"ip redirect" of Cisco Nexus

Network packet loss scenario caused by ICMP Redirect

When the network device sends its own packet, the next hop address and the sending source address are in the same subnet, it will reply an icmp redirect message to indicate to the sending source device that there is a more optimized route. When the source device receives the icmp redirect message, the subsequent packets are sent directly to the new next hop.

Cause

This is usually caused by policy routing (PBR) or unoptimized routes (incorrect static).

+-+ | +-+ +-+ | | 10.1.1.2 go 24 | +-- + | LAN +-+ | | +-+ | | | 10.1.1.1Compact 24 +-- + | +-+-- + +-+ 10.1.1.3Univer 24 |

As shown in the figure, suppose that a packet of 10.1.1.2 destined for 10.1.1.3 is forcibly sent by a policy route to 10.1.1.1, 10.1.1.1 will find that the next hop is in the same segment as itself, and then an ICMP Redirect will be replied to 10.1.1.2.

+-+ | +-+-- +-+ | | 10.1.1.2 + 24 | +-- + | | LAN +-+ | | +-+ | | 10.1.1.1-+ | 192.168.1.0-- + | 192.168.1.0