v0.0.12 — Kubernetes membership health and guided recovery

Highlights

• A node that’s reachable but hasn’t properly joined the cluster is now diagnosed, not just shown as “not ready.” A health card on the node’s own screen and on the Cluster screen names the actual problem — grown from a real case on the team’s own test cluster where a worker “wouldn’t join.”
• Guided, one-click repair. A Repair Kubernetes action rebuilds a node’s membership for you — confirm, watch the steps run live, done — always something you start yourself, never automatic.
• The same fixes that make repair reliable were folded into the ordinary add-a-worker flow too, so a new node benefits from the same hardening.

Known limitations

• Recovery is always user-initiated. There’s no automatic, unattended repair — you always confirm before anything runs.
• Multiple, redundant control-plane machines and a general-purpose Kubernetes diagnostics console remain out of scope.