Reference Guide

• OSPFv2 and OSPFv3 support planned-only and/or unplanned-only restarts. The default is support for both
planned and unplanned restarts.
A planned restart occurs when you enter the redundancy force-failover rpm command to force the
primary RPM to switch to the backup RPM. During a planned restart, OSPF sends out a Grace LSA before the
system switches over to the backup RPM.
An unplanned restart occurs when an unplanned event causes the active RPM to switch to the backup RPM,
such as when an active process crashes, the active RPM is removed, or a power failure happens. During an
unplanned restart, OSPF sends out a Grace LSA when the backup RPM comes online.
To display the configuration values for OSPF graceful restart, enter the show run ospf command for OSPFv2 and
the
show run ospf and show ipv6 ospf database database-summary commands for OSPFv3.
Fast Convergence (OSPFv2, IPv4 Only)
Fast convergence allows you to define the speeds at which LSAs are originated and accepted, and reduce OSPFv2 end-
to-end convergence time.
FTOS allows you to accept and originate LSAa as soon as they are available to speed up route information propagation.
NOTE: The faster the convergence, the more frequent the route calculations and updates. This impacts CPU
utilization and may impact adjacency stability in larger topologies.
Multi-Process OSPFv2 (IPv4 only)
Multi-process OSPF is supported on the Z9000 platform with FTOS version 7.8.1.0 and later, and is supported on OSPFv2
with IPv4 only.
Multi-process OSPF allows multiple OSPFv2 processes on a single router. Multiple OSPFv2 processes allow for isolating
routing domains, supporting multiple route policies and priorities in different domains, and creating smaller domains for
easier management.
The Z9000 supports up to 32 OSPFv2 processes.
Each OSPFv2 process has a unique process ID and must have an associated router ID. There must be an equal number
of interfaces and must be in Layer-3 mode for the number of processes created. For example, if you create five OSPFv2
processes on a system, there must be at least five interfaces assigned in Layer 3 mode.
Each OSPFv2 process is independent. If one process loses adjacency, the other processes continue to function.
Processing SNMP and Sending SNMP Traps
Though there are may be several OSPFv2 processes, only one process can process simple network management
protocol (SNMP) requests and send SNMP traps.
The mib-binding command identifies one of the OSPVFv2 processes as the process responsible for SNMP
management. If you do not specify the
mib-binding command, the first OSPFv2 process created manages the SNMP
processes and traps.
RFC-2328 Compliant OSPF Flooding
In OSPF, flooding is the most resource-consuming task. The flooding algorithm described in RFC 2328 requires that OSPF
flood LSAs on all interfaces, as governed by LSA’s flooding scope (refer to Section 13 of the RFC.)
When multiple direct links connect two routers, the RFC 2328 flooding algorithm generates significant redundant
information across all links.
By default, FTOS implements an enhanced flooding procedure which dynamically and intelligently detects when to
optimize flooding. Wherever possible, the OSPF task attempts to reduce flooding overhead by selectively flooding on a
subset of the interfaces between two routers.
481