[HBASE-19102] TestZooKeeperMainServer fails with KeeperException$ConnectionLossException

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.4.0, 1.3.2, 2.0.0-beta-1, 2.0.0, 1.2.7
• Assignee: Michael Stack
• Reporter: Michael Stack
• Created: 27/Oct/17 04:03
• Updated: 01/Aug/18 06:23
• Resolved: 07/Nov/17 04:42

Description

I'm trying to run test suite on a local machine. I never get to the second part because I fail on below test with below exception near every time (and an ipv6 test... will do that next).

1 -------------------------------------------------------------------------------
2 Test set: org.apache.hadoop.hbase.zookeeper.TestZooKeeperMainServer
3 -------------------------------------------------------------------------------
4 Tests run: 2, Failures: 0, Errors: 1, Skipped: 0, Time elapsed: 16.161 s <<< FAILURE! - in org.apache.hadoop.hbase.zookeeper.TestZooKeeperMainServer
5 testCommandLineWorks(org.apache.hadoop.hbase.zookeeper.TestZooKeeperMainServer) Time elapsed: 15.848 s <<< ERROR!
6 org.apache.zookeeper.KeeperException$ConnectionLossException: KeeperErrorCode = ConnectionLoss for /testCommandLineWorks
7 at org.apache.hadoop.hbase.zookeeper.TestZooKeeperMainServer.testCommandLineWorks(TestZooKeeperMainServer.java:81)

Looks like running the command before we are connected causes the above – we pause 15 seconds and then throw the above. If I wait until connected before proceding, stuff seems to work reliably. I don't have access to the watcher on connections since we override the zk main class... so this seems only avenue available at mo (This zk main thing is all a hack around zk main because it had bugs ... but I think we have to keep the hack because folks use different versions of zk. My $workplace defaults to something that is years old, 3.4.5 for instance).

Issue Links

• links to: Review Board (master)