[HADOOP-4489] Blocks have been scheduled to delete by NameNode but are not deleted on the DataNode

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Won't Fix
• Affects Version/s: 0.17.0
• Assignee: Unassigned
• Reporter: Hairong Kuang
• Created: 22/Oct/08 18:51
• Updated: 23/Apr/09 19:25
• Resolved: 31/Mar/09 00:24

Description

NameNode sometimes scheduled a replica to be deleted as shown in the NameNode log:

INFO org.apache.hadoop.dfs.StateChange: BLOCK* ask DataNode XX to delete blk_YY

But block YY did not get deleted on the DataNode XX. The datanode log had no indication of where the deletion was successful or failed. The block YY remains on the DataNode XX's disk with no timestamp change.

Issue Links

• is related to: HDFS-15 Rack replication policy can be violated for over replicated blocks