Network & Infrastructure / IPv6, TCP & Transport Engineering
IPv6 neighbor states: STALE is not the same as FAILED
Monitoring reports every STALE IPv6 neighbor as an outage even while traffic works. An incident response then clears caches unnecessarily and increases neighbor-discovery load.
Environment and inputs: Dual-stack routed networks and transport endpoints; validate the complete path and the active operating-system version. Diagnostic/design example; not tested on production equipment.
Collect the evidence
Record neighbor states, reachability probes, recent traffic and transitions through delay and probe states. Compare actual application failures with neighbor entries that are merely awaiting reconfirmation.
Explain the result
Neighbor unreachability detection uses several states to balance cached forwarding and active validation. STALE can be usable cached information; FAILED indicates a different outcome after attempted resolution or probing.
Plan the fix
Update monitoring around failed resolution, persistent probe failure and service impact. Investigate Layer-2 delivery or endpoint ownership only where evidence supports a real reachability defect.
Check the result and rollback
Verify alert behavior during idle periods, resumed traffic and a controlled endpoint removal. Restore the old rule only if the new logic misses a demonstrated failure, then refine its state and duration criteria.