Vanilla cluster can't decommission node if replication factor < 3

[Vadim Rovachev]

Steps to reproduce:
1. Start vanilla cluster with replication factor 2 and with next structure:
NN+JT x 1 node
SNN x 1 node
TT+DN x 2 nodes
TT x 1 node
DN x 1 node
2. Start scaling with next parameters:
decommission TT x 1 node
decommission DN x 1 node
add TT x 1 node
add DN x 1 node

Cluster receives state "Error".

log in master node:
2014-02-20 11:57:44,558 INFO org.apache.hadoop.hdfs.server.namenode.FSNamesystem: Start Decommissioning node 192.168.111.4:50010
2014-02-20 11:57:44,558 INFO org.apache.hadoop.hdfs.server.namenode.FSNamesystem: Block: blk_-777377507232042508_1130, Expected Replicas: 3, live replicas: 2, corrupt replicas: 0, decommissioned replicas: 1, excess replicas: 0, Is Open File: false, Datanodes having this block: 192.168.111.4:50010 192.168.111.8:50010 192.168.111.5:50010 , Current Datanode: 192.168.111.4:50010, Is current datanode decommissioning: true

[Michael McCune (mimccune)]

I am unable to reproduce this error. I am using a trunk version of Sahara (commit c1d6d02ab7b6c8e50bbe637d9f68832f19dbcdc6) with a stable/icehouse version of devstack. My image is ubuntu vanilla hdp1 created using sahara-image-elements.

Step 1, no problems. Step 2, I am able to remove and then add all the nodes listed and the cluster returns to "Active" status.

[Andrew Lazarev (alazarev)]

I can't reproduce the issue too. Scaled well on vanilla 1.2.1.

[Andrew Lazarev (alazarev)]

Issue was reported a year ago and there is no reproduction since then. Closing as invalid.