apache / apache/gluten

[CH] Six ClickHouse suites run only on gluten-ut/spark33 and will stop running when it is removed

Open
#12,891 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Scala
Stars
1.6k
Forks
657
Avg merge
3d 17h
Merged PRs (30d)
86

Description

### What

Six ClickHouse suites are enabled on `gluten-ut/spark33` and on no surviving version module:

`GlutenInnerJoinSuite`, `GlutenOuterJoinSuite`, `GlutenJsonSuite`, `GlutenOrcQuerySuite`, `GlutenPercentileSuite`, `QueryTestSuite`

`backends-clickhouse/pom.xml` has only `spark-3.3` and `spark-3.5` profiles, so ClickHouse coverage exists on exactly those two modules. When `gluten-ut/spark33` is removed under #12807, whatever these six cover on ClickHouse stops running anywhere.

### Two of the six are less clear-cut than the list suggests

`GlutenInnerJoinSuite` and `GlutenOuterJoinSuite` have renamed variants on spark35, `GlutenInnerJoinSuiteForceShjOn` and `...ForceShjOff`, so the coverage may already be there under a different name. `QueryTestSuite` comes from Spark's own test jar rather than from a module file. That leaves `GlutenJsonSuite`, `GlutenOrcQuerySuite` and `GlutenPercentileSuite` as the ones most likely to be a genuine hole.

### Why this is a separate issue

Whether to enable them on the ClickHouse 3.5 module is a ClickHouse-maintainer call, not part of removing Spark 3.3 support: enabling a suite that was never run there is likely to surface real failures that need real fixes. #12863 settled the vanilla-case axis and did not cover this. Filing it so the decision is recorded rather than made silently by a deletion.

Related: #12807, #12863, #12889.

Contributor guide

Open the contributing guide

Research direction

Start with backends-clickhouse/pom.xml and compare the spark-3.3 and spark-3.5 profiles, then inspect the six named suites and their renamed Spark 3.5 variants. Determine which suites still lack ClickHouse coverage after #12807; done means recording the maintainer decision and enabling the appropriate suites on the surviving module, with any resulting failures identified.

Written by the indexing model from the issue text.

Assessment

Tech stack
clickhouse, scala
Domain
databases, testing-qa
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.