall be blamed for this

Amy•4/28/25, 9:07 PM

what does that mean-

tulipOP•4/28/25, 9:07 PM

https://github.com/ublue-os/bluefin/pull/1251

Amy•4/28/25, 9:09 PM

Not sure if that'd be an issue tho, secureblue having this issue kinda suggest it is an upstream thing (or maybe it is a seperate issue with the same symptom

)

tulipOP•4/28/25, 9:09 PM

OH im sorry this isnt related to what we are talking about

AAmy Not sure if that'd be an issue tho, secureblue having this issue kinda suggest i...

tulipOP•4/28/25, 9:09 PM

man even centos got that issue

tulipOP•4/28/25, 9:09 PM

its definitely something upstream

Amy•4/28/25, 9:09 PM

oh I'm srry

tulipOP•4/28/25, 9:09 PM

maybe podman?

tulipOP•4/28/25, 9:09 PM

no idea.

AAmy oh I'm srry

tulipOP•4/28/25, 9:10 PM

M2•4/28/25, 9:10 PM

Okay, I wonder if a version of distrobox volume mounted /etc/hostname

M2•4/28/25, 9:10 PM

name of the offending distrobox replace fedora toolbox

Robert•4/28/25, 9:11 PM

The problem is only when a Distrobox was created when ~~os-release~~ hostname was present, then it was removed. Because the bind mounts no longer work

M2•4/28/25, 9:12 PM

podman inspect fedora-toolbox | jq -r 'any(.[].Config.CreateCommand.[]; contains("/etc/hostname"))'

podman inspect fedora-toolbox | jq -r 'any(.[].Config.CreateCommand.[]; contains("/etc/hostname"))'

Robert•4/28/25, 9:12 PM

So the real fix is finding what removes ~~/etc/os-release~~ /etc/hostname

M2•4/28/25, 9:12 PM

/etc/os-release is a symlink

M2•4/28/25, 9:12 PM

the error says cannot stat /etc/hostname

tulipOP•4/28/25, 9:13 PM

i do wonder like

tulipOP•4/28/25, 9:13 PM

did we run into an ostree regression?

tulipOP•4/28/25, 9:14 PM

doesnt seem like it

tulipOP•4/28/25, 9:14 PM

oh my god https://github.com/bootc-dev/bootc/pull/1167

GitHub

store: Really remove empty /etc/resolv.conf and /etc/hostname by cg...

The previous change here was a no-op for two reasons:

It's actually usr/etc at this point
We were operating on the wrong rootfs

Fixes: 57bd0dc

j0rge•4/28/25, 9:14 PM

THERE SHE IS!!!!!!!!!!!!!!

tulipOP•4/28/25, 9:14 PM

maybe thats it?

tulipOP•4/28/25, 9:14 PM

the changes seem a bit unrelated

tulipOP•4/28/25, 9:15 PM

but man it would make a lot of sense for this thing to be a bootc regression

Robert•4/28/25, 9:15 PM

So the file was managed by bootc, then the file was removed from the images

j0rge•4/28/25, 9:15 PM

timeline matches

j0rge•4/28/25, 9:15 PM

and we just got 1.1.7 recently everywhere

tulipOP•4/28/25, 9:15 PM

YES.

j0rge•4/28/25, 9:15 PM

I remember reading the changelog with p5 in it.

j0rge•4/28/25, 9:16 PM

ok is our workaround still valid?

tulipOP•4/28/25, 9:16 PM

i think so

Robert•4/28/25, 9:16 PM

IMO it's not really a regression, but rather fixing something that shouldn't have been there in the first place.
But fixing that caused the issue

Robert•4/28/25, 9:16 PM

New installs probably don't have this problem

j0rge•4/28/25, 9:16 PM

alright then

j0rge•4/28/25, 9:16 PM

let's roll with this workaround

j0rge•4/28/25, 9:17 PM

LTS is a beta so no problemo there, once this merges I'll repush gts and stable and then we'll be done.

Amy•4/28/25, 9:17 PM

tulipOP•4/28/25, 9:17 PM

@j0rge

tulipOP•4/28/25, 9:17 PM

@bsherman we can now default to systemd-resolved on bluefin LTS

j0rge•4/28/25, 9:18 PM

wait what are we using now?

Jj0rge wait what are we using now?

tulipOP•4/28/25, 9:18 PM

networkmanager's default resolv.conf thing

j0rge•4/28/25, 9:18 PM

tulipOP•4/28/25, 9:18 PM

yes.

tulipOP•4/28/25, 9:18 PM

i hate it too

tulipOP•4/28/25, 9:18 PM

and it breaks tailscale

j0rge•4/28/25, 9:18 PM

I know I get the errors

do you know why we have this on bluefin?

Similar Threads