sundapeng opened a new pull request, #9019: URL: https://github.com/apache/paimon/pull/9019
### Purpose Listing a directory through `JindoFileIO` is slow when it contains many subdirectories, and partition discovery on format tables hits this path on every plan. The time is spent resolving timestamps for directory entries, controlled by `fs.oss.show-dir-timestamp` which Jindo enables by default. Measured on Jindo 6.10.2 in the same JVM, three rounds with rotated execution order: | `fs.oss.show-dir-timestamp` | `listStatus` of 1,095 subdirectories | |---|---:| | unset (Jindo defaults to true) | 10,760–11,953 ms | | `false` | 190–240 ms | Both arms returned the same 1,095 entries with the same path/type/length checksum. Paimon does not read directory timestamps from listings, so this PR defaults the option to `false` in `JindoFileIO`. An explicit user setting is kept unchanged. ### Tests `JindoFileIOConfigurationTest` verifies the default is applied and an explicit `true` setting wins. `mvn -pl paimon-filesystems/paimon-jindo test` passes. -- 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]
