andygrove opened a new pull request, #3997:
URL: https://github.com/apache/datafusion-comet/pull/3997
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases. You can
link an issue to this PR using the GitHub syntax. For example `Closes #123`
indicates that this PR will close issue #123.
-->
Closes #.
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in
the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your
changes and offer better suggestions for fixes.
-->
The existing `.github/workflows/docker-publish.yml` triggered on release
tags but only built a single Spark 3.5 / Scala 2.12 / Java 11 variant and
pushed to GHCR. The Docker Hub images users actually reference in the
kubernetes docs
(`apache/datafusion-comet:<version>-spark<X.Y.Z>-scala<X.Y>-java<XX>`) cover
the full supported matrix and have been published out-of-band by the release
manager. This replaces that ad-hoc process with a single repeatable manual
script.
## What changes are included in this PR?
- New `kube/release.Dockerfile`: a thin, parameterized image that drops a
pre-built Comet uber-jar into an `apache/spark` base. Uses `ARG SPARK_IMAGE` +
`ARG COMET_JAR`, restores the non-root `spark_uid` (defaulted to 185 to match
upstream). Leaves the existing `kube/Dockerfile` (build-from-source, referenced
in the user guide) untouched.
- New `dev/release/build-docker-images.sh`: orchestration script that
iterates the five supported Spark x Scala combos (all Java 17), resolves jars
from the staging Maven repo printed by `dev/release/build-release-comet.sh`,
and invokes `docker buildx build --platform linux/amd64,linux/arm64 --push`
once per combo. Includes arg parsing, pre-flight checks (docker, buildx,
platforms, credentials), `--dry-run` mode, temp-context staging, EXIT-trap
cleanup, and a final summary. The same uber-jar works on both architectures
because the jar already bundles `linux/amd64` and `linux/aarch64` native libs.
- Removed `.github/workflows/docker-publish.yml`: replaced by the manual
script.
- Updated `docs/source/contributor-guide/release_process.md`: added a new
"Publish Docker images to Docker Hub" section, two checklist items (one for the
RC stage, one for the final-release stage), and replaced the two inline
references to the deleted GHCR workflow with pointers to the new section.
Published tag scheme (five multi-arch manifests per release):
- `apache/datafusion-comet:<version>-spark3.4.3-scala2.12-java17`
- `apache/datafusion-comet:<version>-spark3.4.3-scala2.13-java17`
- `apache/datafusion-comet:<version>-spark3.5.8-scala2.12-java17`
- `apache/datafusion-comet:<version>-spark3.5.8-scala2.13-java17`
- `apache/datafusion-comet:<version>-spark4.0.1-scala2.13-java17`
## How are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are
they covered by existing tests)?
-->
This is release tooling that cannot be CI-tested without actually publishing
to Docker Hub, so verification is manual:
- `kube/release.Dockerfile` built locally with a stub jar against
`apache/spark:3.5.8-java17`; confirmed the jar lands at `/opt/spark/jars/` and
the container runs as `uid=185(spark)`.
- `build-docker-images.sh` exercised via `--dry-run` against a stub Maven
repo layout: verified arg parsing (happy path, missing args, unknown args,
non-existent repo path), pre-flight checks (docker, buildx, platform list,
credentials), missing-jar failure path, and end-to-end dry-run output showing
all five combos plus the final summary.
- A first real release using this tool on a personal Docker Hub namespace is
described as the canonical end-to-end smoke test in the script's top-of-file
comment. The first actual release using the tool will be the final validation.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
