andygrove commented on issue #2452: URL: https://github.com/apache/datafusion-comet/issues/2452#issuecomment-3369204525
I learned more about this, and the issue may be in Comet rather than DataFusion. Because we share one pool between concurrent tasks, when one SMJ instance fails to allocate memory and returns some memory to the pool before attempting to spill, another task can grab that memory before the spilling code can allocate it. I started testing using the `fair_unified` pool instance, but ran into some issues with that, so I'll debug that next. -- 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]
