Nonstop Routing (NSR) is provided for Open Shortest Path First (OSPF) and Label Distribution Protocol (LDP) protocols for the following events:

• Route Processor (RP) or Distributed Route Processor (DRP) failover

• Process restart for either OSPF, LDP, or TCP

• Minimum Disruption Restart (MDR)

In the case of the RP failover, NSR is achieved by for both TCP and the applications (OSPF or LDP).

NSR is a method to achieve High Availability (HA) of the routing protocols. TCP connections and the routing protocol sessions are migrated from the active RP to standby RP after the RP failover without letting the peers know about the failover. Currently, the sessions terminate and the protocols running on the standby RP reestablish the sessions after the standby RP goes active. Graceful Restart (GR) extensions are used in place of NSR to prevent traffic loss during an RP failover but GR has several drawbacks.

You can use the nsr process-failures switchover command to let the RP failover be used as a recovery action when the active TCP or active LDP restarts. When standby TCP or LDP restarts, only the NSR capability is lost till the standby instances come up and the sessions are resynchronized but the sessions do not go down. In the case of the process failure of an active OSPF, a fault-management policy is used. For more information, refer to Implementing OSPF on Cisco IOS XR Routing Configuration Guide for the Cisco XR 12000 Series Router.

Stream Control Transmission Protocol (SCTP) is a reliable transport protocol that provides multihoming, stream support, and partial reliability. Multihoming occurs when one (or both) endpoints of a connection can consist of more than one IP address, which enables transparent failover between redundant network paths. SCTP can transport multiple message-streams.

TCP is a connection-oriented protocol that specifies the format of data and acknowledgments that two computer systems exchange to transfer data. TCP also specifies the procedures the computers use to ensure that the data arrives correctly. TCP allows multiple applications on a system to communicate concurrently, because it handles all demultiplexing of the incoming traffic among the application programs.

The User Datagram Protocol (UDP) is a connectionless transport-layer protocol that belongs to the IP family. UDP is the transport protocol for several well-known application-layer protocols, including Network File System (NFS), Simple Network Management Protocol (SNMP), Domain Name System (DNS), and TFTP.

Any IP protocol other than TCP, UDP, or SCTP is known as a RAW protocol.

For most sites, the default settings for the TCP, UDP, and RAW transports need not be changed.

This task allows you to configure failover as a recovery action to process failures of active instances.

When the active TCP or the NSR client of the active TCP terminates or restarts, the TCP sessions go down. To continue to provide NSR, failover is configured as a recovery action. If failover is configured, a switchover is initiated if the active TCP or an active application (for example, LDP, OSPF, and so forth) restarts or terminates.

For information on how to configure MPLS Label Distribution Protocol (LDP) for NSR, refer to the Cisco IOS XR MPLS Configuration Guide for the Cisco XR 12000 Series Router.

For information on how to configure NSR on a per-process level for each Open Shortest Path First (OSPF) process, refer to the Cisco IOS XR Routing Configuration Guide for the Cisco XR 12000 Series Router.

Note	Before performing this procedure, enable RP isolation using the isolation enable command for improved troubleshooting. Without enabling RP isolation, the failing process will not generate the logs required to find the root cause of the failure.

1.    configure


2.    nsr process-failures switchover


3.    commit

RP/0/0/CPU0:router(config)# nsr process-failures switchover