[HADOOP-11000] HAServiceProtocol's health state is incorrectly transitioned to SERVICE_NOT_RESPONDING

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.7.0
• Hadoop Flags: Reviewed
• Assignee: Ming Ma
• Reporter: Ming Ma
• Created: 25/Aug/14 17:20
• Updated: 24/Apr/15 22:49
• Resolved: 17/Feb/15 09:51

Description

When HAServiceProtocol.monitorHealth throws a HealthCheckFailedException, the actual exception from protocol buffer RPC is a RemoteException that wraps the real exception. Thus the state is incorrectly transitioned to SERVICE_NOT_RESPONDING

HealthMonitor.java
doHealthChecks

      try {
        status = proxy.getServiceStatus();
        proxy.monitorHealth();
        healthy = true;
      } catch (HealthCheckFailedException e) {
        .....
        enterState(State.SERVICE_UNHEALTHY);
      } catch (Throwable t) {
        .....
        enterState(State.SERVICE_NOT_RESPONDING);
        .....
      }