[AMBARI-21539] Resource Manager fails to restart properly during an IOP to HDP upgrade

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.5.2
• Fix Version/s: 2.5.2
• Component/s: ambari-server
• Assignee: Attila Magyar
• Reporter: Tim Thorpe
• Created: 20/Jul/17 18:40
• Updated: 23/Jul/17 18:17
• Resolved: 23/Jul/17 18:17

Description

2017-07-20 06:19:28,250 INFO service.AbstractService (AbstractService.java:noteFailure(272)) - Service org.apache.hadoop.yarn.server.resourcemanager.recovery.RMStateStore failed in state STARTED; cause: org.apache.zookeeper.KeeperException$NoAuthException: KeeperErrorCode = NoAuth for /rmstore/ZKRMStateRoot
org.apache.zookeeper.KeeperException$NoAuthException: KeeperErrorCode = NoAuth for /rmstore/ZKRMStateRoot
at org.apache.zookeeper.KeeperException.create(KeeperException.java:113)
at org.apache.zookeeper.KeeperException.create(KeeperException.java:51)
at org.apache.zookeeper.ZooKeeper.setACL(ZooKeeper.java:1399)
at org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore$3.run(ZKRMStateStore.java:372)

Workaround:
/usr/hdp/current/zookeeper-server/bin/zkCli.sh -server 127.0.0.1:2181
rmr /rmstore

This workaround fails in a kerberized environment with this error:
Authentication is not valid : /rmstore/ZKRMStateRoot

Issue Links

• links to: review board