IgorBerman commented on issue #3703: URL: https://github.com/apache/iceberg/issues/3703#issuecomment-2747907962
It seems that after all it's related to broadcast that happens unconditionally in [contentFileDS](https://github.com/apache/iceberg/blob/1.8.x/spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/BaseSparkAction.java#L153) method the broadcast variable contains SerializableTableWithSize which is accumulated over time if same action is run in the same driver there is no code that unpersists this broadcast. I can't see a reason why setting spark.sql.autoBroadcastJoinThreshold=-1 should help here, since this is core spark api and not sql api -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
