[AMBARI-13320] Cannot Perform RU If Ambari Server Is Not A Part of the Cluster

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.1.2
• Fix Version/s: 2.2.0
• Component/s: ambari-server
• Assignee: Dmitry Lysnichenko
• Reporter: Dmitry Lysnichenko
• Created: 06/Oct/15 10:48
• Updated: 07/Oct/15 19:23
• Resolved: 07/Oct/15 16:33

Description

A commit was merged into branch-2.1 and branch-2.1.2 which prevents a rolling upgrade from starting unless the Ambari Server is a part of the cluster. When Ambari is not a part of the cluster, a hostname is not correctly chosen for server-side actions and the following exception is thrown at upgrade time:

java.lang.NullPointerException
	at org.apache.ambari.server.actionmanager.ActionDBAccessorImpl.persistActions(ActionDBAccessorImpl.java:300)
	at org.apache.ambari.server.orm.AmbariJpaLocalTxnInterceptor.invoke(AmbariJpaLocalTxnInterceptor.java:68)
	at org.apache.ambari.server.actionmanager.ActionManager.sendActions(ActionManager.java:99)
	at org.apache.ambari.server.controller.internal.RequestStageContainer.persist(RequestStageContainer.java:216)
	at org.apache.ambari.server.controller.internal.UpgradeResourceProvider.createUpgrade(UpgradeResourceProvider.java:752)
	at org.apache.ambari.server.controller.internal.UpgradeResourceProvider.access$100(UpgradeResourceProvider.java:116)
	at org.apache.ambari.server.controller.internal.UpgradeResourceProvider$1.invoke(UpgradeResourceProvider.java:284)
	at org.apache.ambari.server.controller.internal.UpgradeResourceProvider$1.invoke(UpgradeResourceProvider.java:274)
	at org.apache.ambari.server.controller.internal.AbstractResourceProvider.createResources(AbstractResourceProvider.java:272)
	at org.apache.ambari.server.controller.internal.UpgradeResourceProvider.createResources(UpgradeResourceProvider.java:274)
	at org.apache.ambari.server.controller.internal.ClusterControllerImpl.createResources(ClusterControllerImpl.java:289)

The problem is the following line of code:
https://github.com/apache/ambari/blob/branch-2.1.2/ambari-server/src/main/java/org/apache/ambari/server/controller/internal/UpgradeResourceProvider.java#L1281

    String hostName = null;
    Collection<Long> hostIds = cluster.getAllHostsDesiredConfigs().keySet();
    if (!hostIds.isEmpty()) {
      Long hostId = hostIds.iterator().next();
      HostEntity hostEntity = s_hostDAO.findById(hostId);
      if (hostEntity != null) {
        hostName = hostEntity.getHostName();
      }
    }

...

    stage.addServerActionCommand(task.getImplementationClass(), Role.AMBARI_SERVER_ACTION,
        RoleCommand.EXECUTE, cluster.getClusterName(),
        new ServiceComponentHostServerActionEvent(null, System.currentTimeMillis()), commandParams,
        itemDetail, null, Integer.valueOf(1200), allowRetry,
        context.isComponentFailureAutoSkipped());

The calculated hostName is never used to construct the server-side action. The fix is simple - use the hostName when creating the action.

Issue Links

• links to: review_request_