[Failing Test]: SqsIOWriteBatchesTest is flaky (Mockito ArgumentsAreDifferent in timeout-related tests)

[tkaymak]

What happened?

org.apache.beam.sdk.io.aws2.sqs.SqsIOWriteBatchesTest is flaky in the beam_PreCommit_Java_Amazon-Web-Services2_IO_Direct workflow. Different test methods of the same class fail intermittently with Mockito verification errors (org.mockito.exceptions.verification.junit.ArgumentsAreDifferent), which points to a timing-sensitive interaction with the batching/timeout logic under test:

• testWriteBatchesWithTimeout — https://github.com/apache/beam/actions/runs/27417091199/job/81032556571 (2026-06-12, PR [mqtt] Add portable MqttIO Read/Write transforms for batch and streaming (revives #32385) #38493: 369 tests completed, 1 failed)
• testWriteBatchesToDynamicWithTimeout and testWriteBatchesToDynamicWithStrictTimeout — observed earlier on PR [runners-spark] Add Spark 4 runner #38255 runs (2026-04-28, e.g. commits e8e107ce, b344c3ca, see [runners-spark] Add Spark 4 runner #38255 (comment))

The job is green on most master runs (last several scheduled runs passed), so this is an intermittent flake rather than a consistent breakage, but it has been recurring for months and regularly turns unrelated PRs red.

Issue Priority

Priority: 2 (default / most failing tests should be filed as P2)

Issue Components

• Component: Python SDK
• Component: Java SDK

[Kriti-dev07]

Hi,
I'd like to investigate this test issue.
Before I start, do we currently have a reliable way to reproduce the failures locally (e.g. specific Gradle/Maven commands, repeated test runs, or known stress settings)or is reproducing the flake itself part of the investigation?