Skip to content

[Spark] Fix GCS connector compatibility with Spark 3.5.x #284

Open
alxtkr77 wants to merge 1 commit intomlrun:developmentfrom
alxtkr77:CEML-669-gcs-connector-fix
Open

[Spark] Fix GCS connector compatibility with Spark 3.5.x #284
alxtkr77 wants to merge 1 commit intomlrun:developmentfrom
alxtkr77:CEML-669-gcs-connector-fix

Conversation

@alxtkr77
Copy link
Copy Markdown
Member

@alxtkr77 alxtkr77 commented Apr 5, 2026

Summary

Downgrade gcs-connector from 3.1.10 to hadoop3-2.2.33, the stable Hadoop 3.x compatible version that works with Spark 3.5.x. The 3.1.10 version caused classpath/compatibility issues.

Changes Made

  • Swap gcs-connector-3.1.10-shaded.jar for gcs-connector-hadoop3-2.2.33-shaded.jar in Spark Dockerfile
  • Bump Spark image tag to 3.5.6-scala2.12-java17-ubuntu-1

Testing

  • Verified GCS connector jar exists and is valid in the built image
  • Verified GoogleHadoopFileSystem and GoogleHadoopFS classes load successfully in Spark
  • Verified GCS FileSystem can be instantiated (no classpath errors)

Reference

  • Jira: CEML-573

Downgrade gcs-connector from 3.1.10 to hadoop3-2.2.33, the stable
Hadoop 3.x compatible version that works with Spark 3.5.x.

- Swap gcs-connector-3.1.10-shaded.jar for gcs-connector-hadoop3-2.2.33-shaded.jar
- Bump Spark image tag to 3.5.6-scala2.12-java17-ubuntu-1
- Bump chart version to 0.11.0-rc.30

Reference: CEML-669
@alxtkr77 alxtkr77 force-pushed the CEML-669-gcs-connector-fix branch from c4dc227 to bef70bc Compare April 5, 2026 12:08
@alxtkr77 alxtkr77 changed the title [Spark] Fix GCS connector compatibility with Spark 3.5.x (CEML-669) [Spark] Fix GCS connector compatibility with Spark 3.5.x Apr 5, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant