Hi guys,
I want to share with you a little problem that I have recently encountered.
Networking Topology
Fault symptom
Set the IS-IS neighbor timeout period to 3×4=12 seconds on the interface.
#
interface GigabitEthernet0/0/0
isis enable 1
isis timer hello 3 level-1
isis timer holding-multiplier 4 level-1
However, the IS-IS neighbor timeout period of the two devices is 300 seconds.
[AR2]dis isis peer
Peer information for ISIS(1)
System Id Interface Circuit Id State HoldTime Type PRI
-------------------------------------------------------------------------------
0000.0000.0001 GE0/0/0 0000.0000.0002.01 Up 300s L1 64
Cause
Analyze the device configuration. It is found that ISIS GR is configured for both AR1 and AR2, and the GR interval is 300 seconds.
As a result, the timeout period of the IS-IS neighbor relationship is 300 seconds. Although the timeout period configured for IS-IS on the interface is 12 seconds, it does not take effect.
[AR2]dis cu configuration isis
[V200R003C00]
#
isis 1
graceful-restart
is-level level-1
network-entity 10.0000.0000.0002.00
#
Solution
Using the graceful-restart no-impact-holdtime command to configure the holdtime of an IS-IS neighbor to remain unchanged after IS-IS GR is enabled
Summary
To configure GR, you need to enable GR for IGP, BGP, and LDP. After GR is configured, the timeout period of the protocol is prolonged. The specific values are as follows:
The default interval of OSPF GR is 120 seconds.
The default interval of ISIS GR is 300 seconds.
BGP GR: The default maximum waiting time for the local end to wait for GR recovery of the peer is 150 seconds.
The default period during which the BGP restarter waits for the End-Of-RIB flag is 600 seconds.
(The operation of enabling or disabling the BGP Graceful Restart feature may delete and reestablish all sessions and instances.)
MPLS LDP GR: The default value of the neighbor-liveness timer is 600 seconds.
The default value of the Reconnect timer of an LDP session is 300s.
The default value of the LSP Recovery timer is 300s.
As shown in the preceding information, the GR values are large. If fast protocol convergence is required, you can deploy BFD and other fault detection mechanisms.
