StephanEwen commented on a change in pull request #8756: [FLINK-12406]
[Runtime] Report BLOCKING_PERSISTENT result partition meta back to client
URL: https://github.com/apache/flink/pull/8756#discussion_r299701049
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/ExecutionGraph.java
##########
@@ -814,6 +820,69 @@ public Executor getFutureExecutor() {
entry -> serializeAccumulator(entry.getKey(),
entry.getValue())));
}
+ @Override
+ public BlockingPersistentResultPartitionMeta
getBlockingPersistentResultPartitionMeta() {
+ Map<AbstractID, Map<AbstractID, ResultPartitionDescriptor>>
resultPartitionDescriptors = new HashMap<>();
+
+ // keep record of all failed IntermediateDataSetID
+ Set<AbstractID> failedIntermediateDataSetIds = new HashSet<>();
+
+ for (ExecutionVertex executionVertex :
getAllExecutionVertices()) {
+ for (IntermediateResultPartition
intermediateResultPartition : executionVertex.getProducedPartitions().values())
{
+ if (intermediateResultPartition.getResultType()
== ResultPartitionType.BLOCKING_PERSISTENT) {
+ try {
+
addLocation(resultPartitionDescriptors, intermediateResultPartition);
+ } catch (Throwable throwable) {
+ LOG.error("Failed to get
location of ResultPartition: " + intermediateResultPartition.getPartitionId(),
throwable);
+
failedIntermediateDataSetIds.add(
+ new
AbstractID(intermediateResultPartition.getIntermediateResult().getId()));
+ }
+ }
+ }
+ }
+
+ return new
BlockingPersistentResultPartitionMeta(resultPartitionDescriptors,
failedIntermediateDataSetIds);
+ }
+
+ /**
+ *
+ * @param resultPartitionDescriptors
+ * @param intermediateResultPartition
+ * throw exception if any error occurs.
+ */
+ public void addLocation(
+ Map<AbstractID, Map<AbstractID, ResultPartitionDescriptor>>
resultPartitionDescriptors,
+ IntermediateResultPartition intermediateResultPartition) {
+
+ IntermediateDataSetID dataSetID =
intermediateResultPartition.getIntermediateResult().getId();
+
+ Map<AbstractID, ResultPartitionDescriptor> map =
resultPartitionDescriptors.computeIfAbsent(
+ new AbstractID(dataSetID), key -> new HashMap<>()
+ );
+
+ TaskManagerLocation taskManagerLocation = null;
+
+ // The taskManagerLocation should be ready already since the
previous job is done.
+ try {
+ taskManagerLocation = intermediateResultPartition
+
.getProducer().getCurrentExecutionAttempt().getTaskManagerLocationFuture().get(1,
TimeUnit.SECONDS);
Review comment:
This is a blocking waiting call, which cannot be used in a non blocking data
structure like the execution graph.
The call to the future needs to complete or fail instantly.
----------------------------------------------------------------
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:
us...@infra.apache.org
With regards,
Apache Git Services
