debian custom image machine-id is the same on all provisioned servers

[jsfrerot]

Hi, I noticed that /etc/machine-id is the same on all maas provisioned servers.
What is the proper way to fix this?

I used the following command to generate my image:
make debian SERIES=trixie CUSTOM_SCRIPT=scripts/my-changes.sh
in this my-changes.sh script, I'm simply adding repos and packages.

[jsfrerot]

After digging a little, I found there is a script that is supposed to do the cleanup of machine-id in scripts/fuse-tar-root
Looking at the build logs I see

==> cloudimg.image.qemu.cloudimg (shell-local): ++ '[' -f /tmp/packer-maas-ils7/root/etc/machine-id ']'
==> cloudimg.image.qemu.cloudimg (shell-local): ++ echo 'reseting machine-id...'
==> cloudimg.image.qemu.cloudimg (shell-local): ++ :
==> cloudimg.image.qemu.cloudimg (shell-local): ++ '[' -f /tmp/packer-maas-ils7/root/var/lib/cloud/instance ']'
==> cloudimg.image.qemu.cloudimg (shell-local): reseting machine-id...
==> cloudimg.image.qemu.cloudimg (shell-local): ++ '[' -d curtin ']'
==> cloudimg.image.qemu.cloudimg (shell-local): ++ '[' -d '' ']'

and after looking at the content of the generated tar.gz, the machine-id is still not empty.
For now, I added this line to my custom script my-changes.sh:
truncate -s 0 /etc/machine-id
And it fixes my issue

I think this line may not work properly: https://github.com/canonical/packer-maas/blob/main/scripts/fuse-tar-root#L32

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[jsfrerot]

Not stale

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.