[SPARK-33578] enableHiveSupport is invalid after sparkContext that without hive support created

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Won't Fix
• Affects Version/s: 3.1.0
• Fix Version/s: 3.1.0
• Component/s: Spark Core
• Assignee: Unassigned
• Reporter: steven zhang
• Created: 27/Nov/20 12:17
• Updated: 30/Nov/20 02:16
• Resolved: 30/Nov/20 02:16

Description

reproduce as follow code:

        SparkConf sparkConf = new SparkConf().setAppName("hello");

        sparkConf.set("spark.master", "local");

        JavaSparkContext jssc = new JavaSparkContext(sparkConf);

        spark = SparkSession.builder()

                .config("spark.serializer", "org.apache.spark.serializer.KryoSerializer")

                .config("hive.exec.dynamici.partition", true).config("hive.exec.dynamic.partition.mode", "nonstrict")

                .config("hive.metastore.uris", "thrift://hivemetastore:9083")

                .enableHiveSupport()

                .master("local")

                .getOrCreate();

       spark.sql("select * from hudi_db.hudi_test_order").show();

 

 it will produce follow Exception    

AssertionError: assertion failed: No plan for HiveTableRelation [`hudi_db`.`hudi_test_order` … (at current master branch)  org.apache.spark.sql.AnalysisException: Table or view not found: `hudi_db`.`hudi_test_order`;  (at spark v2.4.4)

  

 the reason is SparkContext#getOrCreate(SparkConf) will return activeContext that include previous spark config if it has

but the input SparkConf is the newest which include previous spark config and options.

  enableHiveSupport will set options (“spark.sql.catalogImplementation", "hive”) when spark session created it will miss this conf

SharedState load conf from sparkContext and will miss hive catalog

Issue Links

• links to: [Github] Pull Request #30522 (yui2010)