mDNS Cluster Discovery and Auto-Recovery in rpi_kubernetes
How mDNS, health monitoring, and k3s agent recovery scripts reduced cluster fragility in the Raspberry Pi environment.
Why This Change Was High Impact
Static-IP workflows are brittle in home and lab networks.
rpi_kubernetes moved to mDNS-centered discovery with Avahi, plus health and recovery scripts:
bootstrap/scripts/discover_cluster.pybootstrap/scripts/cluster_health_monitor.pybootstrap/scripts/k3s-agent-recovery.sh
That combination improved both day-one setup and day-two recovery.
The Operational Pattern
The pattern is straightforward:
- discover nodes by hostname where possible
- validate cluster health continuously
- auto-recover workers when agent connectivity fails
This is exactly the right level of automation for small clusters: simple enough to trust, strong enough to prevent repeated manual intervention.
Known Edge Cases
DEPLOYMENT_STATUS.md documents a real issue: control-plane hostname resolution mismatch. That transparency matters. A good runbook includes partial failures, not only success paths.
The project also keeps fallback discovery paths for environments where multicast DNS is unreliable.
Why This Is Better Than "Just Use Static IPs"
Static mapping can work, but it increases maintenance overhead after reimaging, DHCP churn, or network changes. Discovery + health checks shifts that burden from humans to repeatable scripts.
Practical Takeaway
For Raspberry Pi clusters, networking automation is reliability work, not convenience work. mDNS discovery paired with health and recovery scripts is one of the highest ROI improvements you can make.