dnamaz opened a new issue, #37035:
URL: https://github.com/apache/beam/issues/37035
### What happened?
# GitHub Issue: Flink Runner - No translator known for
PrimitiveUnboundedRead after SDF fallback
## Title
`[Flink Runner] No translator known for PrimitiveUnboundedRead after
SDF-to-primitive-read conversion`
## Labels
- `flink-runner`
- `bug`
- `P2`
## Body
### What happened?
When using the Flink classic runner (non-portable) with unbounded source
connectors like `KinesisIO.read()`, the pipeline fails with:
```
java.lang.IllegalStateException: No translator known for
org.apache.beam.sdk.util.construction.SplittableParDo$PrimitiveUnboundedRead
```
### Environment
- **Apache Beam version**: 2.60.0+ (tested with 2.69.0)
- **Flink version**: 1.18, 1.19, 1.20
- **Runner**: Flink classic runner (non-portable, without `beam_fn_api`
experiment)
- **Platform**: AWS Managed Apache Flink
### Steps to reproduce
1. Create a Beam pipeline using `KinesisIO.read()` (or any
`UnboundedSource`-based IO)
2. Run with FlinkRunner on AWS Managed Flink (or any Flink environment
without portable runner)
3. Pipeline fails during translation
```java
Pipeline p = Pipeline.create(options);
p.apply(KinesisIO.read()
.withStreamName("my-stream")
.withInitialPositionInStream(InitialPositionInStream.TRIM_HORIZON))
.apply(ParDo.of(new ProcessFn()));
p.run();
```
### Root Cause Analysis
The Flink classic runner calls
`SplittableParDo.convertReadBasedSplittableDoFnsToPrimitiveReadsIfNecessary()`
in `FlinkRunner.run()` when NOT using the `beam_fn_api` experiment:
```java
// FlinkRunner.java line 76-79
// TODO(https://github.com/apache/beam/issues/20530): Use SDF read as
default when we address performance issue.
if (!ExperimentalOptions.hasExperiment(pipeline.getOptions(),
"beam_fn_api")) {
SplittableParDo.convertReadBasedSplittableDoFnsToPrimitiveReadsIfNecessary(pipeline);
}
```
This conversion transforms:
- `Read.Unbounded` → `SplittableParDo.PrimitiveUnboundedRead`
- `Read.Bounded` → `SplittableParDo.PrimitiveBoundedRead`
However, `FlinkStreamingTransformTranslators` had no registered translators
for these `PrimitiveUnboundedRead` and `PrimitiveBoundedRead` transforms,
causing the "No translator known" error.
### Related Issues
- Related to #20530 (Use SDF read as default when performance issue is
addressed)
### Expected Behavior
The pipeline should successfully translate and execute on the Flink classic
runner.
### Actual Behavior
Pipeline fails during translation with `IllegalStateException: No translator
known for PrimitiveUnboundedRead`.
### Proposed Fix
Add explicit translators for `PrimitiveUnboundedRead` and
`PrimitiveBoundedRead` in `FlinkStreamingTransformTranslators.java` that
delegate to the existing `FlinkUnboundedSource` and `FlinkBoundedSource`
implementations.
### Issue Priority
Priority: 2 (default / most bugs should be filed as P2)
### Issue Components
- [ ] Component: Python SDK
- [ ] Component: Java SDK
- [ ] Component: Go SDK
- [ ] Component: Typescript SDK
- [ ] Component: IO connector
- [ ] Component: Beam YAML
- [ ] Component: Beam examples
- [ ] Component: Beam playground
- [ ] Component: Beam katas
- [ ] Component: Website
- [ ] Component: Infrastructure
- [ ] Component: Spark Runner
- [x] Component: Flink Runner
- [ ] Component: Samza Runner
- [ ] Component: Twister2 Runner
- [ ] Component: Hazelcast Jet Runner
- [ ] Component: Google Cloud Dataflow Runner
--
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]
