[HIVE-18352] introduce a METADATAONLY option while doing REPL DUMP to allow integrations of other tools

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 3.0.0
• Labels: pull-request-available
• Assignee: Anishek Agarwal
• Reporter: Anishek Agarwal
• Created: 02/Jan/18 04:58
• Updated: 27/Jul/18 05:13
• Resolved: 12/Jan/18 05:44

Description

• Introduce a METADATAONLY option as part of the REPL DUMP command which will only try and dump out events for DDL changes, this will be faster as we wont need scan of files on HDFS for DML changes.
• Additionally since we are only going to dump metadata operations, it might be useful to include acid tables as well via an option as well. This option can be removed when ACID support is complete via HIVE-18320

it will be good to support the "WITH" clause as part of REPL DUMP command as well (repl dump already supports it viaHIVE-17757) to achieve the above as that will prevent less changes to the syntax of the statement and provide more flexibility in future to include additional options as well.

REPL DUMP [db_name] {FROM [event_id]} {TO [event_id]} {WITH (['key'='value'],.....)}

This will enable other tools like security / schema registry / metadata discovery to use replication related subsystem for their needs as well.

Issue Links

• links to: GitHub Pull Request #286