mridulm commented on code in PR #38371:
URL: https://github.com/apache/spark/pull/38371#discussion_r1009018882
##########
core/src/test/scala/org/apache/spark/scheduler/DAGSchedulerSuite.scala:
##########
@@ -3089,13 +3089,14 @@ class DAGSchedulerSuite extends SparkFunSuite with
TempLocalSparkContext with Ti
submit(finalRdd, Array(0, 1), properties = new Properties())
// Finish the first 2 shuffle map stages.
- completeShuffleMapStageSuccessfully(0, 0, 2)
+ completeShuffleMapStageSuccessfully(0, 0, 2, Seq("hostA", "hostB"))
assert(mapOutputTracker.findMissingPartitions(shuffleId1) ===
Some(Seq.empty))
completeShuffleMapStageSuccessfully(1, 0, 2, Seq("hostB", "hostD"))
assert(mapOutputTracker.findMissingPartitions(shuffleId2) ===
Some(Seq.empty))
- // Executor lost on hostB, both of stage 0 and 1 should be reran.
+ // FetchFailed on stage 2, both of stage 1 and 2 should be reran. Besides,
executor lost on
+ // hostB, both of stage 0 and 1 should be reran.
Review Comment:
The comment could read - "Executor lost on hostB, both of stage 0 (if
missing partitions) and 1 should be reran - as part of recomputation of stage
2" : if no output on hostB for stage 0, we can omit that from the list of
recomputed stages.
Will that clarify ?
--
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]
