Trino: Add Trino Docker Compose for integration testing #2220

dingo4dev · 2025-07-18T11:10:24Z

Rationale for this change

This pull request introduces a Trino service to the Docker Compose development environment, addressing the need for broader integration testing capabilities as discussed in issue #2219

Are these changes tested?

Are there any user-facing changes?

Copilot

Pull Request Overview

This PR introduces Trino integration testing capabilities to the PyIceberg project by adding Docker Compose configuration and test fixtures for Trino services. The changes enable testing of Iceberg operations through Trino's SQL interface to ensure compatibility and proper integration between PyIceberg and Trino.

Key changes include:

Added Trino Docker services configuration with both REST and Hive catalog support
Introduced new test fixtures for Trino database connections
Created integration tests that verify PyIceberg operations work correctly with Trino

Reviewed Changes

Copilot reviewed 11 out of 12 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
`tests/integration/test_writes/test_writes.py`	Added new Trino-specific UUID partitioning test and skipped existing Spark test
`tests/integration/test_rest_catalog.py`	Added test to verify Iceberg namespace appears as Trino schema
`tests/integration/test_register_table.py`	Added test to verify table registration works with Trino
`tests/conftest.py`	Added Trino connection fixtures and command-line options
`pyproject.toml`	Added Trino dependency and integration test marker
`dev/trino/catalog/*.properties`	Trino catalog configuration files for REST and Hive catalogs
`dev/run-trino.sh`	Shell script to manage Trino Docker services
`dev/docker-compose-trino.yml`	Standalone Trino Docker Compose configuration
`dev/docker-compose-integration.yml`	Added Trino service to main integration environment
`Makefile`	Added target for running Trino integration tests

Comments suppressed due to low confidence (2)

tests/integration/test_register_table.py:109

[nitpick] The test doesn't handle the case where Trino connection might fail or timeout. Consider adding error handling or retries for database connectivity issues that could make the test flaky.

    assert table_name in inspect(trino_conn).get_table_names(schema=namespace)

tests/integration/test_rest_catalog.py

Adds a Docker Compose configuration for setting up Trino with Iceberg, including support for Hive Metastore and REST catalog types. This allows for easier testing and development with Trino and Iceberg. closes apache#2219

Add Trino as alternative tool for test uuid partitions as Java Iceberg 1.9.2 in spark is not yet supported.

raulcd

Thank for the PR!

raulcd · 2025-11-14T08:44:31Z

tests/integration/test_writes/test_writes.py

    assert tbl.scan().to_arrow() == arrow_table


+@pytest.mark.skip("UUID BucketTransform is not supported in Spark Iceberg 1.9.2 yet")


is this test skip related to the trino work?
I am curious why this wasn't skipped before

@raulcd, Thanks for your reviews.

Before this PR, the bucket transform is not test in this testcase because we found the issue in Iceberg Java.

iceberg-python/tests/integration/test_writes/test_writes.py

Lines 2093 to 2102 in f2131e6

@pytest.mark.parametrize(

"transform",

[

IdentityTransform(),

# Bucket is disabled because of an issue in Iceberg Java:

# https://github.com/apache/iceberg/pull/13324

# BucketTransform(32)

],

)

def test_uuid_partitioning(session_catalog: Catalog, spark: SparkSession, transform: Transform) -> None: # type: ignore

My thoughts is using Trino to test this uuid partition testcase for Identity & Bucket Transforms, instead of using Spark Iceberg to test it, because the UUID Bucket Transform is not supported in Spark but it is worked in Trino.

To isolate the current issue in Java Iceberg, I added new testcase for it in: https://github.com/dingo4dev/iceberg-python/blob/3eef2566e02342d65a2c2f650bfb6c62d0221cf3/tests/integration/test_writes/test_writes.py#L2161

I see, thanks for the comment. I understand you are adding a new test that is covering both the BucketTransform and the IdentityTransform on test_uuid_partitioning_with_trino but I don't think adding the conditional skip on this specific test is necessary. Either we leave it as is, the test will still run on non-trino set up or we remove it entirely as the transform is covered on the trino setup. I would rather just maintain it to be honest because if the Iceberg Java issue is fixed in the future we can just also add the BucketTransform to run unconditionally.

@raulcd You're right! I just found that is a way to add the skip marker to specific parameter, so that we can keep the partition test running on the Spark, and also able to skip the BucketTransform and easy to tracked it instead of using comments, tho. How does that sound?

sounds good in principle, if you can push or show the diff I might understand better what you mean :)
Thanks for looking at this!

Sure ! I just updated it. Please have a look ;)

The change lgtm! Thanks, I'll see if I have some time to go over the changes in detail to properly review but we will still need a committer review :)

ebyhr · 2025-11-29T08:05:25Z

dev/trino/config.properties

@@ -0,0 +1,23 @@
+# Licensed to the Apache Software Foundation (ASF) under one


Why do we want to add config.properties‎ file?

Hi @ebyhr , Thanks for your review.
My first initial thought is to configure the catalog.management to dynamic which help to add and test the glue catalog and dynamo catalog without restart the container. Then, I though adding it will help development experience, you can modify server configure for your own resources without update dockerfile yourself

Thank you, I understand the motivation now. The next question is why enabling CATALOG_MANAGEMENT environment variable is insufficient.

I added catalog.management=${ENV:CATALOG_MANAGEMENT} to make sure it consistence while someone updating config.properties explicitly. Or we can just use catalog.management=dynamic and remove the env var.

ebyhr · 2025-11-29T08:06:01Z

dev/docker-compose-integration.yml

+    environment:
+      - CATALOG_MANAGEMENT=dynamic


This environment variable looks redundant since we're adding catalogs with properties files instead of CREATE CATALOG statement.

ebyhr · 2025-11-29T08:14:27Z

dev/trino/catalog/warehouse_rest.properties

+s3.aws-access-key=admin
+s3.aws-secret-key=password
+s3.endpoint=http://minio:9000
+s3.path-style-access=false


nit: We could remove this s3.path-style-access=false. It's disabled by default.

ebyhr · 2025-12-03T02:32:14Z

dev/docker-compose-integration.yml

+      - rest
+      - hive
+    volumes:
+      - ./trino/catalog:/etc/trino/catalog


nit: How about specifying the full path of properties files? Thus, we can use existing catalogs such memory and TPCH catalogs - they are helpful during development.

Sounds great!

kevinjqliu

Thanks for working on this! Its great to see another engine integration.

A couple of general comments:

Make Trino infra and integration tests optional for now and hook it up to the CI later
Consolidate the 3 different ways to spin up trino infra. My preference would be to create a new docker-compose-trino with just the trino infra and spin up that alongside the original dev/docker-compose-integration.yml.
Group all the infra related changes to dev/trino/, similar to what we do for spark
Create a separate test file for trino tests

Let me know what you think!

dingo4dev force-pushed the test-trino-integration branch 4 times, most recently from 04d1f61 to a0e1a11 Compare July 22, 2025 19:01

dingo4dev marked this pull request as ready for review July 22, 2025 19:02

Copilot AI review requested due to automatic review settings July 22, 2025 19:02

dingo4dev changed the title ~~[WIP] Trino: Add Trino Docker Compose for integration testing~~ Trino: Add Trino Docker Compose for integration testing Jul 22, 2025

Copilot AI reviewed Jul 22, 2025

View reviewed changes

tests/integration/test_rest_catalog.py Outdated Show resolved Hide resolved

dingo4dev force-pushed the test-trino-integration branch from df48ce0 to f46d70f Compare August 28, 2025 19:02

dingo4dev added 3 commits November 13, 2025 01:09

Adds Docker Compose setup for Trino

0c57d0c

Adds a Docker Compose configuration for setting up Trino with Iceberg, including support for Hive Metastore and REST catalog types. This allows for easier testing and development with Trino and Iceberg. closes apache#2219

add uuid partitions test with trino

60c4d13

Add Trino as alternative tool for test uuid partitions as Java Iceberg 1.9.2 in spark is not yet supported.

fix: correct conditions in namespace

06f7419

dingo4dev force-pushed the test-trino-integration branch from f46d70f to 06f7419 Compare November 12, 2025 17:11

dingo4dev added 2 commits November 13, 2025 01:15

fix: add license to trino config file

93b1187

remove precommit as prek exist

3eef256

raulcd reviewed Nov 14, 2025

View reviewed changes

use mark to skip BucketTransform in Spark uuid partition test

311f7b6

ebyhr reviewed Nov 29, 2025

View reviewed changes

ebyhr reviewed Dec 3, 2025

View reviewed changes

kevinjqliu reviewed Dec 27, 2025

View reviewed changes

		assert tbl.scan().to_arrow() == arrow_table


		@pytest.mark.skip("UUID BucketTransform is not supported in Spark Iceberg 1.9.2 yet")

	@pytest.mark.parametrize(
	"transform",
	[
	IdentityTransform(),
	# Bucket is disabled because of an issue in Iceberg Java:
	# https://github.com/apache/iceberg/pull/13324
	# BucketTransform(32)
	],
	)
	def test_uuid_partitioning(session_catalog: Catalog, spark: SparkSession, transform: Transform) -> None: # type: ignore

		@@ -0,0 +1,23 @@
		# Licensed to the Apache Software Foundation (ASF) under one

Trino: Add Trino Docker Compose for integration testing #2220

Are you sure you want to change the base?

Trino: Add Trino Docker Compose for integration testing #2220

Conversation

dingo4dev commented Jul 18, 2025

Rationale for this change

Are these changes tested?

Are there any user-facing changes?

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Uh oh!

raulcd left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dingo4dev Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ebyhr Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kevinjqliu left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

dingo4dev Dec 3, 2025 •

edited

Loading

ebyhr Dec 3, 2025 •

edited

Loading