Github user GJL commented on a diff in the pull request:
https://github.com/apache/flink/pull/5397#discussion_r165298568
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/taskexecutor/TaskExecutor.java
---
@@ -305,6 +311,84 @@ public void postStop() throws Exception {
// RPC methods
//
======================================================================
+ @Override
+ public CompletableFuture<StackTraceSampleResponse>
requestStackTraceSample(
+ final ExecutionAttemptID executionAttemptId,
+ final int sampleId,
+ final int numSamples,
+ final Time delayBetweenSamples,
+ final int maxStackTraceDepth,
+ final Time timeout) {
+ return requestStackTraceSample(
+ executionAttemptId,
+ sampleId,
+ numSamples,
+ delayBetweenSamples,
+ maxStackTraceDepth,
+ new ArrayList<>(numSamples),
+ new CompletableFuture<>());
+ }
+
+ private CompletableFuture<StackTraceSampleResponse>
requestStackTraceSample(
+ final ExecutionAttemptID executionAttemptId,
+ final int sampleId,
+ final int numSamples,
+ final Time delayBetweenSamples,
+ final int maxStackTraceDepth,
+ final List<StackTraceElement[]> currentTraces,
+ final CompletableFuture<StackTraceSampleResponse>
resultFuture) {
+
+ if (numSamples > 0) {
+ getRpcService().getScheduledExecutor().schedule(() ->
runAsync(() -> {
+ final Optional<StackTraceElement[]> stackTrace
= getStackTrace(executionAttemptId, maxStackTraceDepth);
--- End diff --
First call to `getStackTrace` should not be delayed. I think it is enough
to move this line out of the `schedule` call..
---
