Part Number: AM5716
Hi,
Customer met one ETH issue. Here is the describtion from customer.
The eth1 network port of AM5716 and the peer device are both connected to the switch. Under normal circumstances, a TCP link will be established and messages will be continuously exchanged through the MODBUS protocol. When running stably for a long time, the application will suddenly report a communication failure. The detection mechanism is that when the application on AM5716 cannot receive data, it will continuously try to re-create the socket for connection. When the blocking time exceeds 10s, a communication failure is determined. At this time, I logged in to the device through SSH and found that pinging the peer IP also failed, and the fault would not be recovered. At this time, communication can be restored by running ifconfig eth1 down/up or restarting the AM5716 device.
Due to the low frequency of the problem, it only occurs every few days or even once a week, and no effective information has been collected to help locate the problem before. Based on the existing information, please ask TI engineers to help find out if they have encountered similar problems?
In addition, I have encountered a similar problem on AM335x before. After running for a period of time, the tester used a computer to continuously ping the eth1 port of AM335x through the switch for testing, and the ping was suddenly interrupted. After plugging and unplugging the network cable, the device still did not recover. I tried to log in to the device through the serial port and captured packets on the eth1 port. At this time, the network card entered promiscuous mode and found that the ping was successful again. However, after exiting the promiscuous mode, the ping failed again. Finally, try ifconfig eth1 down/up on the eth1 port. At this time, ping is normal again. It is suspected that the AM5716 has the same problem as the previous AM335x, and the problem is not easy to reproduce.
Regards
Zekun
