Leanken.Lin created SPARK-33016:
-----------------------------------

             Summary: Potential SQLMetrics missed which might cause WEB UI 
display issue while AQE is on.
                 Key: SPARK-33016
                 URL: https://issues.apache.org/jira/browse/SPARK-33016
             Project: Spark
          Issue Type: Bug
          Components: SQL
    Affects Versions: 3.0.1
            Reporter: Leanken.Lin


In current AQE execution, there might be a following scenario which might cause 
SQLMetrics being incorrectly override.
 # Stage A and B are created, and UI updated thru event 
onAdaptiveExecutionUpdate.
 # Stage A and B are running. Subquery in stage A keep updating metrics thru 
event onAdaptiveSQLMetricUpdate.
 # Stage B completes, while stage A's subquery is still running, updating 
metrics.
 # Completion of stage B triggers new stage creation and UI update thru event 
onAdaptiveExecutionUpdate again (just like step 1).

 

But it's very hard to re-produce this issue, since it was only happened with 
high concurrency. For the fix, I suggested that we might be able to keep all 
duplicated metrics instead of updating it every time.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to