Closed lewiszlw closed 4 months ago
Closes #.
When partition count of execution plan is 0, the job will hang because available tasks of running stage will be empty, then get_running_stage_id method of ExecutionGraph will return None. https://github.com/apache/datafusion-ballista/blob/e7cb3d57f61900ea56107c6477c5754429ba821e/ballista/scheduler/src/state/execution_graph.rs#L969 So the running stage will keep running and won't be converted to successful/failed stage.
get_running_stage_id
ExecutionGraph
Thanks @lewiszlw. Could you rebase/upmerge (to fix the CI failure) then I can review
Which issue does this PR close?
Closes #.
Rationale for this change
When partition count of execution plan is 0, the job will hang because available tasks of running stage will be empty, then
get_running_stage_id
method ofExecutionGraph
will return None. https://github.com/apache/datafusion-ballista/blob/e7cb3d57f61900ea56107c6477c5754429ba821e/ballista/scheduler/src/state/execution_graph.rs#L969 So the running stage will keep running and won't be converted to successful/failed stage.What changes are included in this PR?
Are there any user-facing changes?