[YARN-10459] containerLaunchedOnNode method not need to hold schedulerApptemt lock

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.2.0, 3.1.3
• Fix Version/s: 3.4.0
• Component/s: scheduler
• Labels: pull-request-available
• Target Version/s: 3.2.0, 3.1.3, 3.4.0
• Hadoop Flags: Reviewed
• Assignee: Minni Mittal
• Reporter: Ryan Wu
• Created: 14/Oct/20 10:28
• Updated: 11/Feb/24 10:22
• Resolved: 01/Feb/22 17:38

Description

 

Now, the containerLaunchedOnNode method hold the SchedulerApplicationAttempt writelock, but looking at the method, it does not change any field. And more seriously, this will affect the scheduler.

 public void containerLaunchedOnNode(ContainerId containerId, NodeId nodeId) { 
    // Inform the container 
    writelock.lock 
    try { 
      RMContainer rmContainer = getRMContainer(containerId); 
      if (rmContainer == null) { 
      // Some unknown container sneaked into the system. Kill it.          
        rmContext.getDispatcher().getEventHandler().handle( new RMNodeCleanContainerEvent(nodeId, containerId)); return; 
      } 
      rmContainer.handle( new RMContainerEvent(containerId, RMContainerEventType.LAUNCHED)); 
   }finally { 
      writeLock.unlock(); 
   } 
}  

 

Issue Links

• links to: GitHub Pull Request #3195