[TEZ-148] DAG state machine does not handle successfully completing vertices at kill wait

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Labels: TEZ-0.2.0 TEZ-1
• Assignee: Mike Liddell
• Reporter: Hitesh Shah
• Created: 20/May/13 21:08
• Updated: 09/Jul/13 00:52
• Resolved: 09/Jul/13 00:52

Description

A race between a kill and vertices completing successfully seems to not be handled.

impl.DAGImpl (DAGImpl.java:handle(592)) - Can't handle this event at current state
org.apache.hadoop.yarn.state.InvalidStateTransitonException: Invalid event: DAG_VERTEX_COMPLETED at KILL_WAIT
at org.apache.hadoop.yarn.state.StateMachineFactory$MultipleInternalArc.doTransition(StateMachineFactory.java:385)
at org.apache.hadoop.yarn.state.StateMachineFactory.doTransition(StateMachineFactory.java:299)
at org.apache.hadoop.yarn.state.StateMachineFactory.access$300(StateMachineFactory.java:43)
at org.apache.hadoop.yarn.state.StateMachineFactory$InternalStateMachine.doTransition(StateMachineFactory.java:445)

checkForCompletion ignores the current kill wait state and sends back a final state of succeeded causing a problem in the state machine as it can only go to KILL_WAIT or KILLED from KILL_WAIT.

Issue Links

• relates to: TEZ-141 DAG does not kill running vertices when going into failed state