juliuszsompolski opened a new pull request, #42282:
URL: https://github.com/apache/spark/pull/42282

   ### What changes were proposed in this pull request?
   
   If the ExecutePlan never reached the server, a ReattachExecute will fail 
with INVALID_HANDLE.OPERATION_NOT_FOUND. In that case, we could try to send 
ExecutePlan again.
   
   Note: This is an "overkill" version of the change.
   Even though we empirically observed that error is throws only from first 
next() or hasNext() of the response StreamObserver, wrap the initial call in 
retries as well to not depend on it in case it's just an quirk of GRPC that's 
not fully dependable.
   
   ### Why are the changes needed?
   
   This solves an edge case of reattachable execution where the initial 
execution never reached the server.
   
   ### Does this PR introduce _any_ user-facing change?
   
   No.
   
   ### How was this patch tested?
   
   Testing these failures is difficult, will require some special testing setup


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


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

Reply via email to