Configuring VRRP with concurrent meshing and routing

  • The VRRP configuration parameter preempt-delay-time should be set to at least 60 for each virtual router. This is because meshing is a distributed protocol that takes some time to stabilize when a new switch enters the mesh. A failing switch will be treated as new when it reenters the mesh. Since meshing is an L2 protocol, it must be stable before L3 VRRP can become active. The preempt-delay-time is only observed in the case where an already active VRRP router exists. This will allow the existing router to continue serving its role until the preferred router, owner or higher priority backup, is truly ready to take over.

  • The VRRP configuration parameter nonstop will not be configurable when VRRP is configured within a mesh. This is because the VRRP nonstop configuration attempts to make VRRP hitless in the case of a failed management module. Meshing, however, is not hitless in this case. Having the nonstop parameter set will cause the VRRP virtual router to ignore the preempt-delay-time and will have the virtual router attempt to become active before meshing is ready. This will result in a potential 30 or more second routed traffic gap while meshing becomes stable.

  • For best network resiliency during a VRRP failover event, all switches in the mesh domain must be running x.15.09 or later version of the switch software. This is because changes in versions x.15.09 and later allow a VRRP virtual router MAC address to move from the master to the backup without being blocked by meshing on connected switches in the mesh.

For more information, see the ArubaOS-Switch Multicast and Routing Guide for your switch.