XComp commented on code in PR #24382:
URL: https://github.com/apache/flink/pull/24382#discussion_r1502831868
##########
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/DefaultExecutionGraph.java:
##########
@@ -1031,34 +1028,18 @@ public void suspend(Throwable suspensionCause) {
return;
} else if (transitionState(state, JobStatus.SUSPENDED,
suspensionCause)) {
initFailureCause(suspensionCause, System.currentTimeMillis());
+ resetExecutionGraph(ExecutionJobVertex::suspend)
+ .whenComplete(
+ (Void ignored, Throwable throwable) -> {
+ if (throwable != null) {
+ LOG.debug(
+ "Could not properly suspend the
execution graph.",
+ throwable);
+ }
- incrementRestarts();
-
- // cancel ongoing scheduling action
- if (schedulingFuture != null) {
- schedulingFuture.cancel(false);
- }
- final ArrayList<CompletableFuture<Void>>
executionJobVertexTerminationFutures =
- new ArrayList<>(verticesInCreationOrder.size());
-
- for (ExecutionJobVertex ejv : verticesInCreationOrder) {
- executionJobVertexTerminationFutures.add(ejv.suspend());
- }
-
- final ConjunctFuture<Void> jobVerticesTerminationFuture =
-
FutureUtils.waitForAll(executionJobVertexTerminationFutures);
-
- checkState(jobVerticesTerminationFuture.isDone(), "Suspend needs
to happen atomically");
Review Comment:
thinking about it once more: It does change the behavior because the old
implementation would result in a fatal error. The new version doesn't do this.
I will add the precondition back. :innocent:
--
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]
