figured it out. We were facing a scenario where some of the files had replication factor more than the actual number of datanodes.
https://issues.apache.org/jira/browse/MAPREDUCE-2845 That is why the Namenode was unable to decommission the machine, since it could not get the replica count settled. Thanks Divye Sheth On Fri, Mar 7, 2014 at 6:33 PM, divye sheth <divs.sh...@gmail.com> wrote: > Hi, > > I am using hadoop 0.20.2 and have decommissioned one of my servers. It has > around 300G's of data which if I am correct will be distributed across the > other machines and slowly all the blocks will be moved from the > decommissioned machine. > > The issue I am facing is from the past 6hours I do not see any change in > capacity of neither the machine which has been decommissioned nor the other > machines. > > How should I verify if the decommissioning process is in progress? > > Note: Namanode says it has started decommission of the machine, even its > UI show decommission in progress. > > Please help me out here. > > Thanks > Divye Sheth >