Hive on Spark will resubmitted application when not enough resouces to launch yarn application master

Hive on spark will estimate reducer number when the query is not set reduce number,which cause a application submit.The application will pending if the yarn queue's resources is insufficient.
So there are more than one pending applications probably because
there are more than one estimate call.The failure is soft, so it doesn't prevent subsequent processings. We can make that a hard failure

That code is found in
at org.apache.hadoop.hive.ql.exec.spark.SparkUtilities.getSparkSession(SparkUtilities.java:112)
at org.apache.hadoop.hive.ql.optimizer.spark.SetSparkReducerParallelism.process(SetSparkReducerParallelism.java:115)