flyrain commented on issue #214: URL: https://github.com/apache/polaris/issues/214#issuecomment-2316234502
This is an Iceberg issue instead of a Polaris one. To summarize, DML commands and procedures usually use `FileIO` object provided by the catalog for read and write files. However, the procedure `RemoveOrphanFile` uses the [Spark configuration](https://github.com/apache/iceberg/blob/79620e198009fa243c278c66fd442d107b46206a/spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java#L356-L356) to get the FileSystem object for listing, which is a Hadoop s3a File System. It couldn't recognize the `s3://`. Solutions would be 1. Using catalog `FileIO` instead of the File System from Spark config. `ResolvingFileIO` is the default one used by REST catalog, which delegates to `S3FileIO` in this case, [it supports `listPrefix`](https://github.com/apache/iceberg/blob/79620e198009fa243c278c66fd442d107b46206a/aws/src/main/java/org/apache/iceberg/aws/s3/S3FileIO.java#L299-L299). 2. Using aws s3 client instead of Hadoop s3a client in Spark, I guess this only needs a config change, I'm not familiar with that though. Recommend to check with the Iceberg community. -- 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]
