AC stack is connected alongside the core switch with the gateway on the core switch; both ACs have two cards each; the stacking ports between AC and SW are f2/2/17 and f1/2/17;
Terminal centralized forwarding, fault location by point.
Upper-layer terminals can ping the gateway normally.
Lower-layer terminals cannot ping the gateway but can communicate with AC normally. Repeated tests show that when pinging the gateway, only the first packet succeeds while subsequent packets fail.
1. Changed to local forwarding, and the issue disappeared;
2. The fault phenomenon distinguishes the point, but ping statistics show packet loss at AC;
3. Changed the AC forwarding mode to soft forwarding; wlan fast-forwarding mode hardware changed to software; // The issue disappeared, so the problem lies in the 58x FPGA;
4. Investigated the device networking and found unreasonable networking for the faulty AP; As shown in the figure above, the faulty AP is online on AC G1/1/1;
The orange line indicates the traffic flow. The AP egress is on card 1, but the traffic needs to enter from card 2. Check the forwarding entry, and the traffic will enter from card 2. Verify the egress card number and port number. Since one stacking port is on card 2 and the AP is on card 1, the verification will fail, resulting in packet loss.
The first ping packet is sent to the processor for processing, so it is not subject to the above restrictions.
1. The most reasonable networking state requires that the service ports of AC and SW be allocated to two cards with equal quantities.
2. Temporary mitigation involves moving the AP at the faulty point to card 2.