rmetzger commented on a change in pull request #13424:
URL: https://github.com/apache/flink/pull/13424#discussion_r492636298



##########
File path: 
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/DispatcherJob.java
##########
@@ -119,11 +119,17 @@ private DispatcherJob(
                                                }
                                        });
                                } else { // failure during initialization
-                                       final Throwable strippedThrowable = 
ExceptionUtils.stripCompletionException(throwable);
+                                       JobStatus finalStatus = 
JobStatus.FAILED;
+                                       Throwable strippedThrowable = 
ExceptionUtils.stripCompletionException(throwable);
+                                       if (throwable instanceof 
JobInitializationCancelledException) {

Review comment:
       but there is no jobManagerRunner available if it's initialization failed 
exceptionally.
   
   I will add a unit test that ensures the correct exception types to avoid 
accidental changes. 

##########
File path: 
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/DispatcherJob.java
##########
@@ -165,17 +171,10 @@ private DispatcherJob(
                        if (isInitialized()) {
                                return 
getJobMasterGateway().thenCompose(jobMasterGateway -> 
jobMasterGateway.cancel(timeout));
                        } else {
-                               log.info("Cancellation during initialization 
requested for job {}. Job will be cancelled once JobManager has been 
initialized.", jobId);
+                               log.info("Cancellation during initialization 
for job {}.", jobId);

Review comment:
       Since it is quite rare to cancel jobs during initialization, I don't 
believe this log message will hurt

##########
File path: 
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/InitializingJobManagerRunner.java
##########
@@ -0,0 +1,105 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.flink.runtime.dispatcher;
+
+import org.apache.flink.api.common.JobID;
+import org.apache.flink.runtime.client.JobInitializationException;
+import org.apache.flink.runtime.concurrent.FutureUtils;
+import org.apache.flink.runtime.jobmaster.JobManagerRunner;
+import org.apache.flink.runtime.messages.Acknowledge;
+import org.apache.flink.runtime.rpc.RpcUtils;
+import org.apache.flink.util.function.SupplierWithException;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import javax.annotation.concurrent.GuardedBy;
+
+import java.util.concurrent.CompletableFuture;
+
+/**
+ * Wrapper for an initializing JobManagerRunner.
+ */
+public class InitializingJobManagerRunner {
+       private final Logger log = 
LoggerFactory.getLogger(InitializingJobManagerRunner.class);
+
+       private final CompletableFuture<JobManagerRunner> 
jobManagerRunnerFuture = new CompletableFuture<>();
+
+       private final Thread initializationRunner;
+
+       private final Object lock = new Object();
+
+       @GuardedBy("lock")
+       private CompletableFuture<Acknowledge> cancellationFuture;
+
+       public 
InitializingJobManagerRunner(SupplierWithException<JobManagerRunner, 
JobInitializationException> initializationRunnable, JobID jobID) {
+               String threadName = "jobmanager-initialization-" + jobID;
+               this.initializationRunner = new Thread(() -> {

Review comment:
       I thought I can only set the interrupt flag of a thread if I control it 
myself. Is there a way of setting the flag when using a pool?

##########
File path: 
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/InitializingJobManagerRunner.java
##########
@@ -0,0 +1,105 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.flink.runtime.dispatcher;
+
+import org.apache.flink.api.common.JobID;
+import org.apache.flink.runtime.client.JobInitializationException;
+import org.apache.flink.runtime.concurrent.FutureUtils;
+import org.apache.flink.runtime.jobmaster.JobManagerRunner;
+import org.apache.flink.runtime.messages.Acknowledge;
+import org.apache.flink.runtime.rpc.RpcUtils;
+import org.apache.flink.util.function.SupplierWithException;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import javax.annotation.concurrent.GuardedBy;
+
+import java.util.concurrent.CompletableFuture;
+
+/**
+ * Wrapper for an initializing JobManagerRunner.
+ */
+public class InitializingJobManagerRunner {
+       private final Logger log = 
LoggerFactory.getLogger(InitializingJobManagerRunner.class);
+
+       private final CompletableFuture<JobManagerRunner> 
jobManagerRunnerFuture = new CompletableFuture<>();
+
+       private final Thread initializationRunner;
+
+       private final Object lock = new Object();
+
+       @GuardedBy("lock")
+       private CompletableFuture<Acknowledge> cancellationFuture;
+
+       public 
InitializingJobManagerRunner(SupplierWithException<JobManagerRunner, 
JobInitializationException> initializationRunnable, JobID jobID) {
+               String threadName = "jobmanager-initialization-" + jobID;
+               this.initializationRunner = new Thread(() -> {
+                       log.debug("Starting JobManager initialization thread: 
{}", threadName);
+                       try {
+                               runInitialization(initializationRunnable);
+                       } catch (Throwable threadError) {
+                               log.warn("Unexpected error in thread '{}'", 
threadName, threadError);
+                       } finally {
+                               log.debug("Stopping JobManager initialization 
thread: {}", threadName);
+                       }
+               }, threadName);
+               this.initializationRunner.setDaemon(true); // this should not 
block the JVM from shutting down
+               this.initializationRunner.start();
+       }
+
+       private void runInitialization(SupplierWithException<JobManagerRunner, 
JobInitializationException> initializationRunnable) {
+               try {
+                       JobManagerRunner runner = initializationRunnable.get();
+                       synchronized (lock) {
+                               if (cancellationFuture == null) {
+                                       jobManagerRunnerFuture.complete(runner);
+                               } else {
+                                       // jobMaster has finished despite the 
interruption. Cancel job.
+                                       
FutureUtils.forward(runner.getJobMasterGateway().thenCompose(gateway -> 
gateway.cancel(RpcUtils.INF_TIMEOUT)),
+                                               cancellationFuture);
+                                       jobManagerRunnerFuture.complete(runner);
+                               }
+                       }
+               } catch (JobInitializationException throwable) {
+                       synchronized (lock) {
+                               if (cancellationFuture == null) {
+                                       
jobManagerRunnerFuture.completeExceptionally(throwable);
+                               } else {
+                                       // an exception related to an 
interruption is expected here.
+                                       log.debug("Cancellation finished with 
(expected) exception", throwable);
+                                       
jobManagerRunnerFuture.completeExceptionally(new 
JobInitializationCancelledException("Job initialization has been cancelled", 
throwable));
+                                       
cancellationFuture.complete(Acknowledge.get());
+                               }
+                       }
+               }
+       }
+
+       public CompletableFuture<Acknowledge> cancelInitialization() {
+               synchronized (lock) {
+                       initializationRunner.interrupt();

Review comment:
       The thread will also stop if it enters a (properly implemented) blocking 
operation.
   
   The classical case we are targeting is that a remote system (triggered by 
initializeOnMaster(), or some state restore operation) is not available. The 
code will be in blocking operations all the time (waiting for network, 
sleeping, ...).
   The bigger question is whether the thread is cooperative in reacting to the 
interrupt. Often, interrupt exceptions are ignored. 




----------------------------------------------------------------
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]


Reply via email to