JNCIS-ENT: High Availability

VRRP and first-hop redundancy

set protocols vrrp interface ge-0/0/0.0 virtual-address 10.0.0.254 gives hosts a shared gateway. Priority defaults to 100; the highest priority becomes master, and a tie is broken by the highest interface IP address. Junos enables preemption by default, so a returning higher-priority router takes the master role back - a very common exam trap. The advertise interval defaults to 1 second, and failover takes roughly three intervals plus skew. accept-data allows the master to accept traffic destined to the virtual address for some services, and multiple VRIDs on one interface let you load-share between routers.

Other HA mechanisms

• LAG/LACP: set interfaces ae0 aggregated-ether-options lacp active; at least one side must be active for the bundle to form, and minimum-links protects capacity after a member fails.
• GRES, NSR, and nonstop bridging keep forwarding and sessions alive across a Routing Engine switchover.
• BFD provides sub-second failure detection; detection time is approximately minimum-interval multiplied by the multiplier, negotiated with the neighbour, so it must be enabled on both ends.
• Redundant trunk groups give loop-free dual-homed Layer 2 connectivity without relying on STP.

Traps

• Assuming preemption is off because another vendor defaults it off.
• Priority 255 is reserved and 0 means the router gives up mastership.
• LACP passive on both sides never bundles.
• BFD with mismatched timers or enabled at only one end silently does nothing.

Worked example

R1 uses priority 150 and R2 uses 100 on the same VRID. R1 is master while healthy; when R1 fails, R2 takes over within roughly three advertise intervals, and when R1 returns it preempts and resumes mastership automatically.