配置参数:
spark.hadoop.hive.exec.orc.default.stripe.size=78643200
spark.hadoop.orc.stripe.size=78643200
spark.hadoopRDD.targetBytesInPartition=78643200
spark.hadoop.hive.exec.dynamic.partition.mode=nonstrict
spark.sql.sources.partitionOverwriteMode=dynamic
spark.sql.hive.convertMetastoreOrc=true
代码里落表前的.repartition(5000)
这种要删掉
spark.sql.shuffle.partitions=5000
这个配置参数也要删掉