ostree-sysroot-deploy: check if deployments are in the same stateroot.

[jmarrero]

Our code currently does not check for stateroot as a differentiator between deployments, however given that we allow multiple stateroots, we should see them as different deployments in the same way we check for kargs for example.

[openshift-ci]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[ericcurtin]

I agree with this, but I am curious what is the use case where all the previous checks have not returned FALSE already. The case where this happens in real-life... We can't depend on changing kargs, this is correct, we use ostree=true in our Android Bootloader images for example.

[jmarrero]

thanks for the review!
Single Node Openshift uses stateroots for managing faster upgrades. And they hit this.

See OCPBUGS-30276

just trying to confirm locally this works as intended before moving this out of draft.

[cgwalters]

Ooh, good catch. Fun bug. I am actually looking forward to us trying to reuse the stateroot stuff more broadly as it's a very powerful thing.

I think it'd be relatively straightforward to do a test case for this in tests/admin-test.sh or so, I can help with that

[cgwalters]

FWIW I'm also OK adding and shipping this without an additional test, since this looks like a relatively critical bug for that use case and I am very confident the fix is correct.

[jmarrero]

I can see the failure reported in the original issue with the test when I remove my changes:

error: Reloading deployments after commit: loading sysroot: Parsing deployment /ostree/boot.0/testos2/7bd091ad6955c0e7ef41bab65adb016275664b55efc0d10a0c8cb263da3c1e4e/0 in stateroot 'testos2': readlinkat: No such file or directory

[cgwalters]

[2024-05-02T00:14:50.093Z] FAIL: tests/test-libarchive-import 11 /libarchive/selinux - OSTree:ERROR:tests/test-libarchive-import.c:681:test_libarchive_selinux: assertion failed (error == NULL): Child process exited with code 1 (g-spawn-exit-error-quark, 1)

Hmm that one is weird.

[cgwalters]

@jmarrero I think this just needs a rebase 🏄