rpm-ostree status after ostree container image deploy

[kwozyman]

Hi, folks!

I'm playing with the new ostreecontainer command in Anaconda/Kickstart in Fedora 39 (which I've succesfully backported to RHEL 9 branch) and I noticed that deploying a fedora container image (for example quay.io/fedora/fedora-silverblue:rawhide) works like a charm.

However, if I deploy a container image of Red Hat Device Edge (built using osbuild-composer and then ostree container encapsulate , for example quay.io/kwozyman/r4e:92) it gets written to disk, it boots correctly, but rpm-ostree status gives me "error fetching image metadata" and a subsequent rpm-ostree upgrade will repull the container image manifest and creates a new "fixed" deployment. Vm screenshot after boot

[cgwalters]

Can you try using e.g. rpm-ostree compose container-encapsulate instead? I wonder if there's a regression in deploying "flattened" images...

Ah yes this trivially reproduces outside of Anaconda too with a plain ostree container image deploy --image quay.io/kwozyman/r4e:92 --no-signature-verification. Probably an ostree-rs-ext bug.

[cgwalters]

Fix in ostreedev/ostree-rs-ext#556

[kwozyman]

That's great, thanks a lot @cgwalters! I'm afraid I don't quite understand the bug, is this at the container image creation step or at the deployment? Also, will this be backported in RHEL?

[cgwalters]

Deployment side.

Also, will this be backported in RHEL?

Yes

[kwozyman]

Thanks a lot!