KAFKA-17479 Fail the whole pipeline if junit step times out [4/n] (#17121)

Fixes an issue where the CI workflow could appear to be successful in the event of a timeout and no failing tests. Instead of using Github Action's timeout, this patch makes use of the linux `timeout` command. This lets us capture the exit code and handle timeouts separately from a failed execution.

Reviewers: Chia-Ping Tsai <chia7712@gmail.com>

.github/scripts/junit.py

@@ -37,10 +37,14 @@ FLAKY = "FLAKY ⚠️ "
 SKIPPED = "SKIPPED 🙈"
 
 
-def get_env(key: str) -> str:
+def get_env(key: str, fn = str) -> Optional:
     value = os.getenv(key)
-    logger.debug(f"Read env {key}: {value}")
-    return value
+    if value is None:
+        logger.debug(f"Could not find env {key}")
+        return None
+    else:
+        logger.debug(f"Read env {key}: {value}")
+        return fn(value)
 
 
 @dataclasses.dataclass
@@ -248,12 +252,23 @@ if __name__ == "__main__":
             print(f"| {row_joined} |")
         print("\n</details>")
 
-    logger.debug(summary)
-    if total_failures > 0:
-        logger.debug(f"Failing this step due to {total_failures} test failures")
-        exit(1)
-    elif total_errors > 0:
-        logger.debug(f"Failing this step due to {total_errors} test errors")
+    # Print special message if there was a timeout
+    exit_code = get_env("GRADLE_EXIT_CODE", int)
+    if exit_code == 124:
+        logger.debug(f"Gradle command timed out. These are partial results!")
+        logger.debug(summary)
+        logger.debug("Failing this step because the tests timed out.")
         exit(1)
+    elif exit_code in (0, 1):
+        logger.debug(summary)
+        if total_failures > 0:
+            logger.debug(f"Failing this step due to {total_failures} test failures")
+            exit(1)
+        elif total_errors > 0:
+            logger.debug(f"Failing this step due to {total_errors} test errors")
+            exit(1)
+        else:
+            exit(0)
     else:
-        exit(0)
+        logger.debug(f"Gradle had unexpected exit code {exit_code}. Failing this step")
+        exit(1)

.github/workflows/build.yml

@@ -107,17 +107,18 @@ jobs:
         # --scan:         Attempt to publish build scans in PRs. This will only work on PRs from apache/kafka, not public forks.
         # --continue:     Keep running even if a test fails
         # -PcommitId      Prevent the Git SHA being written into the jar files (which breaks caching)
-        timeout-minutes: 180  # 3 hours
-        continue-on-error: true
+        id: junit-test
         run: |
-          ./gradlew --build-cache --scan --continue \
+          set +e
+          timeout 180m ./gradlew --build-cache --scan --continue \
           -PtestLoggingEvents=started,passed,skipped,failed \
           -PmaxParallelForks=2 \
           -PmaxTestRetries=1 -PmaxTestRetryFailures=10 \
           -PcommitId=xxxxxxxxxxxxxxxx \
           test
+          exitcode="$?"
+          echo "exitcode=$exitcode" >> $GITHUB_OUTPUT
       - name: Archive JUnit reports
-        if: always()
         uses: actions/upload-artifact@v4
         id: junit-upload-artifact
         with:
@@ -126,8 +127,8 @@ jobs:
             **/build/reports/tests/test/*
           if-no-files-found: ignore
       - name: Parse JUnit tests
-        if: always()
         run: python .github/scripts/junit.py >> $GITHUB_STEP_SUMMARY
         env:
           GITHUB_WORKSPACE: ${{ github.workspace }}
           REPORT_URL: ${{ steps.junit-upload-artifact.outputs.artifact-url }}
+          GRADLE_EXIT_CODE: ${{ steps.junit-test.outputs.exitcode }}