Hive run-time job initialization Failed:java.io.IOException:Filesystem closed

A new problem appears in the Hivesql statement that has recently run a read-write multiple times, as shown in the figure:

In the MapReduce Management page of Hadoop, the following error message is shown:

At present, there is no solution in ............

It's strange that the same SQL statement has such a problem that remains to be seen.

For this issue, because FileSystem is a single object, when the JVM is set to reuse this parameter, the previous task will be closed after the completion of the filesystem, then the
This error is thrown when the task determines whether this file object is open.
Two ways to solve the problem:
One, do not set JVM reuse.
One, the configuration filesystem is not cached and an object is re-created each time.