Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
2.7.5
-
None
-
None
Description
Currently, when we execute
hdfs mover -p /some/path
the moverThreadAllocator in org.apache.hadoop.hdfs.server.balancer.Dispatcher does not initialize lotSize according to dfs.mover.moverThreads and dfs.datanode.balance.max.concurrent.moves. So, when we invoke moverThreadAllocator.allocate() method, it will always return 1.