My proposal is to try to use TCP sockets.
We could increase number of workers, but we still could have problems at the peak load. TCP socket backlog should help with that. (http://veithen.github.io/2014/01/01/how-tcp-backlog-works-in-linux.html)
My proposal is to try to use TCP sockets.
We could increase number of workers, but we still could have problems at the peak load. TCP socket backlog should help with that. (http:// veithen. github. io/2014/ 01/01/how- tcp-backlog- works-in- linux.html)