Extreme EXOS VRRP: VRID Configuration and Failover

VRRP on ExtremeXOS is configured per VLAN rather than per interface, which trips up engineers coming from Cisco or Juniper where the virtual router hangs off a physical interface. Once the model is clear it is a four-command operation per VLAN: create the virtual router with a VRID, add the virtual IP, optionally set priority and preemption, then enable it. This article covers the exact EXOS syntax, the VRID scoping rules that determine whether two instances collide, and the verification output that proves master election worked.

The EXOS VRRP Model

Each VRRP instance is identified by three things together: VLAN name, VRID, and virtual IP address. Two routers configured with the same VLAN name, the same VRID and the same virtual IP are members of the same instance — one becomes master, the rest become backups. VRIDs are scoped to the VLAN, so VRID 1 can be reused in every VLAN without conflict, and a VRID may be any value between 1 and 255.

Base Configuration (Both Routers)

create vlan v10 tag 10
configure vlan v10 add ports 1-48 untagged
configure vlan v10 ipaddress 10.1.1.2/24        ! 10.1.1.3/24 on the peer
enable ipforwarding vlan v10

create vrrp vlan v10 vrid 1
configure vrrp vlan v10 vrid 1 add 10.1.1.1      ! same VIP on both routers
enable vrrp vlan v10 vrid 1

The virtual IP is what clients use as their default gateway. Note the order: the VR must exist before you can add an address, and it must have at least one virtual IP before it can be enabled. Extreme documents both of those as explicit error conditions, and the failure messages are unambiguous:

# enable vrrp
Failed to enable VR for vlan Data, vrid 10. VR must have at least 1 virtual IP address!

Priority and Preemption

Master election is by priority, highest wins. Ties are broken by the highest IP address. Preemption lets a higher-priority router reclaim the master role after it returns.

configure vrrp vlan v10 vrid 1 priority 110
configure vrrp vlan v10 vrid 1 preempt delay 30

A 30-second preemption delay is a reasonable default: it stops a router that has just rebooted from grabbing the master role before its routing protocols have reconverged, which would otherwise black-hole traffic for the duration of the convergence.

Failover Timing

EXOS uses a 1-second advertisement interval by default, and a backup waits for three consecutive missed advertisements before declaring the master down — roughly three seconds of failover time. That can be tuned when the application cannot tolerate it:

configure vrrp vlan v10 vrid 1 advertisement-interval 1 centiseconds

Shrinking the interval increases VRRP control traffic and makes the instance sensitive to transient CPU load; shrink it only after testing under real load.

Fabric Routing (Optional)

One EXOS-specific extra is VRRP fabric routing. It lets a VRRP backup router forward traffic that would otherwise be forwarded by the master, but only for destinations in a directly attached subnet. The practical effect is that upstream traffic is not forced through the master during normal operation.

configure vrrp vlan v10 vrid 1 fabric-routing on

Verification

show vrrp
show vrrp detail
show vrrp vlan v10
show vrrp vlan v10 stats
show configuration vrrp

show vrrp gives you the state per instance (MASTER / BACKUP / INIT), the priority, the virtual IP and the advertisement interval. If both routers claim MASTER, the two most likely causes are a mismatch in the virtual IP list or an L2 path problem preventing the advertisements from reaching the peer — VRRP advertisements use multicast 224.0.0.18, so a VLAN that filters multicast will silently break the instance.

Scaling and Caveats

• Multiple VRIDs per VLAN let you run several virtual routers with different VIP sets on the same VLAN, which is useful for load sharing between two routers across different subnets.
• VRRP does not load-balance by default; only the master forwards.
• Remember that VRRP protects the gateway, not the uplink. Pair it with a link-monitoring mechanism if you need failover when the uplink dies but the VRRP instance itself stays healthy.

Related articles on this site: HSRP vs VRRP vs GLBP: FHRP Configuration Compared for the cross-vendor first-hop redundancy comparison, Junos VRRP Configuration: VIP, Priority and Accept-Data configuration notes, and ExtremeXOS Stacking Configuration Step by Step general platform coverage.

原文链接：https://extreme-networks.my.site.com/ExtrArticleDetail?an=000083202