Configuring the graceful restart helper mode
Many routers have separate control plane routing and forwarding plane processes. Therefore, it is possible for a particular routing process to fail independently of the forwarding plane. In such a scenario, it is often desirable for the forwarding plane to be unaffected by the routing process failure and for this process to restart and reconstitute its prefailure control plane state, typically with assistance from its neighbors or peers.
Many protocols, including RSVP-TE (as described in RFC 3209, RFC 3473, and RFC 5063), describe that a graceful-restart procedure works as follows.
- MPLS neighbors detect, usually by means of a liveliness detection mechanism (for example, Hello messages), the failure and restarting of the control plane process for the RSVP-TE protocol on a neighboring router.
- Neighbors of the failing router attempt to minimize data-plane disruption (usually by pretending that the neighbor is still up).
- Neighbors of the failing router attempt to help it recover its control plane state after it restarts, usually by retransmitting the appropriate protocol messages.
A router whose routing process fails goes into the restart mode and the neighbors of the router switch to the helper mode. However, because routing processes cannot restart on the vRouter, the vRouter supports only the RSVP-TE graceful-restart helper mode. The restarting mode is not supported.
When in the RSVP-TE graceful-restart helper mode, a vRouter can help, with respect to the LSP direction, an upstream or a downstream neighbor that is restarting as long as the neighbor indicates that it supports RSVP-TE graceful restart in accordance with the relevant RFCs.
To configure graceful restart on your vRouter, use the following command:
set protocols mpls-rsvp globals graceful-restart
vyatta@vyatta# set protocols mpls-rsvp globals graceful-restart
vyatta@vyatta# commit