Nonstop forwarding with OSPFv2 and OSPFv3

On a Nonstop OSPFv2 router, failover of a MM does not result in the OSPF v2 router being removed from the OSPFv2 domain. A restart request is sent by the Nonstop OSPFv2 router to the neighboring OSPFv2 routers, after which the graceful restart process begins. This behavior applies to OSPFv3 as well.

A graceful restart allows an OSPF routing switch to stay on the forwarding path while being restarted. The routing switch sends “grace LSAs” that notify its neighbors that it intends to perform a graceful restart. During the configurable grace period, the restarting switch’s neighbors continue to announce the routing switch in their LSAs as long as the network topology remains unchanged. The neighbors run in “helper mode” while the routing switch restarts.

Graceful restart will fail under these conditions:

  • There is a topology change during the graceful restart period. The helper switches exit helper mode and adjacencies are lost until the restarting switch rebuilds the adjacencies.

  • The neighbor switches do not support helper mode.

  • Configure router-id or IPv4 loopback address for OSPFv3 Non-Stop Forwarding to work on the switch.

  • For OSPF nonstop switching to work without traffic loss (after switchover), redundancy rapid-switchover time should be greater than(~10secs) restart interval time of OSPF.

For more information on OSPFv2 and OSPFv3 graceful restart, see RFC 3623 and RFC 5187.