Details
-
Sub-task
-
Status: Closed
-
Major
-
Resolution: Abandoned
-
None
-
None
-
None
Description
Block replication strategies need topology information for ideal block placements. This information is available in resource managers and/or can be provided separately through scripts/services/classes, as in the case of HDFS. This jira focuses on enhancing spark-yarn package to suitably extract topology information from YARN.