You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Spark treats all ORC files to have file statistics which does not follow ORC specifications
Spark-3.3.0 is expected to be released before fixing this issue. So, we need to alert the users that for Spark-330 there might be a runtime exception reading ORC file generated by the GPU.
The text was updated successfully, but these errors were encountered:
Report needed documentation
Based on the discussion in SPARK-34960: Aggregate (Min/Max/Count) push down for ORC, we need to document that the flag
spark.sql.orc.aggregatePushdown
should be disabled while reading ORC files created by the GPU.Why do we need this
file statistics
which does not follow ORC specificationsThe text was updated successfully, but these errors were encountered: