garyli1019 commented on pull request #3093: URL: https://github.com/apache/hudi/pull/3093#issuecomment-866035898
> Generally speaking, I am bit skeptical on overall use of speculative execution. Most of the policies are pretty naive. Anyways, @garyli1019 Would love to get a JIRA with some steps to repro if possible. The writer is already resistant to stage retries, so it should no cause such dirty/uncommitted files lying around. Do you have the state of marker files when this happened? There is very little chance that the executor being killed before create the marker file and after drop the parquet. Will follow up with this issue once I have the fix. https://issues.apache.org/jira/browse/HUDI-2056 -- 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. For queries about this service, please contact Infrastructure at: [email protected]
