Installation guide

Local Ethernet
GRP Redundant Processor Support 5
Note When the startup configuration is updated with copy [tftp | rcp | ... ] startup-config, and
auto-sync is enabled, the boot variables on the secondary GRP will be updated according to the values specified
in the current running-config. These values may or may not correspond to what is set in the new startup-config,
just as the values of these variable on the primary may or may not correspond to the startup-config. To avoid
this problem, use copy running-config starup-config to update the startup configuration.
Local Ethernet
The GSR router provides one Ethernet port for each GRP. However, software restrictions limit the
router to using only the Ethernet port on the primary GRP. The MAC address for the Ethernet port
is held in the chassis NVRAM, while the IP address is be taken from the startup-config. The same
MAC address and, typically, the same IP address will be used regardless of which GRP is primary.
The router loses no functionality by not using the Ethernet port on the secondary. For security
reasons, access to the secondary is mediated by the primary. Restricting Ethernet access to the
primary Ethernet port fits into this access model.
There are some consequences due to these restrictions, including the following:
• Network booting from local Ethernet is not currently allowed. Booting is restricted to Flash.
• If a primary GRP fails, it does not attempt to use its Ethernet to send a core dump. Instead, it informs the
secondary GRP that a dump is available. When the secondary is ready to process the dump, it will request
the crash data from the failed primary, then send it to the network destination defined in the
exception
dump global configuration command.
Failover
The term failover describes the event when one GRP card takes over operation from another. A
failover can be forced, by entering the redundancy force-failover EXEC command, or a failover
occurs when the primary GRP fails and the secondary GRP takes over.
The secondary GRP monitors so-called heartbeat messages from the primary GRP. If no heartbeat
messages are detected for a few seconds, the secondary takes over router operations from the failed
primary. The primary also monitors heartbeat messages from the secondary and attempts to reload
the secondary if the heartbeats cease.
When failover occurs, it interrupts system operation momentarily. The following events occur during
failover:
• The failed primary GRP will be reset and reloaded, assuming the GRP is properly configured.
• Linecards are reloaded to bring them to a known state.
• The router stops passing traffic.
• Route information is lost.
• All connections are lost.
• The secondary GRP card takes over as the primary GRP. Because this formerly-secondary GRP
card has separate image and configuration files, it can act as the sole processor. It will establish
connections with any peer routers to discover route information. The routing tables are thereby
regenerated and downloaded to the linecards.
When a primary GRP detects a fatal hardware or software error, it sends a message to the secondary
GRP. In this instance, the secondary takes over immediately, without waiting the few seconds for the
heartbeat detection logic to discover the failure.