Re: [PR] [FLINK-34227][runtime] Fixes concurrency issue in JobMaster shutdown [flink]

["zentol (via GitHub)" <gi...@apache.org>]

zentol commented on code in PR #24489:
URL: https://github.com/apache/flink/pull/24489#discussion_r1941204106


##########
flink-runtime/src/test/java/org/apache/flink/runtime/jobmaster/JobMasterTest.java:
##########
@@ -2097,6 +2098,87 @@ public void testResourceRequirementsAreRequestedFromTheScheduler() throws Except
         }
     }
 
+    @Test
+    public void testJobMasterDoesNotReconnectToResourceManagerEvenIfCleanupStalls()
+            throws Exception {
+
+        // the ResourceManager should count the connect attempts
+        final TestingResourceManagerGateway resourceManagerGateway =
+                createAndRegisterTestingResourceManagerGateway();
+        final AtomicInteger connectCount = new AtomicInteger();
+        final OneShotLatch firstRegistrationLatch = new OneShotLatch();
+        resourceManagerGateway.setRegisterJobManagerFunction(
+                (jobMasterId, resourceID, s, jobID) -> {
+                    connectCount.incrementAndGet();
+                    firstRegistrationLatch.trigger();
+
+                    return CompletableFuture.completedFuture(
+                            resourceManagerGateway.getJobMasterRegistrationSuccess());
+                });
+
+        // the scheduler close logic should be handled outside the main thread
+        final OneShotLatch schedulerCloseLatch = new OneShotLatch();
+        final CompletableFuture<Void> schedulerCloseFuture = new CompletableFuture<>();
+        final TestingSchedulerNG scheduler =
+                TestingSchedulerNG.newBuilder()
+                        .setCloseAsyncSupplier(
+                                () -> {
+                                    schedulerCloseLatch.trigger();
+                                    return schedulerCloseFuture;
+                                })
+                        .build();
+
+        final OneShotLatch closeSlotPoolLatch = new OneShotLatch();
+        final JobMaster jobMaster =
+                new JobMasterBuilder(jobGraph, rpcService)
+                        .withHighAvailabilityServices(haServices)
+                        .withSlotPoolServiceSchedulerFactory(
+                                DefaultSlotPoolServiceSchedulerFactory.create(
+                                        TestingSlotPoolServiceBuilder.newBuilder()
+                                                .setCloseRunnable(closeSlotPoolLatch::awaitQuietly),
+                                        new TestingSchedulerNGFactory(scheduler)))
+                        .createJobMaster();
+        jobMaster.start();
+
+        notifyResourceManagerLeaderListeners(resourceManagerGateway);
+        firstRegistrationLatch.await();
+
+        final CompletableFuture<Void> jobMasterCloseFuture = jobMaster.closeAsync();
+
+        // force the scheduler closing to happen outside the main thread

Review Comment:
   we should keep both bits; the scheduler ideally doesnt complete things in the IO executor, but the JobMaster needs safeguards in case it does (or maybe even fail hard if it is violated? 🤔 ).



-- 
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: issues-unsubscribe@flink.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org