Yep. Both the rx and tx buffers were full and the port was dropping packets. (There were both container ships waiting to offload and trucks waiting to offload full and empty containers, respectively.) Now that buffer size has been increased, there is more bandwidth to available to actually move containers. Since the spike is temporary, the problem has gone away. If we were permanently faced with more containers then, yes, we'd need another port.
Was this just caused by just a spike in demand? It seems to me the pretty would only full up with empties if there is some kind of imbalance between shipping and receiving, right?
As Ryan pointed out in his Twitter thread the bottleneck "should" be the cranes.
While it's a long-running gag, I would like to remind everyone that swapping to zram is a thing on Linux (I don't know whether there's an equivalent on Darwin, and I think NT ships it by default), and the tech goes all the way back to RAM Doubler for MacOS (or at least, that's the earliest implementation that I know about). So for quite a long time you kind of have been able to do exactly that...
