The redirect-reverse command enables the IPv4 packets and response IPv4 packets to be transferred on the same interface.
The undo redirect-reverse command cancels the configuration.
redirect-reverse
redirect-reverse next-hop ip-address [ per-packet ]
undo redirect-reverse [ per-packet ]
| Parameter | Description | Value |
|---|---|---|
next-hop ip-address |
Specifies the IPv4 address of the next hop. When you run the redirect-reverse command in the view of a GE interface, GE subinterfaces, Eth-Trunk interface, Eth-Trunk subinterfaces, VLAN interfaces, or VXLAN interfaces, you must specify the next-hop ip-address parameter. When you run the command in the view of a dialer, tunnel interface, do not specify the next-hop ip-address parameter. |
- |
per-packet |
Indicates per-packet sticky load balancing. Only Ethernet interfaces and subinterfaces, Eth-Trunk interfaces and subinterfaces support this parameter. |
- |
Application Scenarios
The FW uses the incoming interface of the forward packets as the outgoing interface of return packets instead of looking up the routing table.
In the multi-ISP load balancing NAT server scenario, the FW looks up the routing table for an outgoing interface to send the return traffic from a server. As a result, the return traffic from the server may take a path on ISP2, although the request to the server takes a link on ISP1. The inconsistent forward and return paths may slow down or even interrupt services. To resolve this issue, configure the sticky load balancing function on the incoming interface of ISP1.
Precautions
If you run the redirect-reverse command in the view of a subinterface, you must add the subinterface to a VLAN. Otherwise, the next hop specified in the command does not take effect.
After per-packet sticky load balancing is enabled on an interface, hardware fast forwarding does not apply to the traffic entering this interface.
The priority of direct routes is higher than that of the sticky load balancing function. The device preferentially forwards response packets based on direct routes even if the sticky load balancing function is configured.
If equal-cost multipath (ECMP) routes are configured, the sticky load balancing function is enabled by default. In case of non-equal-cost routes, the sticky load balancing function is disabled by default, and you need to run this command to enable the function.
If the NAT Server global address is in the virtual system and the inside address is in the root system, configuring the sticky load balancing function on the inbound interface of the FW may prevent ISP users from properly accessing the internal server. This is because that the response packets of the internal server cannot find the correct cross-root system route and are forwarded along the root system route, causing a traffic forwarding failure.