marc-pydantic opened a new issue, #22363:
URL: https://github.com/apache/datafusion/issues/22363
### Describe the bug
The partitioned_by_file_group field in `FileScanConfig` is not included in
the `FileScanExecConf` protobuf message. After deserialization, the field
defaults to false, enabling work stealing? even when the original plan had it
disabled.
This does seem to manifest as duplicated rows in results when using
`datafusion-distributed` latest `main`.
### To Reproduce
```rust
let scan_config = FileScanConfigBuilder::new(url, file_source)
.with_partitioned_by_file_group(true)
.build();
assert!(scan_config.partitioned_by_file_group); // true
let bytes =
physical_plan_to_bytes(DataSourceExec::from_data_source(scan_config))?;
let result = physical_plan_from_bytes(bytes)?;
let config = result.downcast_ref::<DataSourceExec>()
.data_source().downcast_ref::<FileScanConfig>();
assert!(config.partitioned_by_file_group); // FAILS - now false
```
### Expected behavior
`partitioned_by_file_group` should survive protobuf roundtrip
### Additional context
Field was introduced in #21351 and #21342 but not added to the protobuf
schema. With 3 workers, aggregations return 3x expected values (e.g., count =
198 instead of 66).
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
