tillrohrmann commented on a change in pull request #7356:
[FLINK-10868][flink-yarn] Enforce maximum failed TMs in YarnResourceManager
URL: https://github.com/apache/flink/pull/7356#discussion_r250561743
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java
##########
@@ -825,6 +843,12 @@ protected void closeTaskManagerConnection(final
ResourceID resourceID, final Exc
slotManager.unregisterTaskManager(workerRegistration.getInstanceID());
workerRegistration.getTaskExecutorGateway().disconnectResourceManager(cause);
+ failedContainerSoFar.getAndAdd(1);
+ if (failedContainerSoFar.intValue() >=
maximumAllowedTaskManagerFailureCount) {
+ rejectAllPendingSlotRequests(new
MaximumFailedTaskManagerExceedingException(
+ String.format("Maximum number of failed
container %d "
+ + "is detected in Resource
Manager", failedContainerSoFar.intValue())));
+ }
Review comment:
By counting it here we also count heartbeat timeouts where the container has
not died as container failures. I think this is not good. Instead I would
suggest to only count true container failures.
----------------------------------------------------------------
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
