If pods die ungracefully, ghost nodes linger in your Tailscale admin console forever.
Ideas
- Reconciliation loop that compares running pods to Tailscale nodes
- Clean up nodes that no longer have matching pods
- Maybe just a manual `kubectl ts cleanup` command
If pods die ungracefully, ghost nodes linger in your Tailscale admin console forever.
Ideas