We should also provide some guidance/doc/readme to tell users to limit the number of max_workers as https://github.com/netty/netty-incubator-transport-io_uring/issues/106 shows that people would get a huge amount of time spent by lot of kernel threads to async complete unbound I/O work
https://github.com/torvalds/linux/commit/90fa02883f063b971ebfd9f5b2184b38b83b7ee3 has introduced IOSQE_ASYNC_HYBRID that would be beneficial "in cases where unbound max_worker is relatively small" (quoting the commit msg).
We should also provide some guidance/doc/readme to tell users to limit the number of max_workers as https://github.com/netty/netty-incubator-transport-io_uring/issues/106 shows that people would get a huge amount of time spent by lot of kernel threads to async complete unbound I/O work