Kevin Sweeney created AURORA-778:
------------------------------------

             Summary: Persist cause of task state transitions in 
machine-readable format
                 Key: AURORA-778
                 URL: https://issues.apache.org/jira/browse/AURORA-778
             Project: Aurora
          Issue Type: Story
          Components: Scheduler
            Reporter: Kevin Sweeney


Add a new field to TaskEvents that identifies the cause of a transition and 
export metrics based on the cause (in addition to state). This will be useful 
to track down the cause of LOST tasks.

Possible implementation:
{noformat}
enum TransitionCause {
  UNKNOWN = 0,
  SLAVE_REMOVED = 1,
  EXECUTOR_CRASHED = 2,
  TIMED_OUT_BY_SCHEDULER = 3,
  FORCE_TASK_STATE_RPC = 4,
  // ...
}
{noformat}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to