Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1 KB

YARN-8650.md

File metadata and controls

35 lines (22 loc) · 1 KB

YARN-8650

Scenario

YARN-8650
  1. NM is launching the container
 protected int prepareForLaunch(ContainerStartContext ctx) throws IOException {
   ContainerId containerId = container.getContainerId();//crash point
   dispatcher.getEventHandler().handle(new ContainerEvent(
       containerId,
       ContainerEventType.CONTAINER_LAUNCHED));
 }
  1. Shutdown command is sent to NM who holds the container.
  2. Recovery process will send KILL event to Container, and the state change to DONE
  3. Then Launcher will send LAUNCH_FAILED event to RMContainer( in code#6)

Trigger analysis

  1. The sleep after shutdown is very important to flip the KILL event and LAUNCHED event.

Crash Point

pre-read

org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch call 289