we are using mapr 3.x with hadoop 0.20. We tried to decommission a node by moving it to /decommission topology. Its a 10TB hard drive on 10GB ethernet. Its been more than 72 hours and looks like process is stuck when node still has 2% of data. We have monitored draining process via MCS and using `maprcli dump volumenodes` . Other nodes has more than enough space to accommodate.