Hi, I have not seen any errors with ping on when performing a long duration test on a TI board. I typically see errors when bringing up a board or through misuse, both neither of those are applicable here. The only suggestions I can make is that you will have to prove the endpoint that broke the cycle. Running over a million packets must have taken hours to hit this issue. ICMP is not a considered a reliable protocol since there is not any guarantees that all packets will delivered. Since there are only two points I would check to see if there are any interface HW errors such as RX CRC, overruns etc. that are reported with ethtool on both end points here. Best Regards, Schuyler
↧
