Peter Toth created SPARK-60381:
----------------------------------
Summary: Account the late-materialization survivor buffer to the
task's execution memory
Key: SPARK-60381
URL: https://issues.apache.org/jira/browse/SPARK-60381
Project: Spark
Issue Type: Improvement
Components: SQL
Affects Versions: 5.0.0
Reporter: Peter Toth
Account the survivor buffer of the late-materialization Parquet reader
(SPARK-59620) to the task's execution memory through a MemoryConsumer. A denied
reservation gives splicing up, as passing
spark.sql.parquet.storageFilterPushdown.maxSplicedRowGroupBytes does, and a
spill frees the survivor key values not yet emitted, which the reader then
reads again.
Suggested in https://github.com/apache/spark/pull/58895#issuecomment-6064730845.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
