[SPARK-24766] CreateHiveTableAsSelect and InsertIntoHiveDir won't generate decimal column stats in parquet

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.0
• Fix Version/s: 3.0.0
• Component/s: SQL
• Labels: Parquet
• Assignee: Unassigned
• Reporter: Yuming Wang
• Created: 10/Jul/18 15:27
• Updated: 27/May/19 03:20
• Resolved: 27/May/19 03:20

Description

How to reproduce:

INSERT OVERWRITE LOCAL DIRECTORY '/tmp/spark/parquet/dir' STORED AS parquet select cast(1 as decimal) as decimal1;

create table test_parquet stored as parquet as select cast(1 as decimal) as decimal1;

$ java -jar ./parquet-tools/target/parquet-tools-1.10.1-SNAPSHOT.jar meta  file:/tmp/spark/parquet/dir/part-00000-cb96a617-4759-4b21-a222-2153ca0e8951-c000

file:        file:/tmp/spark/parquet/dir/part-00000-cb96a617-4759-4b21-a222-2153ca0e8951-c000

creator:     parquet-mr version 1.6.0 (build 6aa21f8776625b5fa6b18059cfebe7549f2e00cb)



file schema: hive_schema

--------------------------------------------------------------------------------

decimal1:    OPTIONAL FIXED_LEN_BYTE_ARRAY O:DECIMAL R:0 D:1



row group 1: RC:1 TS:46 OFFSET:4

--------------------------------------------------------------------------------

decimal1:     FIXED_LEN_BYTE_ARRAY SNAPPY DO:0 FPO:4 SZ:48/46/0.96 VC:1 ENC:BIT_PACKED,PLAIN,RLE ST:[no stats for this column]

because spark still use com.twitter.parquet-hadoop-bundle.1.6.0.

May be we should refactor CreateHiveTableAsSelectCommand and InsertIntoHiveDirCommand or upgrade built-in Hive.

Issue Links

• relates to: SPARK-23710 Upgrade the built-in Hive to 2.3.5 for hadoop-3.2