[YARN-1071] ResourceManager's decommissioned and lost node count is 0 after restart

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.1.0-beta
• Fix Version/s: 2.4.0
• Component/s: resourcemanager
• Hadoop Flags: Reviewed
• Assignee: Jian He
• Reporter: Srimanth Gunturi
• Created: 16/Aug/13 17:24
• Updated: 18/Sep/14 18:14
• Resolved: 21/Feb/14 06:52

Description

I had 6 nodes in a cluster with 2 NMs stopped. Then I put a host into YARN's yarn.resourcemanager.nodes.exclude-path. After running yarn rmadmin -refreshNodes, RM's JMX correctly showed decommissioned node count:

"NumActiveNMs" : 3,
"NumDecommissionedNMs" : 1,
"NumLostNMs" : 2,
"NumUnhealthyNMs" : 0,
"NumRebootedNMs" : 0

After restarting RM, the counts were shown as below in JMX.

"NumActiveNMs" : 3,
"NumDecommissionedNMs" : 0,
"NumLostNMs" : 0,
"NumUnhealthyNMs" : 0,
"NumRebootedNMs" : 0

Notice that the lost and decommissioned NM counts are both 0.

Issue Links

• relates to: YARN-2567 Add a percentage-node threshold for RM to wait for new allocations after restart/failover
• relates to: AMBARI-2940 After restarting YARN, the number of lost nodes is incorrect