Aha! Redstone wire performs sub-optimally near chunk boundaries, causing mechanisms to fail to correctly function.
How is it that Linux/MCServer consistently responds faster on a port that is software-redirected to another (say, redirect 80 to 25565) than the original port (25565)? Surely the software would increase latency, not decrease it?
It seems something strange is happening in the saving/loading process of blocks near chunks. Freshly placed wire/repeaters perform perfectly, as expected. However, a reload introduces strange behaviour. This isn't related to the recent load-redstone-on-chunk-load changes.
How is it that Linux/MCServer consistently responds faster on a port that is software-redirected to another (say, redirect 80 to 25565) than the original port (25565)? Surely the software would increase latency, not decrease it?
It seems something strange is happening in the saving/loading process of blocks near chunks. Freshly placed wire/repeaters perform perfectly, as expected. However, a reload introduces strange behaviour. This isn't related to the recent load-redstone-on-chunk-load changes.