I am seeing some strange behaviour where I send a continuous stream of non-reliable rpc’s from the client with a couple of transforms of data on a tick (so very often) - Then at some point I need to send a stream of data reliably from the client to the server. AFter I do this - and just try to resume sending the non-reliable RPC’s - they all fail to send. After some digging I find that the queued bits in the network driver start to accumulate and grow substantially every frame. (this is all on the client)
Along the way there are other intermittent RPC’s (reliable) that get mixed in a little with the stream of data.
My suspicion is that during the time when I send a bunch of data reliably some other reliable data gets queued and then never gets un-queued and ends up blocking other transmission.
Any input out there? Looking through the code it seems that all non-reliable rpc’s get blocked if the queued bits is above some threshold.
Also, any suggestions on debugging this?