There is no problem when the pyspark program starts running. If I let the job run for a long time, there are py4j errors.
2020-09-11 06:55:00 ERROR JobScheduler:
