dwsmith1983 commented on PR #5613: URL: https://github.com/apache/datafusion-comet/pull/5613#issuecomment-5998784142
On the exact-fit wait: it is real, and we are keeping the anchor for it. The anchor is what stops a release from removing the task's entry under a parked acquire, which fails that acquire, and the extra wait only happens when the request would take exactly the memory Spark has free and ends at the next release. 518dc62ec pins it: below the minimum share the request parks until another task releases memory and then gets its full grant, and at the share `try_grow` is refused one byte short while `grow` carries one byte of overcommit. The memory guide and the description now describe both cases. -- 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]
