RocMarshal commented on code in PR #27539:
URL: https://github.com/apache/flink/pull/27539#discussion_r3006364815
##########
flink-runtime/src/main/java/org/apache/flink/runtime/scheduler/adaptive/Executing.java:
##########
@@ -99,6 +98,7 @@ class Executing extends StateWithExecutionGraph
this.rescaleOnFailedCheckpointCount = rescaleOnFailedCheckpointCount;
this.failedCheckpointCountdown = null;
+ // 如果不是从sp过来则记录，否则从sp过来则不记录
Review Comment:
Hi, @ferenc-csaky Thank you for the reminder.
To be exactly, It should be a `TODO` line in English.
With offline-discussion with @WeiZhong94 , We believe there may be some
uncertainty and potential risks around this part of the code, so I’ve added a
marker here.
Currently, in AdaptiveScheduler, when stopWithSavepoint fails, the job is
not stopped but instead transitions back to the Executing state. This may
result in an extra rescale record being generated without any actual change in
parallelism.
I will make corresponding adjustments once we reach a clearer conclusion.
Of course, we would greatly appreciate your thoughts on this as well.
A candidate solution is here
https://github.com/apache/flink/pull/27539/changes#r3007348820
--
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]
