[HIVE-24574] Add DIAGNOSE Statement

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Assignee: Unassigned
• Reporter: David Mollitor
• Created: 30/Dec/20 16:42
• Updated: 07/Jan/21 23:09

Description

Add a new statement to Hive called DIAGNOSE

DIAGNOSE [WITH (PARQUET|ORC|JSON|AVRO) RESULTS] ...

Returns a single binary (BLOB) column which contains a TAR-GZ file comprised of several other files:

• A JSON file containing HS2 version information, HS2 host name, date or query submission, query id(s) etc.
• The query itself (file name is MD5 of the query)
• EXPLAIN plan (file name is MD5 of the explain plan)
• SHOW CREATE for each table in the query (<table_name.sql>)
• The configuration of the session (set)
• The Hive logs generated by the query
• The processing engine logs generated by the query
• Any counters associated with the processing engine
• Optionally, the results of the query in a single file (file name is MD5 of the results)

Issue Links

• relates to: HIVE-24596 Explain ddl for debugging