Github user rvesse commented on the issue: https://github.com/apache/spark/pull/21669 @ifilonenko I think the issue with the `UnixUsername` might possibly be avoided by exporting `HADOOP_USER_NAME` as an environment variable in the pod spec set to the same value as `SPARK_USER`. This should cause the Hadoop UGI machinery to simply take the provided username rather than attempting to obtain it from the container which fails because the user does not exist as a Unix user in `/etc/passwd`
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org