jackylee-ch opened a new issue, #6995: URL: https://github.com/apache/incubator-gluten/issues/6995
### Backend VL (Velox) ### Bug description Currently, we monitor TaskEnd/StageEnd events to update and clean up FilePartition and Executor data for duplicate reading. However, in actual execution, we found that the current monitoring logic is slow, and it would cause spark event accumulation. When the events eventually exceed the capacity defined by Spark and they will be discarded. As a result, a large amount of information remains in rddPartitionInfoMap and stageInfoMap and cannot be discarded, which leads to frequent FGC in driver and finnaly OOM. Bellow is the log for Event discard.  Bellow is the memory analyze for our online driver. The driver memory is 2.2G and the memory used by SoftAffinityManager is 1.5G. <img width="687" alt="image" src="https://github.com/user-attachments/assets/a1f364d7-e49a-4802-a83b-840f650879b8";> ### Spark version Spark-3.5.x ### Spark configurations _No response_ ### System information _No response_ ### Relevant logs _No response_ -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
