High Availability, IRF, and Resiliency
IRF (Intelligent Resilient Framework) virtualizes multiple physical switches into one logical device. Candidates must know member IDs, priority (higher wins election), and the IRF port binding two physical ports. The master handles control plane; others are standby/slave. Traps: duplicate member IDs prevent formation; mismatched software versions block merge; and MAD (multi-active detection) is required to prevent a split brain. Know BFD MAD vs. LACP MAD vs. ARP MAD and when each applies.
Beyond IRF, resiliency includes VRRP (virtual router, priority, preempt), Smart Link, RRPP, and link aggregation as HA primitives. VRRP traps: higher priority wins, preempt is often disabled by default, and tracking objects can lower priority. Also understand graceful restart and ISSU for hitless upgrades.
Worked example: two switches form IRF, then a link failure splits them. Without MAD, both become master and duplicate IPs cause chaos. Configure BFD MAD on a dedicated link; when the split is detected, one member enters recovery and shuts its ports. Verify with display irf and display mad verbose.
Rule of thumb: IRF priority is election-only; a higher priority does not guarantee master after a merge unless preempt conditions are met. Always pair IRF with MAD and verify member IDs and software match before deployment.