Closed gizahNL closed 9 months ago
There are many rx_hw_dropped_packets
errors which usually caused by software is not fetching the packet in time.
Can you run below command and share us the log to analyze the possible cause? Tweak the PCIE port in 1080p59_1v.json to the actual ports in your setup.
./build/app/RxTxApp --config_file tests/script/loop_json/1080p59_1v.json --tasklet_time
This was solved. It was an issue inside our app related with thread priorities under linux. Please feel free to close this issue.
Thanks to let's know the cause. We has a ebpf tool https://github.com/OpenVisualCloud/Media-Transport-Library/blob/main/tools/ebpf/lcore_monitor.c
which can be used to monitor the IMTL lcore status.
Free to open a new one if any other issue.
We're testing the receiving end of our application but we're seeing significant loss of frames consistently on the input, I'm seeing 20 incomplete frames in every stat report. The amount of incomplete frames per stat period seems to be the same regardless of the framerate we operate at. Which would suggest it's not related to performance. In our test equipment we've tested a single leg going over a switch back into the other port on our E810 card, and we've tested sending with a VF device and receiving with another VF device on the same physical device.
logs (RX):
Since the sender is also an MTL utilizing application here are the TX logs: