balodesecurity opened a new pull request, #8322: URL: https://github.com/apache/hadoop/pull/8322
## Summary `DistributedFileSystem.getFileChecksum()` on an empty (zero-byte) file always returned an `MD5MD5CRC32GzipFileChecksum` value, even when the client was configured with `dfs.checksum.combine.mode=COMPOSITE_CRC`. This was misleading and inconsistent with what non-empty files return in the same mode. ### Root Cause In `FileChecksumHelper.FileChecksumComputer.compute()`, the empty-file fast path unconditionally constructed an `MD5MD5CRC32GzipFileChecksum` without checking `combineMode`. ### Fix Check `combineMode` in the empty-file path: - **`COMPOSITE_CRC`**: return `CompositeCrcFileChecksum(crc=0, configuredChecksumType, configuredBytesPerCrc)` using the client's configured `ChecksumOpt`. - **`MD5MD5CRC`**: retain the existing backward-compatible magic value (`MD5MD5CRC32GzipFileChecksum(0, 0, md5OfZeros)`) to avoid breaking existing tools. ### Changes - `FileChecksumHelper.java`: split the empty-file branch on `combineMode`. - `TestGetFileChecksum.java`: add `testEmptyFileChecksumType` verifying both modes return the expected `FileChecksum` subclass for a zero-byte file. ## Test plan - [ ] `TestGetFileChecksum#testEmptyFileChecksumType` passes locally ✅ - [ ] Full module build passes (`mvn package ... -DskipTests`) ✅ -- 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]
