[HBASE-13733] Failed MiniZooKeeperCluster startup did not shutdown ZK servers

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.1.0, 1.2.0, 2.0.0
• Fix Version/s: 1.2.0, 1.1.1, 2.0.0
• Component/s: Zookeeper
• Hadoop Flags: Reviewed
• Assignee: Stephen Yuan Jiang
• Reporter: Stephen Yuan Jiang
• Created: 21/May/15 01:43
• Updated: 24/Jun/22 18:58
• Resolved: 21/May/15 21:05

Description

MiniZooKeeperCluster#startup() starts servers one-by-one, if everything is good, it would declare success of start:

  public int startup(File baseDir, int numZooKeeperServers) 
    ...
    // running all the ZK servers
    for (int i = 0; i < numZooKeeperServers; i++) {
    ...===> could throw exception in the loop and end the startup
      // Start up this ZK server
      standaloneServerFactory.startup(server);
      ...
      standaloneServerFactoryList.add(standaloneServerFactory);
      zooKeeperServers.add(server);
    }
   ...
    started = true;
    ...
  }

However, if exception throws in the middle of start up (eg. some servers already started), the MiniZooKeeperCluster#shutdown() would not shut down them and clean up resources.

  public void shutdown() throws IOException {
    if (!started) {
      return;
    }
    ...
  }