[YARN-2371] Wrong NMToken is issued when NM preserving restarts with containers running

• Type: Sub-task
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Fix Version/s: 2.5.0
• Component/s: resourcemanager
• Assignee: Hong Zhiguo
• Reporter: Hong Zhiguo
• Created: 30/Jul/14 06:45
• Updated: 30/Jun/15 07:20
• Resolved: 28/Aug/14 20:17

Description

When application is submitted with "ApplicationSubmissionContext.getKeepContainersAcrossApplicationAttempts() == true", and NM is restarted with containers running, wrong NMToken is issued to AM through RegisterApplicationMasterResponse.
See the NM log:

2014-07-30 11:59:58,941 ERROR org.apache.hadoop.yarn.server.nodemanager.containermanager.ContainerManagerImpl: Unauthorized request to start container.-
NMToken for application attempt : appattempt_1406691610864_0002_000001 was used for starting container with container token issued for application attempt : appattempt_1406691610864_0002_000002

The reason is in below code:

 
createAndGetNMToken(String applicationSubmitter,
      ApplicationAttemptId appAttemptId, Container container) {
      ......
          Token token =
              createNMToken(container.getId().getApplicationAttemptId(),
                container.getNodeId(), applicationSubmitter);
     ......
}

"appAttemptId" instead of "container.getId().getApplicationAttemptId()" should be passed to "createNMToken".

Issue Links

• is duplicated by: YARN-2065 AM cannot create new containers after restart-NM token from previous attempt used
• relates to: YARN-2433 Stale token used by restarted AM (with previous containers retained) to request new container