If the partition count or kafka IO size is large, then skip committin… by kishorepola · Pull Request #37510 · apache/beam

kishorepola · 2026-02-05T02:32:12Z

…g offsets that are not changed. Reduce kafka commit load

Please add a meaningful description for your change here

While committing offsets back to Kafka, Beam commits offsets back to all the topics and partitions in the KafkaIO. If some topic-partitions are idle, even then the same old offset is committed back. This causes lot of metadata pressure on the brokers if the kafka cluster has lot of idle partitions or cluster size is decently big.
Added in memory tracking for offsets.
Commit back only those offsets that are modified.

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)

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

…g offsets that are not changed. Reduce kafka commit load

gemini-code-assist · 2026-02-05T02:32:28Z

Summary of Changes

Hello @kishorepola, 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 optimizes the Kafka offset commitment process within Apache Beam's KafkaIO connector. Previously, the system would commit offsets for all topic-partitions, even if some were idle and their offsets hadn't changed. This behavior could lead to excessive metadata load on Kafka brokers. The implemented solution introduces in-memory tracking of the last committed offsets and ensures that only partitions with advanced offsets are committed, thereby reducing the commit load and improving overall system efficiency.

Highlights

Optimized Kafka Offset Commits: Introduced a mechanism to track the last successfully committed offsets for each topic-partition, preventing redundant commits for idle partitions.
Reduced Kafka Broker Load: Modified the commitCheckpointMark method to only commit offsets that have advanced, significantly reducing metadata pressure on Kafka brokers, especially in large-scale deployments.
Improved Efficiency: The change avoids unnecessary consumer.commitSync() calls when no offsets have changed, leading to more efficient resource utilization.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java
- Added java.util.HashMap import to support the new offset tracking mechanism.
- Introduced a new private field lastCommittedOffsets (a Map<TopicPartition, Long>) to store the last known committed offset for each topic-partition.
- Refactored the commitCheckpointMark method to: 1) Filter out partitions with uninitialized offsets. 2) Compare the current nextOffset with the lastCommittedOffsets to identify only those partitions whose offsets have genuinely advanced. 3) Construct a map (toCommit) containing only the advanced offsets. 4) Skip the consumer.commitSync() call if toCommit is empty, avoiding unnecessary operations. 5) Update lastCommittedOffsets only after a successful consumer.commitSync() operation.

Activity

The pull request was created by kishorepola to address high Kafka commit load due to redundant offset commits for idle partitions.

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.

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 · 2026-02-05T04:18:00Z

Assigning reviewers:

R: @ahmedabu98 for label java.
R: @sjvanrossum for label kafka.

Note: If you would like to opt out of this review, comment assign to next reviewer.

Available commands:

stop reviewer notifications - opt out of the automated review tooling
remind me after tests pass - tag the comment author after tests pass
waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

github-actions · 2026-02-12T12:27:07Z

Reminder, please take a look at this pr: @ahmedabu98 @sjvanrossum

gauravmishraitbhu · 2026-02-14T16:03:19Z

Hi @tomstepp , can you help review this.

tomstepp

Thanks for contributing this!

tomstepp · 2026-02-17T15:40:19Z

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java

+          toCommit.put(tp, new OffsetAndMetadata(next));
+        }
+      }
+


Maybe we could we add a debug log for # idle partitions?

tomstepp · 2026-02-17T15:54:24Z

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java

 class KafkaUnboundedReader<K, V> extends UnboundedReader<KafkaRecord<K, V>> {

+  // Track last successfully committed offsets to suppress no-op commits for idle partitions.
+  private final Map<TopicPartition, Long> lastCommittedOffsets = new HashMap<>();


Maybe we can also track last commit time per partition? We could try to commit if idle for more than some time (10 minutes for example).

I think this could also help for cases where customers may use time lag monitoring (tracking time since last commit).

tomstepp · 2026-02-17T15:58:52Z

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java

Can we add a unit test to cover this new behavior?

From a quick search maybe reuse or similar to sdks/java/io/kafka/src/test/java/org/apache/beam/sdk/io/kafka/KafkaCommitOffsetTest.java

johnjcasey · 2026-02-17T16:17:37Z

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java

+        Long prev = lastCommittedOffsets.get(tp);
+
+        if (prev == null || next > prev) {
+          toCommit.put(tp, new OffsetAndMetadata(next));


I like the idea of this change. Can we keep the existing java streams logic, but simply add a new filter step?

If the partition count or kafka IO size is large, then skip committin…

68bd4b2

…g offsets that are not changed. Reduce kafka commit load

github-actions bot added java io kafka labels Feb 5, 2026

github-actions bot added the Next Action: Reviewers label Feb 5, 2026

github-actions bot added the slow-review label Feb 12, 2026

github-actions bot removed the slow-review label Feb 14, 2026

tomstepp suggested changes Feb 17, 2026

View reviewed changes

johnjcasey reviewed Feb 17, 2026

View reviewed changes

Abacn added Next Action: Author and removed Next Action: Reviewers labels Feb 17, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

If the partition count or kafka IO size is large, then skip committin…#37510

If the partition count or kafka IO size is large, then skip committin…#37510
kishorepola wants to merge 1 commit intoapache:masterfrom
kishorepola:commit-only-modified-offsets-for-big-kafka-cluster

kishorepola commented Feb 5, 2026

Uh oh!

gemini-code-assist bot commented Feb 5, 2026

Uh oh!

github-actions bot commented Feb 5, 2026

Uh oh!

github-actions bot commented Feb 12, 2026

Uh oh!

gauravmishraitbhu commented Feb 14, 2026

Uh oh!

tomstepp left a comment

Uh oh!

tomstepp Feb 17, 2026

Uh oh!

tomstepp Feb 17, 2026

Uh oh!

tomstepp Feb 17, 2026

Uh oh!

johnjcasey Feb 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

kishorepola commented Feb 5, 2026

GitHub Actions Tests Status (on master branch)

Uh oh!

gemini-code-assist bot commented Feb 5, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

github-actions bot commented Feb 5, 2026

Uh oh!

github-actions bot commented Feb 12, 2026

Uh oh!

gauravmishraitbhu commented Feb 14, 2026

Uh oh!

tomstepp left a comment

Choose a reason for hiding this comment

Uh oh!

tomstepp Feb 17, 2026

Choose a reason for hiding this comment

Uh oh!

tomstepp Feb 17, 2026

Choose a reason for hiding this comment

Uh oh!

tomstepp Feb 17, 2026

Choose a reason for hiding this comment

Uh oh!

johnjcasey Feb 17, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants