One of datanodes in my cluster is down due to the storage being full.
Then I clean some space, however, the datanode throws following error:
Caught exception while adding replicas from /sata/hadoop/hdfs/data/current. Will throw later
block pool BP-***** is not found
BlockPoolSliceStorage.recoverTransitionRead:attempt to load an used block storage: /sata/hadoop/hdfs/data/current/BP-*****
Then even I delete the whole hdfs data dir, and datanode make a new one automatically, it still throw the same error message.