Skip to content

Conversation

@stankiewicz
Copy link
Contributor

@stankiewicz stankiewicz commented Dec 2, 2025

Changes:

  • dependencies definitions to otel
  • WindowedValue interface change to expose context for propagation - context is not read by user
  • changes are behind external_metadata experiment

Next PRs:

  • propagation to windmill sink
  • reading from windmill reader
  • enabling default otel implementation for dataflow runner (v1 with propagation, v2 no propagation yet)

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

  • Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
  • Update CHANGES.md with noteworthy changes.
  • If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

Build python source distribution and wheels
Python tests
Java tests
Go tests

See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.

@stankiewicz
Copy link
Contributor Author

R: @sjvanrossum

@gemini-code-assist
Copy link
Contributor

Summary of Changes

Hello @stankiewicz, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request lays the foundational groundwork for integrating OpenTelemetry context propagation within Apache Beam. It involves updating OpenTelemetry dependencies, extending the internal data model (beam_fn_api.proto and WindowedValue interface) to carry trace context information, and implementing the necessary serialization/deserialization logic. These changes are part of an ongoing effort to enhance observability by enabling end-to-end distributed tracing across Beam pipelines, currently under an experimental flag.

Highlights

  • OpenTelemetry Dependency Updates: Updated and added several OpenTelemetry dependencies in BeamModulePlugin.groovy and various build.gradle files to support OpenTelemetry context propagation.
  • Protocol Buffer Extension for Trace Context: Modified beam_fn_api.proto to include traceparent and tracestate fields within the ElementMetadata message, enabling the serialization of W3C Trace Context.
  • WindowedValue Interface Enhancement: Extended the WindowedValue interface and its implementations to include a getContext() method, allowing the OpenTelemetry Context to be associated with and propagated alongside data elements.
  • OpenTelemetry Context Serialization: Introduced OpenTelemetryContextSerializer to handle the encoding and decoding of OpenTelemetry Context objects into and from the ElementMetadata of WindowedValue using the W3C Trace Context Propagator.
  • Experimental Feature: These changes are implemented behind an external_metadata experiment flag, indicating an incremental rollout and testing approach.
Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature Command Description
Code Review /gemini review Performs a code review for the current pull request in its current state.
Pull Request Summary /gemini summary Provides a summary of the current pull request in its current state.
Comment @gemini-code-assist Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help /gemini help Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

  1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution.

@github-actions
Copy link
Contributor

github-actions bot commented Dec 2, 2025

Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control. If you'd like to restart, comment assign set of reviewers

@stankiewicz
Copy link
Contributor Author

/gemini review

Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

The pull request effectively integrates OpenTelemetry context propagation into Apache Beam's core data structures and serialization mechanisms. This foundational change is well-implemented across various files, ensuring the Context is carried alongside WindowedValue elements. Key aspects include updating dependencies, extending the protocol buffer schema for traceparent and tracestate, modifying core WindowedValue interfaces and implementations to include the Context, and implementing robust serialization/deserialization logic via OpenTelemetryContextSerializer. The addition of a test case in WindowedValueTest.java validates the new functionality. The use of null for getContext() in specific overrides is consistent with the stated scope that the context is not yet consumed by users. Overall, this is a solid and necessary step towards comprehensive OpenTelemetry integration.

classesInPackage("com.fasterxml.jackson.annotation"),
classesInPackage("com.google.cloud.hadoop.gcsio"),
classesInPackage("com.google.common.collect"), // Via gcs-connector ReadOptions builder
classesInPackage("io.opentelemetry"), // open telemetry
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@kennknowles I had to at this as we are exposing otel context in

		interface org.apache.beam.sdk.values.WindowedValue
		interface org.apache.beam.sdk.values.OutputBuilder
		interface org.apache.beam.sdk.transforms.DoFn$OutputReceiver

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is fine. I actually think this test code-rotted some time ago, and also its purpose is somewhat obsolete: Before we released "vendor" versions of Guava and Calcite and other things, we relocated and included them in the built jars dynamically - so our codebase looked like it linked to the normal version, but during jar assembly it was relocated to the vendor namespace. So we accidentally could ship something with an API that could never be used. This test was to make sure we didn't do that. Now that we (mostly) vendor things, it is less important. We probably still have some uses of dynamic relocation, but it isn't "everything/everywhere" like it used to be.

(I had a philosophy, which turns out to be a PITA, that we should relocate everything except for allowlisted API surface classes, to keep dependencies to only actual public deps. This is slow, causes massive bloat, and IDEs do not understand it. I still like the idea that "implementation detail" dependencies do not impact users but it needs language and runtime support, not hacking in the build system - the "right" way in Java is probably classloaders, ugh)

classesInPackage("com.fasterxml.jackson.annotation"),
classesInPackage("com.google.cloud.hadoop.gcsio"),
classesInPackage("com.google.common.collect"), // Via gcs-connector ReadOptions builder
classesInPackage("io.opentelemetry"), // open telemetry
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is fine. I actually think this test code-rotted some time ago, and also its purpose is somewhat obsolete: Before we released "vendor" versions of Guava and Calcite and other things, we relocated and included them in the built jars dynamically - so our codebase looked like it linked to the normal version, but during jar assembly it was relocated to the vendor namespace. So we accidentally could ship something with an API that could never be used. This test was to make sure we didn't do that. Now that we (mostly) vendor things, it is less important. We probably still have some uses of dynamic relocation, but it isn't "everything/everywhere" like it used to be.

(I had a philosophy, which turns out to be a PITA, that we should relocate everything except for allowlisted API surface classes, to keep dependencies to only actual public deps. This is slow, causes massive bloat, and IDEs do not understand it. I still like the idea that "implementation detail" dependencies do not impact users but it needs language and runtime support, not hacking in the build system - the "right" way in Java is probably classloaders, ugh)

opentelemetry_api : "io.opentelemetry:opentelemetry-api", // google_cloud_platform_libraries_bom sets version
opentelemetry_api : "io.opentelemetry:opentelemetry-api:$opentelemetry_version",
opentelemetry_bom : "io.opentelemetry:opentelemetry-bom-alpha:$opentelemetry_version-alpha", // alpha required by extensions
opentelemetry_context : "io.opentelemetry:opentelemetry-context:$opentelemetry_version",
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will this change conflict with the GCP BOM? @suztomo do you think this is fine? We might just need to check them together manually.

}

@Override
public @Nullable Context getContext() {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

getContext is pretty generic, even though it is accurate in this context (pun intended). Perhaps getOpenTelemetryContext even though it is longer it'll be more readable.

.setPaneInfo(template.getPaneInfo());
.setPaneInfo(template.getPaneInfo())
.setContext(template.getContext())
.setCausedByDrain(template.causedByDrain())
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All this propagation of other fields would be valuable to commit separately for code history. Perhaps jj split these to other PRs, or make them tiny commits in this PR.

@Override
public WindowedValue<T> decode(InputStream inStream) throws CoderException, IOException {
return decode(inStream, Context.NESTED);
return decode(inStream, org.apache.beam.sdk.coders.Coder.Context.NESTED);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should be able to be Coder.Context if you want.

}

public static Context read(BeamFnApi.Elements.ElementMetadata from) {
return W3CTraceContextPropagator.getInstance().extract(Context.root(), from, GETTER);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK... this way of doing things matches https://opentelemetry.io/docs/specs/otel/context/api-propagators/

But it does seem like a complicated way of doing the equivalent of

W3TraceContextPropagator.getInstance().setTracestate(elementMetadata.getTracestate());
W3TraceContextPropagator.getInstance().setTraceparent(elementMetadata.getTraceparent());

and

elementMetadataBuilder.setTracestate(W3TraceContextPropagator.getInstance().getTracestate());
elementMetadataBuilder.setTraceparent(W3TraceContextPropagator.getInstance().getTraceparent());

Other nit: did we not agree that the ElementMetadata object should be only implementation detail? This class and everything should be private since it is just for encoding/decoding utility. But potentially also you want to do the SETTER on an Builder and the GETTER on a WindowedValue. If it is implementation detail, hidden, then I don't care that much, but if it is public then we have to be careful about what classes we use.

}
}

public static class OpenTelemetryContextSerializer {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

private? Possibly package level access (maybe it is just @VisibleForTesting? I didn't audit the whole PR...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants