Fixing Over-replicated Blocks

[#1]

CentOS 6.6
CDH 5.1.2

Due to space pressure, I need to reduce replication factor of existing files from 3 to 2.

A command like the following is executed

[hdfs]$ hdfs dfs -setrep -R -w 2  /path/of/files

A warning about "waiting time may be long for DECREASING the number of replications" appeared.

I am still waiting after tens of minutes. And fsck still showing over-replication. 

[hdfs]$ hdfs fsck /path/of/files
16/11/02 12:04:42 WARN ssl.FileBasedKeyStoresFactory: The property 'ssl.client.truststore.location' has not been set, no TrustStore will be loaded
Connecting to namenode via http://namenode1:50070
FSCK started by hdfs (auth:SIMPLE) from /192.168.88.38 for path /path/of/files at Wed Nov 02 12:04:43 HKT 2016
....Status: HEALTHY
 Total size:	129643323 B
 Total dirs:	1
 Total files:	4
 Total symlinks:		0
 Total blocks (validated):	4 (avg. block size 32410830 B)
 Minimally replicated blocks:	4 (100.0 %)
 Over-replicated blocks:	4 (75.0 %)
 Under-replicated blocks:	0 (0.0 %)
 Mis-replicated blocks:		0 (0.0 %)
 Default replication factor:	3
 Average block replication:	3
 Corrupt blocks:		0
 Missing replicas:		0 (0.0 %)
 Number of data-nodes:		6
 Number of racks:		1
FSCK ended at Wed Nov 02 12:04:43 HKT 2016 in 1 milliseconds


The filesystem under path ' /path/of/files' is HEALTHY

Is this normal? How long should the wait be?

[#2]

The setrep command just completed. However, the fsck still showing over-replication.

[#3 · Accepted solution]

Answering my question...

The source code of org.apache.hadoop.hdfs.server.blockmanagement.BlockManager says

...
    if (numCurrentReplica > expectedReplication) {
      if (num.replicasOnStaleNodes() > 0) {
        // If any of the replicas of this block are on nodes that are
        // considered "stale", then these replicas may in fact have
        // already been deleted. So, we cannot safely act on the
        // over-replication until a later point in time, when
        // the "stale" nodes have block reported.
        return MisReplicationResult.POSTPONE;
      }
...

So the key point is whether the DataNodes are "stale". I don't know how to force the nodes to have block reported besides restarting. So I restarted all DataNode and over-replicated blocks gone.