Fedora IoT greenboot login errors

Running Fedora IoT 36 on a BIOS Intel x86_64 system.

After logging in I get these lines pasted above my issues, /etc/issue.d is empty so I don’t know where this comes from.

Script '01_update_platforms_check.sh' FAILURE (exit code '1'). Continuing...

Boot Status is GREEN - Health Check SUCCESS

Googled for a bit and found on fedora.pkgs.org that it’s supposed to be provided by greenboot-update-platforms-check, however both this and its parent package greenboot, seem to only have created the directory trees with I believe every file, except the systemd servers/timers, being missing. Is there a fix (at least how to get rid of the login message) or is this intended for some reason?

1 Like

My dude,

While not entirely sure why the URL for the default remote is failing, using the mirror URL does work:

% cat /etc/ostree/remotes.d/fedora-iot.conf
[remote "fedora-iot"]

So, in order to “fix” this issue, we need to create a new OSTree remote pointing to the mirror URL.
However, bear in mind that by doing so we are slightly deviating from the way IoT is supposed to be configured; for instance, we will need to rebase our current deployment with the new remote for rpm-ostree to be able to interact with the remote repository.

Personally, I will keep a close eye on this issue and revert back to the original remote as soon as this issue is fixed.

Enough chatter, let’s tighten those loose nuts! :rocket:



% cat /etc/os-release
PRETTY_NAME="Fedora Linux 37.20221023.0 (IoT Edition)"

% uname -i

Workaround steps:

  1. Let’s create a new OSTree remote:
sudo ostree remote add --if-not-exists \
--gpg-import=/etc/pki/rpm-gpg/RPM-GPG-KEY-fedora-37-x86_64 \
fedora-iot-mirrorlist \
  1. Now we need to remove the broken remote, but before doing so let’s back it up:
backupDir="$HOME/backups/etc/ostree/remotes.d/" \
&& mkdir -p $backupDir \
&& cp /etc/ostree/remotes.d/fedora-iot.conf $backupDir
  1. With peace of mind we can now revert any changes made to the default remote, let’s remove it:
sudo ostree remote delete fedora-iot
  1. If everything went well, the script should now work properly:
deployment="$(ls -Art /sysroot/ostree/deploy/fedora-iot/deploy | tail -n 1)" \
/sysroot/ostree/deploy/fedora-iot/deploy/$(echo "${deployment%%*(.origin)}")/usr/lib/greenboot/check/wanted.d/01_update_platforms_check.sh

# Quick explanation of the oneliner
# The ephemeral 'deployment' variable stores the filename of the newest file
# in the directory.
# We then trim the extension `.origin` because what we are really looking for
# is the newest directory name, which contains the booted OSTree deployment:
# $(echo "${deployment%%*(.origin)}")
# Credits for this trick to https://linuxhint.com/trim_string_bash/
  1. Because we removed the original OSTree remote the system is based on, rpm-ostree will fail when trying to upgrade the system.
    The solution to this is to rebase our deployment on the new remote:
 sudo rpm-ostree upgrade -r
 error: While pulling fedora/devel/x86_64/iot: Remote "fedora-iot" not found

sudo rpm-ostree rebase fedora-iot-mirrorlist:fedora/devel/x86_64/iot
sudo rpm-ostree upgrade -r

Note: when rebasing your deployment, pay attention to the remote branch you want to rebase it on, in my case it’s devel.

You can get a list of the available branches with the refs OSTree remote sub-command:

% sudo ostree remote refs fedora-iot-mirrorlist 

You shouldn’t see anymore the error message on subsequent logins.


I opened a ticket on Pagure (not really sure if there’s the proper place though, we will see…):