Hi, Has anyone else also experienced https://issues.apache.org/jira/browse/SPARK-2604?
It is an edge case scenario of mis configuration, where the executor memory asked is same as the maximum allowed memory by yarn. In such situation, application stays in hang state, and the reason is not logged in verbose manner to be debugged easily. As per the fix, it gets detected and corresponding reasons are logged before failing the application. I will prefer the fix to be in open source code version, please share your thoughts. Thanks,