fix(deploy-kernel): the stage assert accepts an already-booted target #2062
Loading…
Reference in a new issue
No description provided.
Delete branch "fix/kstage-idempotent"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
status.staged is null on an already-converged node (bootc: Image specification is unchanged); the retried sf35 leg crashed on the None while the fleet was 5/5 on target. Accept staged OR booted matching the frozen digest; three-state simulation attached in the commit.
'bootc switch' answers 'Image specification is unchanged' and stages NOTHING when the node already runs the target — status.staged is null and the assert crashed on the None ('NoneType' object is not subscriptable). A retried roll-node leg after a completed stage+reboot lands exactly there: the seanfarm35 workers roll converged 5/5 and still recorded Failed on worker a40389. The assert now accepts the digest in EITHER status.staged or status.booted, and names both images when neither matches. Verified against all three simulated bootc states (staged / booted / neither).