What is the common pool, and why does it cause trouble?

ForkJoinPool.commonPool() is a static, shared, lazily created pool sized at availableProcessors() - 1. Every parallel stream in the JVM uses it, including ones inside libraries, and CompletableFuture's async methods use it when you do not pass an executor.

So a single parallel stream doing blocking I/O occupies a scarce global resource, and every other parallel operation in the process slows down or stalls. On a machine reporting one CPU, the common pool has a single worker, so a parallel stream gets that one thread plus the caller.

You can raise its size with -Djava.util.concurrent.ForkJoinPool.common.parallelism=N, which is a blunt global setting, or run the stream inside your own ForkJoinPool by submitting it there, which works because the stream uses the pool of the thread it runs on.