Details
-
Sub-task
-
Status: Patch Available
-
Major
-
Resolution: Unresolved
-
None
-
None
Description
In order to balance datanode storage utilization of a cluster, Balancer (1) classifies datanodes into different groups (overUtilized, aboveAvgUtilized, belowAvgUtilized and underUtilized), (2) chooses source and target datanode pairs and (3) chooses blocks to move. Some of these logic are in Dispatcher. It is better to separate them out. This JIRA is a further work of HDFS-6828.