Stamatis Zampetakis created HIVE-30034:
------------------------------------------
Summary: Add TPCDS planner tests for Iceberg tables
Key: HIVE-30034
URL: https://issues.apache.org/jira/browse/HIVE-30034
Project: Hive
Issue Type: Improvement
Components: Test
Reporter: Stamatis Zampetakis
Assignee: Stamatis Zampetakis
{{TestTezTPCDS30TBPerfCliDriver}} has planner tests (EXPLAIN variations) for
TPC-DS queries on transactional ORC tables. It relies on metadata and
statistics extracted from a running metastore instance by performing a database
dump.
Iceberg tables are self-contained and their metadata and stats are part of the
table directory (use of metastore is minimal). Query planning for iceberg
tables hits many different APIs so to guard against regressions and also track
plan changes it would be beneficial to extend test coverage for Iceberg by
running EXPLAIN commands for all TPC-DS queries.
The plan is the following:
1. Download the metadata & stat files of Iceberg tables from a real cluster
with TPC-DS dataset loaded (no data files)
2. Restore those in the test infrastructure
3. Create necessary drivers and configs to run the existing TPC-DS explain
statements over the restored Iceberg tables
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
