[AMBARI-21731] Rolling Decommission of HDFS Datanodes

• Type: New Feature
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.5.1
• Component/s: ambari-server
• Labels: datanode decommision hdfs rolling rolling_decommision
• Flags: Important
• Assignee: Unassigned
• Reporter: Shahrukh Khan
• Created: 16/Aug/17 06:57
• Updated: 17/Aug/17 09:16

Description

Implement feature of Rolling Decommission for HDFS Datanodes.

Use "Under-Replicated Blocks" as parameter for Rolling Decommission, Instead of Wait Time.

E.g. - To decommission 100 HDFS Datanodes - with rolling decommission of 3 nodes at a time, a batch of decommission should trigger when a threshold of - "Minimum number of Under-Replicated Blocks" has reached.

*Note - Minimum Threshold of - "Under-Replicated Blocks" has to be taken care so that we don't end up in a situation with huge load on HDFS Namenode and high number of Under Replicated Blocks - Specially with Datanodes of huge Data size ( ~ TB )