MathNodes
commented
1 year ago This seems to be occurring on all platforms. I believe this is due to data not clearing out from the previous node and each node's used bandwidth is being summed. It would be good to create a dictionary of each nodes utilized bandwidth and sum ongoing usage from that with the starting point of the on-chain data.
Due to limitations of NICs on *nix systems, the reported bytes sent/bytes recieved on the wireguard interface is not immediately flushed and so those totals do not reflect accurate data.
This can be workedaround by storing bytes sent/bytes received upon connection in the NodeSwitch dictionary and subtracting those initial amounts from the scheduler task checking to make sure it is greater than or equal to zero.