[SDK Sentinel] Stabilize sticky cache metric assertions

[sdk-sentinel-bot]

Caution

This PR contains untrusted AI-generated code. Do not approve or run CI until a maintainer has reviewed the diff. SDK Sentinel verified that GitHub Actions remained approval-gated with zero executable jobs when this PR was opened.

Summary

• CI evidence: failed CI job · workflow run · last known pass · first known failure

Stabilize a test-only assertion after the Linux Java 23 Edge suite intermittently observed an active-thread gauge of zero. The test passed in the [last known good job](https://github.com/temporalio/sdk-java/actions/runs/34508568447/job/102976712372) and failed in the [first known bad job](https://github.com/temporalio/sdk-java/actions/runs/34634340353/job/103378476256). No public API, breaking, or server changes.

Root cause

The test slept for a fixed 600 ms before sampling asynchronously published gauges once. Runner scheduling could leave the reporter's latest value at zero when sampled.

Fix

Use the existing bounded eventual-assertion helper around both original active-state gauge expectations. This preserves the asserted values while removing dependence on one wall-clock sample.

Validation

Five patched focused stress attempts and formatting passed. Six unchanged attempts did not naturally reproduce the low-rate failure. The unchanged baseline validator passed, but candidate validation stopped at submodule setup because this investigation sandbox makes git submodule metadata read-only. The Linux Java 23 Edge job and canonical validator remain required.

Validation status: validation-incomplete

• Flake confidence: high
• Fix confidence: medium
• Value: medium
• Patch scope: standard — 1 changed file; 1.6 KiB
• Local reproduction: not-reproduced — The unchanged focused test passed six local attempts; the captured same-dimension CI pass/failure establishes the low-rate flake.
• Regression coverage: pass-after-only — The existing focused test is the regression coverage and passed five post-change stress attempts.
• Unchanged baseline (Gradle formatting and test-source compilation) — passed
• Independent candidate (Gradle formatting and test-source compilation) — did not pass (failure)

Investigator-run checks

• GRADLE_USER_HOME=.ci-flake-runtime/tmp/gradle-agent ./gradlew --offline --no-daemon --init-script .ci-flake-runtime/tmp/redirect-build.gradle :temporal-sdk:test --tests "io.temporal.workflow.MetricsTest.testStickyCacheSize" -x :temporal-serviceclient:initSubmodules -x :temporal-serviceclient:updateSubmodules -x :temporal-sdk:compileJava17Java -x :temporal-sdk:compileJava21Java — passed; 5 attempt(s). All five patched focused stress attempts passed; six unchanged attempts had also passed.
• .ci-flake-runtime/input/validate.sh — failed; 1 attempt(s). Formatting checks reached the submodule tasks, then validation was blocked because the sandbox made .git/modules submodule configuration read-only. The trusted unchanged baseline had passed.
• GRADLE_USER_HOME=.ci-flake-runtime/tmp/gradle-agent ./gradlew --offline --no-daemon --init-script .ci-flake-runtime/tmp/redirect-build.gradle spotlessApply -x :temporal-serviceclient:initSubmodules -x :temporal-serviceclient:updateSubmodules — passed; 1 attempt(s). Repository formatting completed successfully.

Required target CI

• Continuous Integration / Unit test with in-memory test service [Edge] on Linux with Java 23
• Canonical spotlessCheck and testClasses validation in an environment with writable git submodule metadata

Residual risks

• The natural failure was not reproduced locally, so CI must confirm the timing fix under Edge-suite load.
• The fixed repository validator could not complete because of the investigation sandbox's read-only submodule metadata.

Automation provenance

• Investigator workflow
• Base SHA: dd28759b03134c0a9d43ffe960061d07f1759c3d

If this finding should not be fixed, apply the sdk-sentinel:false-positive label and close the PR. Sentinel will suppress the finding until its affected source changes.

SDK Sentinel recurrence history

• Unique failed CI run attempts: 1
• First occurrence: 2026-09-11 18:46 UTC
• Latest occurrence: 2026-09-11 18:46 UTC
• Recent occurrences:
  • 2026-09-11 18:46 UTC — failed CI job