The GitHub Actions job "Nightly Builds" on pekko.git/main has failed.
Run started by GitHub user He-Pin (triggered by He-Pin).

Head commit for run:
fc3561b1d9e0c54e2f4a815fe79dcfad546c17f2 / He-Pin(kerr) <[email protected]>
test: avoid busy-spin in stream stress tests (#2996)

* test: reduce stream stress test timing sensitivity

Motivation:
Recent JDK 25 nightly builds repeatedly time out in BoundedSourceQueueSpec and 
FlowMapAsync/FlowMapAsyncUnordered parallelism tests.

Modification:
Give the BoundedSourceQueue completion assertion a dilated timeout, reduce the 
burst wakeup stress loop to a still-repetitive 10000 rounds, and reduce the 
mapAsync parallelism stress input size while preserving the parallelism limit 
assertion. Also use the same 12 second timeout for the ordered mapAsync test as 
the unordered test gets from the JDK 25 timefactor.

Result:
The tests continue to verify queue wakeups and mapAsync parallelism limits 
without depending on excessive runner throughput under JDK 25 nightly load.

Tests:
- scalafmt --mode diff-ref=origin/main
- scalafmt --list --mode diff-ref=origin/main
- git diff --check
- sbt with JDK 25 nightly-style virtualized test-stream-dispatcher flags: 
stream-tests / Test / testOnly 
org.apache.pekko.stream.scaladsl.BoundedSourceQueueSpec 
org.apache.pekko.stream.scaladsl.FlowMapAsyncSpec 
org.apache.pekko.stream.scaladsl.FlowMapAsyncUnorderedSpec

References:
None - nightly-builds.yml failure analysis

* test: avoid busy-spin in mapAsync stress tests

Motivation:
Recent nightly stream stress failures timed out under JDK 25 contention. The 
previous PR revision reduced the stress sizes, but that weakened the tests 
instead of addressing the root timing sensitivity.

Modification:
Keep the existing 10K mapAsync stress inputs and 100K BoundedSourceQueue burst 
loop. Replace the mapAsync stress-test busy-spin delay loops with 
LockSupport.parkNanos so the tests preserve asynchronous completion pressure 
without burning carrier CPU. Keep the BoundedSourceQueue completion assertion 
dilated because that assertion is explicitly timefactor-sensitive.

Result:
The tests retain their original stress coverage while avoiding artificial CPU 
starvation in nightly runners.

Tests:
- JDK 25 nightly-style virtualized flags: stream-tests / Test / testOnly 
org.apache.pekko.stream.scaladsl.FlowMapAsyncSpec 
org.apache.pekko.stream.scaladsl.FlowMapAsyncUnorderedSpec 
org.apache.pekko.stream.scaladsl.FlowMapAsyncPartitionedSpec
- JDK 25 nightly-style virtualized flags: stream-tests / Test / testOnly 
org.apache.pekko.stream.scaladsl.BoundedSourceQueueSpec
- scalafmt --mode diff-ref=origin/main
- scalafmt --list --mode diff-ref=origin/main
- git diff --check

References:
None - nightly-builds.yml failure analysis.

Report URL: https://github.com/apache/pekko/actions/runs/26509448278

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to