Skip to content

Comments

Fix DebeziumIO resuming from worker restart#37689

Open
Abacn wants to merge 1 commit intoapache:masterfrom
Abacn:debezium-offset
Open

Fix DebeziumIO resuming from worker restart#37689
Abacn wants to merge 1 commit intoapache:masterfrom
Abacn:debezium-offset

Conversation

@Abacn
Copy link
Contributor

@Abacn Abacn commented Feb 23, 2026

Part of #28248

Resolves crash loop when worker restart (same pipeline).

Please add a meaningful description for your change here

Tested on Dataflow runner with following error injection procedure:

After pipeline started consuming data, ssh into worker machine and kill java process

logs seen:

DEFAULT 2026-02-23T21:24:32.670187985Z --------------{"metadata":{"connector":"postgresql","version":"3.1.3.Final","name":"beam-debezium-connector","database":"test","schema":"public","table":"test_table"},"before":null,"after":{"fields":{"name":"abc","id":1}}}------------
DEFAULT 2026-02-23T21:24:32.671735590Z --------------{"metadata":{"connector":"postgresql","version":"3.1.3.Final","name":"beam-debezium-connector","database":"test","schema":"inventory","table":"customers"},"before":null,"after":{"fields":{"last_name":"Thomas","id":1001,"first_name":"Sally","email":"sally.thomas@acme.com"}}}------------
...
WARNING 2026-02-23T21:25:33.613167Z [severity: WARNING] Completing Logging stream for SDK harness sdk-0-0. This indicates sdk issues unless the job is being terminated.
CRITICAL 2026-02-23T21:25:33.613200197Z [severity: CRITICAL] Java exited: exit status 143
CRITICAL 2026-02-23T21:25:33.613255955Z [severity: CRITICAL] Completing background processes before exiting...
...
DEFAULT 2026-02-23T21:25:51.338688576Z SDK Fn Harness started

confirmed worker does not read from the beginning again. Then add some rows, see it is consumed:

DEFAULT 2026-02-23T21:26:48.739887663Z --------------{"metadata":{"connector":"postgresql","version":"3.1.3.Final","name":"beam-debezium-connector","database":"test","schema":"public","table":"test_table"},"before":null,"after":{"fields":{"name":"def","id":2}}}------------
INFO 2026-02-23T21:26:49.405Z Debezium ServiceRegistry stopped.
...
DEFAULT 2026-02-23T21:27:44.665167369Z --------------{"metadata":{"connector":"postgresql","version":"3.1.3.Final","name":"beam-debezium-connector","database":"test","schema":"public","table":"test_table"},"before":null,"after":{"fields":{"name":"ghi","id":3}}}------------
INFO 2026-02-23T21:27:45.298Z Debezium ServiceRegistry stopped.

While this has fixed Dataflow runner worker crash / restart, I think there is another FR in #28248 to support start consumer from a specific offset. Keep the issue open for now.


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.

@gemini-code-assist
Copy link
Contributor

Summary of Changes

Hello @Abacn, 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 significantly enhances the robustness and configurability of DebeziumIO within Apache Beam. The primary goal was to prevent pipeline crash loops during worker restarts and ensure seamless data consumption. This was achieved by correctly initializing internal state, refining the record polling mechanism, and exposing a new configuration option for polling timeouts. These changes collectively improve the stability and operational flexibility of Debezium-based data pipelines.

Highlights

  • Worker Restart Resilience: Addressed a crash loop issue that occurred when a worker restarted within the same pipeline, ensuring DebeziumIO can resume correctly without starting from the beginning.
  • NPE Fix: Resolved a NullPointerException by moving the startTime recording for the KafkaSourceConsumerFn into the @Setup method, ensuring it's initialized correctly upon worker restart.
  • Poll Loop Exit Logic: Fixed an issue where the DebeziumIO poll loop would not exit correctly even when records were present, improving the termination behavior of the KafkaSourceConsumerFn.
  • Configurable Polling Timeout: Introduced a configurable pollingTimeout for the KafkaSourceConsumerFn, allowing users to optimize for latency or improve record fetching by adjusting the consumer polling request timeout.
  • Refactored KafkaSourceConsumerFn Initialization: Simplified the constructors for KafkaSourceConsumerFn by passing the DebeziumIO.Read specification object, centralizing configuration management.
  • Simplified Offset Tracking: Modified the OffsetTracker to solely restrict processing based on the maximum number of records, removing the time-based restriction from its tryClaim logic.

🧠 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/debezium/src/README.md
    • Updated the description of DebeziumIO initialization to reflect changes in how time-based restrictions are set.
  • sdks/java/io/debezium/src/main/java/org/apache/beam/io/debezium/DebeziumIO.java
    • Added getPollingTimeout() abstract method to Read class.
    • Added setPollingTimeout() abstract method to Builder class.
    • Introduced withPollingTimeout() public method to configure the consumer polling timeout.
    • Refactored KafkaSourceConsumerFn instantiation to use the DebeziumIO.Read specification object instead of individual parameters.
  • sdks/java/io/debezium/src/main/java/org/apache/beam/io/debezium/KafkaSourceConsumerFn.java
    • Imported java.time.Duration, MoreObjects, and Stopwatch for new timing logic.
    • Removed org.joda.time.Duration import.
    • Defined DEFAULT_POLLING_TIMEOUT constant.
    • Refactored KafkaSourceConsumerFn constructors to accept a DebeziumIO.Read<T> spec object.
    • Moved startTime initialization from getInitialRestriction to a new @Setup method to prevent NPE on deserialization.
    • Modified the process method to incorporate pollingTimeOut and Stopwatch for managing poll loop duration.
    • Adjusted the ProcessContinuation logic to use remainingTimeout for resume delay, ensuring the poll loop respects the configured timeout.
    • Simplified OffsetTracker.tryClaim to only consider maxRecords for restriction, removing time-based checks.
  • sdks/java/io/debezium/src/test/java/org/apache/beam/io/debezium/KafkaSourceConsumerFnTest.java
    • Added a static helper method getSpec to create DebeziumIO.Read instances for tests.
    • Updated existing KafkaSourceConsumerFn instantiations in tests to use the new constructor with DebeziumIO.Read spec.
    • Added testMaxTimeToRun to verify the time-based stopping mechanism.
    • Modified CounterSourceConnectorConfig to include a default value for 'to' and add a 'sleep' parameter.
    • Updated CounterSourceConnector's taskConfigs and start methods to handle the new 'sleep' parameter and default 'to' value.
    • Adjusted CounterTask's poll method to support indefinite running (to == -1) and introduce a sleep duration.
  • sdks/java/io/debezium/src/test/java/org/apache/beam/io/debezium/OffsetTrackerTest.java
    • Updated KafkaSourceConsumerFn instantiation to use the new constructor with DebeziumIO.Read spec.
    • Removed the testRestrictByAmountOfTime test method, as time-based restriction is now handled at the ProcessContinuation level rather than by the OffsetTracker.
Activity
  • The author tested the changes on a Dataflow runner by injecting errors (killing the Java process on a worker machine).
  • Confirmed that after a worker restart, the pipeline did not re-read data from the beginning.
  • Verified that new rows added after the restart were successfully consumed by the pipeline.
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.

@Abacn Abacn marked this pull request as ready for review February 24, 2026 02:14
@github-actions
Copy link
Contributor

Assigning reviewers:

R: @ahmedabu98 for label java.

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).

* Move startTime recording into setup to fix NPE in restarted worker

* Fix DebeziumIO poll loop not exiting when record list isn't empty

* Make pollTimeout configurable
}
pollTimer.reset().start();
task.commit();
records = task.poll();

Choose a reason for hiding this comment

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

This task.poll() could take much longer than pollingTimeOut, if we have a way to set timeout for task.poll() operation which respects the total pollingTimeOut, that would be better!

Copy link

@jiufeng-liu-ck jiufeng-liu-ck left a comment

Choose a reason for hiding this comment

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

Looks good to me!

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