GJL commented on a change in pull request #9860: [FLINK-14331][runtime] Reset
vertices right after they transition to terminated states
URL: https://github.com/apache/flink/pull/9860#discussion_r334401412
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/ExecutionVertex.java
##########
@@ -706,8 +706,10 @@ private Execution resetForNewExecutionInternal(final long
timestamp, final long
public void tryAssignResource(LogicalSlot slot) {
if (!currentExecution.tryAssignResource(slot)) {
- throw new IllegalStateException("Could not assign
resource " + slot + " to current execution " +
- currentExecution + '.');
+ final IllegalStateException e = new
IllegalStateException(
+ "Could not assign resource " + slot + " to
current execution " + currentExecution + '.');
+ slot.releaseSlot(e);
Review comment:
> Actually I can see that in `waitForAllSlotsAndDeploy` such unexpected
errors are already propagated from `assignResourceOrHandleError` to
`deployAll`, then to `assertNoException`
I think it's not enough currently. If there is an error when assigning the
resource, the error will be handled in `deployOrHandleError()` by restarting
the tasks. Because `.handle()` does not propagate the original error,
`assertNoException` will not be triggered.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
