[HBASE-17996] HBase master fails to start sometimes on RHEL7

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Won't Fix
• Component/s: master, test
• Assignee: Unassigned
• Reporter: David Knupp
• Created: 04/May/17 17:23
• Updated: 01/Aug/18 06:22
• Resolved: 16/Jun/17 23:10

Description

Impala includes HBase in its local test environment, and we have found that intermittently, the HBase master node fails to start when we are testing on RHEL7.

In these failures, what we typically see in the logs is this:

17/04/29 21:33:47 INFO zookeeper.ClientCnxn: Session establishment complete on server localhost/0:0:0:0:0:0:0:1:2181, sessionid = 0x15bbd21b797000a, negotiated timeout = 90000
17/04/29 21:33:47 INFO client.ZooKeeperRegistry: ClusterId read in ZooKeeper is null
17/04/29 21:33:48 INFO master.ActiveMasterManager: Deleting ZNode for /hbase/backup-masters/localhost,16000,1493526758211 from backup master directory

On a successful startup, the log looks like this:

17/04/16 21:32:29 INFO zookeeper.ClientCnxn: Session establishment complete on server localhost/0:0:0:0:0:0:0:1:2181, sessionid = 0x15b7a2ed6860005, negotiated timeout = 90000
17/04/16 21:32:29 INFO client.ZooKeeperRegistry: ClusterId read in ZooKeeper is null
17/04/16 21:32:30 INFO util.FSUtils: Created version file at hdfs://localhost:20500/hbase with version=8
17/04/16 21:32:31 INFO master.MasterFileSystem: BOOTSTRAP: creating hbase:meta region

So the event that we don't see in the failed start up attempts is master.MasterFileSystem: BOOTSTRAP: creating hbase:meta region.

The full logs will be attached.