damccorm commented on code in PR #41280:
URL: https://github.com/apache/beam/pull/41280#discussion_r4564542786
##########
sdks/java/io/iceberg/src/test/java/org/apache/beam/sdk/io/iceberg/cdc/sink/DeltaLakeToIcebergCdcIT.java:
##########
@@ -141,7 +142,10 @@ public class DeltaLakeToIcebergCdcIT {
@Rule public final TestPipeline p = TestPipeline.create();
@Rule public final TestName testName = new TestName();
- @Rule public final Timeout globalTimeout = Timeout.seconds(10 * 60);
+
+ @Rule
+ public final Timeout globalTimeout =
+ Timeout.seconds(OPTIONS.getRunner().equals(DirectRunner.class) ? 300 :
20 * 60);
Review Comment:
Nothing changed in the test or pipeline code recently — in 19 of the last 20
scheduled `IcebergIO Integration Tests on Dataflow` runs on `master`, each test
method in `DeltaLakeToIcebergCdcIT` completed in **~5m 00s–5m 25s** (well
within the 10-minute JUnit timeout).
In the single failing run
([43366381845](https://github.com/apache/beam/actions/runs/43366381845)),
Dataflow job `2026-11-12_21_36_57-14384524004107536003`
(`changeFeedAppliesToIcebergTableWithChangeTypeColumn`, which runs on Runner
v2) experienced a one-off slow Runner v2 worker container pull + 644-JAR
staging boot on the GCE VM (`Waiting for 1 of 1 SDK Harnesses to register` took
~6.5 minutes after VM startup instead of the usual ~2 minutes). Once the SDK
harness registered at `05:45:15Z`, the pipeline itself executed in ~2 minutes
and succeeded (`JOB_STATE_DONE` at `05:49:31Z`, 12m 33s total), ~3 minutes
after JUnit's 10-minute `Timeout` terminated the test client at `05:46:33Z`.
For context, when `DeltaLakeToIcebergCdcIT` and `IcebergCdcWriteBaseIT` were
added together in #40230, `IcebergCdcWriteBaseIT` (matching
`IcebergCatalogBaseIT`) used
`Timeout.seconds(OPTIONS.getRunner().equals(DirectRunner.class) ? 300 : 20 *
60)`, whereas `DeltaLakeToIcebergCdcIT` hardcoded `Timeout.seconds(10 * 60)`.
Since this only occurred once in the last 7 days (1/20 runs), let me know if
you'd like to keep this alignment with `IcebergCdcWriteBaseIT` or close this PR.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
