StefanRRichter commented on a change in pull request #6684: [FLINK-10205]
Batch Job: InputSplit Fault tolerant for DataSource…
URL: https://github.com/apache/flink/pull/6684#discussion_r240252512
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/ExecutionVertex.java
##########
@@ -590,6 +604,15 @@ public Execution resetForNewExecution(final long
timestamp, final long originati
this.currentExecution = newExecution;
+ synchronized (this.inputSplits){
Review comment:
I wonder if the synchronization on `inputSplits` is even needed? In the
failover case, `synchronized (priorExecutions)` makes failing exclusive and
failing checks that the previous attempt is no longer running before
manipulating the input splits. Getting the next split should then only be
called from one attempt at a time, once its running and that should be from a
single thread. Did I miss something?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
