Can Spark directory lookup for partition filtering be parallel?

Question

I have a large number of directories and use Spark partition filtering to read a DataFrame. I find that I look up the partition directories directly on the file system, I can do