macdoor opened a new pull request, #27965: URL: https://github.com/apache/flink/pull/27965
## Purpose Fix `ConnectionClosedException: Premature end of Content-Length delimited message body` when reading large Parquet files from S3-compatible storage (e.g. MinIO) via `flink-s3-fs-native`, especially with `seek()` / ranged GETs. ## Root cause Closing a partially consumed `GetObject` response via `BufferedInputStream.close()` lets Apache HttpClient drain the remaining body for connection reuse. If the connection ends early, draining throws. AWS SDK v2 recommends `ResponseInputStream.abort()` when the rest of the body will not be read. ## Change - On reopen and on `close()` when `position < contentLength`, call `abort()` on the `ResponseInputStream` instead of relying on drain-on-close. ## Testing - Local `mvn` compile was not run in this environment (Flink 2.4-SNAPSHOT reactor deps not resolvable against the configured Maven mirror). Relying on Azure Pipelines for `flink-s3-fs-native`. JIRA: https://issues.apache.org/jira/browse/FLINK-39484 -- 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]
