nsivabalan commented on issue #16987: URL: https://github.com/apache/hudi/issues/16987#issuecomment-4800721452
Triage note: this appears fixed on master. - Root-cause fix: [#13882](https://github.com/apache/hudi/pull/13882) (commit `4d95b2c2`) — `HoodieRowParquetWriteSupport` now derives timestamp precision from the writer Avro/Hudi schema. For fields with `logicalType: timestamp-millis`, it emits `INT64` annotated `TIMESTAMP(MILLIS, UTC)` and scales the long via `DateTimeUtils.microsToMillis(...)` instead of falling back to Spark's `spark.sql.parquet.outputTimestampType` default (micros). See `HoodieRowParquetWriteSupport.makeWriter` (~L444) and `convertField` (~L719). - Repair for already-affected tables: [#14161](https://github.com/apache/hudi/pull/14161) (commit `cab988cc`) ships `logical-repair` tooling for tables written by 0.15.x that have the corrupted millis-as-micros encoding. If you can still reproduce on a recent master/1.1.0 build with the script in #13233, please reopen with the exact build hash — otherwise this can likely be closed once 1.1.0 ships. -- 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]
