Martijn Visser created FLINK-41229:
--------------------------------------
Summary: StreamNetworkThroughputBenchmark hangs instead of failing
when the receiver thread fails
Key: FLINK-41229
URL: https://issues.apache.org/jira/browse/FLINK-41229
Project: Flink
Issue Type: Bug
Components: Benchmarks
Affects Versions: 2.4.0
Reporter: Martijn Visser
Assignee: Martijn Visser
When {{ReceiverThread}} hits an exception, {{go()}} prints it and returns
without completing {{recordsProcessed}}.
{{StreamNetworkThroughputBenchmark.executeBenchmark(records)}} waits on that
future with {{Long.MAX_VALUE}}, so the benchmark hangs until something kills
it. In flink-benchmarks, six CI runs between 2026-09-22 and 2026-10-26 sat
there until GitHub killed them after 6h (FLINK-41226).
{code}
"main" ... at
java.util.concurrent.CompletableFuture.get(CompletableFuture.java:2095)
at
org.apache.flink.streaming.runtime.io.benchmark.StreamNetworkThroughputBenchmark.executeBenchmark(StreamNetworkThroughputBenchmark.java:59)
at
org.apache.flink.streaming.runtime.io.benchmark.StreamNetworkThroughputBenchmark.executeBenchmark(StreamNetworkThroughputBenchmark.java:40)
at
org.apache.flink.benchmark.StreamNetworkThroughputBenchmarkExecutor.networkThroughput(StreamNetworkThroughputBenchmarkExecutor.java:64)
{code}
The receiver should complete {{recordsProcessed}} exceptionally, so the
benchmark fails with the actual cause.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
