The OSPFv3 protocol can be configured to participate in the checkpointing service, so that these protocols can execute a graceful restart when the management unit fails. In a graceful restart, the hardware to continues forwarding IPv6 packets using OSPFv3 routes while a backup switch takes over management unit responsibility.
Graceful restart uses the concept of helpful neighbors. A fully adjacent router enters helper mode when it receives a link state announcement (LSA) from the restarting management unit indicating its intention of performing a graceful restart. In helper mode, a switch continues to advertise to the rest of the network that they have full adjacencies with the restarting router, thereby avoiding announcement of a topology change and the potential for flooding of LSAs and shortest-path-first (SPF) runs (which determine OSPF routes). Helpful neighbors continue to forward packets through the restarting router. The restarting router relearns the network topology from its helpful neighbors.
Graceful restart can be enabled for either planned or unplanned restarts, or both. A planned restart is initiated by the operator through the management command initiate failover. The operator may initiate a failover in order to take the management unit out of service (for example, to address a partial hardware failure), to correct faulty system behavior which cannot be corrected through less severe management actions, or other reasons. An unplanned restart is an unexpected failover caused by a fatal hardware failure of the management unit or a software hang or crash on the management unit.
