[GitHub] [samza] alnzng commented on a change in pull request #1529: SAMZA-2685: Add job coordinator which does not do resource management

[GitBox <gi...@apache.org>]

alnzng commented on a change in pull request #1529:
URL: https://github.com/apache/samza/pull/1529#discussion_r709584553



##########
File path: samza-core/src/main/java/org/apache/samza/coordinator/StaticResourceJobCoordinator.java
##########
@@ -0,0 +1,232 @@
+/*
+ * 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.samza.coordinator;
+
+import java.io.IOException;
+import java.util.Set;
+import java.util.concurrent.CountDownLatch;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicBoolean;
+import com.google.common.annotations.VisibleForTesting;
+import org.apache.samza.SamzaException;
+import org.apache.samza.config.Config;
+import org.apache.samza.config.JobConfig;
+import org.apache.samza.container.LocalityManager;
+import org.apache.samza.container.grouper.task.TaskAssignmentManager;
+import org.apache.samza.container.grouper.task.TaskPartitionAssignmentManager;
+import org.apache.samza.coordinator.communication.CoordinatorCommunication;
+import org.apache.samza.coordinator.communication.CoordinatorCommunicationContext;
+import org.apache.samza.coordinator.communication.HttpCoordinatorToWorkerCommunicationFactory;
+import org.apache.samza.coordinator.communication.JobModelServingContext;
+import org.apache.samza.coordinator.metadatastore.NamespaceAwareCoordinatorStreamStore;
+import org.apache.samza.coordinator.stream.messages.SetChangelogMapping;
+import org.apache.samza.coordinator.stream.messages.SetContainerHostMapping;
+import org.apache.samza.coordinator.stream.messages.SetJobCoordinatorMetadataMessage;
+import org.apache.samza.coordinator.stream.messages.SetTaskContainerMapping;
+import org.apache.samza.coordinator.stream.messages.SetTaskModeMapping;
+import org.apache.samza.coordinator.stream.messages.SetTaskPartitionMapping;
+import org.apache.samza.job.JobCoordinatorMetadata;
+import org.apache.samza.job.JobMetadataChange;
+import org.apache.samza.job.metadata.JobCoordinatorMetadataManager;
+import org.apache.samza.job.model.JobModel;
+import org.apache.samza.job.model.JobModelUtil;
+import org.apache.samza.metadatastore.MetadataStore;
+import org.apache.samza.metrics.MetricsRegistry;
+import org.apache.samza.startpoint.StartpointManager;
+import org.apache.samza.storage.ChangelogStreamManager;
+import org.apache.samza.system.StreamMetadataCache;
+import org.apache.samza.system.SystemAdmins;
+import org.apache.samza.util.SystemClock;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+
+/**
+ * Handles coordination with the workers for a Samza job. This includes generating the job model, managing metadata,
+ * and coordinating with the workers when the job model changes.
+ * This does certain similar things as the older ClusterBasedJobCoordinator, but one notable difference is that this
+ * coordinator does no management of execution resources. It relies on an external component to manage those resources
+ * for a Samza job.
+ */
+public class StaticResourceJobCoordinator {
+  private static final Logger LOG = LoggerFactory.getLogger(StaticResourceJobCoordinator.class);
+
+  private final JobModelHelper jobModelHelper;
+  private final JobModelServingContext jobModelServingContext;
+  private final CoordinatorCommunication coordinatorCommunication;
+  private final JobCoordinatorMetadataManager jobCoordinatorMetadataManager;
+  /**
+   * This can be null if startpoints are not enabled.
+   */
+  private final StartpointManager startpointManager;
+  private final ChangelogStreamManager changelogStreamManager;
+  private final MetricsRegistry metrics;
+  private final SystemAdmins systemAdmins;
+  private final Config config;
+
+  private final AtomicBoolean isStarted = new AtomicBoolean(false);
+  private final AtomicBoolean shouldShutdown = new AtomicBoolean(false);
+  private final CountDownLatch shutdownLatch = new CountDownLatch(1);
+
+  public static StaticResourceJobCoordinator build(MetricsRegistry metrics, MetadataStore metadataStore,

Review comment:
       Out of curiosity, any specific reason you prefer to use a static build method to initialize instead of using a constructor?

##########
File path: samza-core/src/main/java/org/apache/samza/clustermanager/JobCoordinatorLaunchUtil.java
##########
@@ -76,11 +85,37 @@ public static void run(SamzaApplication app, Config config) {
     CoordinatorStreamUtil.writeConfigToCoordinatorStream(finalConfig, true);
     DiagnosticsUtil.createDiagnosticsStream(finalConfig);
 
-    ClusterBasedJobCoordinator jc = new ClusterBasedJobCoordinator(
-        metrics,
-        metadataStore,
-        finalConfig);
-    jc.run();
+    if (new JobCoordinatorConfig(finalConfig).getUseStaticResourceJobCoordinator()) {
+      runStaticResourceJobCoordinator(metrics, metadataStore, finalConfig);
+    } else {
+      ClusterBasedJobCoordinator jc = new ClusterBasedJobCoordinator(metrics, metadataStore, finalConfig);
+      jc.run();
+    }
+  }
+
+  private static void runStaticResourceJobCoordinator(MetricsRegistryMap metrics, MetadataStore metadataStore,
+      Config finalConfig) {
+    StaticResourceJobCoordinator staticResourceJobCoordinator =
+        StaticResourceJobCoordinator.build(metrics, metadataStore, finalConfig);
+    addShutdownHook(staticResourceJobCoordinator);
+    Map<String, MetricsReporter> metricsReporters =
+        MetricsReporterLoader.getMetricsReporters(new MetricsConfig(finalConfig), JOB_COORDINATOR_CONTAINER_NAME);
+    metricsReporters.values()
+        .forEach(metricsReporter -> metricsReporter.register(JOB_COORDINATOR_CONTAINER_NAME, metrics));
+    metricsReporters.values().forEach(MetricsReporter::start);
+    staticResourceJobCoordinator.run();
+    metricsReporters.values().forEach(MetricsReporter::stop);

Review comment:
       Any specific reason why we don't put metric reporter related logic inside `StaticResourceJobCoordinator`?

##########
File path: samza-core/src/main/java/org/apache/samza/config/JobCoordinatorConfig.java
##########
@@ -77,4 +79,8 @@ public String getJobCoordinatorFactoryClassName() {
       return jobCoordinatorFactoryClassName;
     }
   }
+
+  public boolean getUseStaticResourceJobCoordinator() {

Review comment:
       Would you think the method name `isStaticResourceJobCoordinatorUsed`/ `isStaticResourceJobCoordinatorEnabled` more readable? 

##########
File path: samza-core/src/main/java/org/apache/samza/clustermanager/JobCoordinatorLaunchUtil.java
##########
@@ -76,11 +85,37 @@ public static void run(SamzaApplication app, Config config) {
     CoordinatorStreamUtil.writeConfigToCoordinatorStream(finalConfig, true);
     DiagnosticsUtil.createDiagnosticsStream(finalConfig);
 
-    ClusterBasedJobCoordinator jc = new ClusterBasedJobCoordinator(
-        metrics,
-        metadataStore,
-        finalConfig);
-    jc.run();
+    if (new JobCoordinatorConfig(finalConfig).getUseStaticResourceJobCoordinator()) {
+      runStaticResourceJobCoordinator(metrics, metadataStore, finalConfig);
+    } else {
+      ClusterBasedJobCoordinator jc = new ClusterBasedJobCoordinator(metrics, metadataStore, finalConfig);
+      jc.run();
+    }
+  }
+
+  private static void runStaticResourceJobCoordinator(MetricsRegistryMap metrics, MetadataStore metadataStore,
+      Config finalConfig) {
+    StaticResourceJobCoordinator staticResourceJobCoordinator =
+        StaticResourceJobCoordinator.build(metrics, metadataStore, finalConfig);
+    addShutdownHook(staticResourceJobCoordinator);
+    Map<String, MetricsReporter> metricsReporters =
+        MetricsReporterLoader.getMetricsReporters(new MetricsConfig(finalConfig), JOB_COORDINATOR_CONTAINER_NAME);
+    metricsReporters.values()
+        .forEach(metricsReporter -> metricsReporter.register(JOB_COORDINATOR_CONTAINER_NAME, metrics));
+    metricsReporters.values().forEach(MetricsReporter::start);
+    staticResourceJobCoordinator.run();
+    metricsReporters.values().forEach(MetricsReporter::stop);
+  }
+
+  /**
+   * This is a separate method so it can be stubbed in tests, since adding a real shutdown hook will cause the hook to
+   * added to the test suite JVM.
+   */
+  @VisibleForTesting
+  static void addShutdownHook(StaticResourceJobCoordinator staticResourceJobCoordinator) {

Review comment:
       Any problem if we apply this shutdown hook function to `ClusterBasedJobCoordinator` as well?

##########
File path: samza-core/src/main/java/org/apache/samza/coordinator/communication/HttpCoordinatorToWorkerCommunicationFactory.java
##########
@@ -0,0 +1,41 @@
+/*
+ * 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.samza.coordinator.communication;
+
+import org.apache.samza.config.ClusterManagerConfig;
+import org.apache.samza.coordinator.server.HttpServer;
+import org.eclipse.jetty.servlet.DefaultServlet;
+import org.eclipse.jetty.servlet.ServletHolder;
+
+
+/**
+ * Implements HTTP-based communication between the coordinator and workers.
+ */
+public class HttpCoordinatorToWorkerCommunicationFactory implements CoordinatorToWorkerCommunicationFactory {
+  @Override
+  public CoordinatorCommunication coordinatorCommunication(CoordinatorCommunicationContext context) {
+    ClusterManagerConfig clusterManagerConfig = new ClusterManagerConfig(context.getConfig());
+    JobModelHttpServlet jobModelHttpServlet = new JobModelHttpServlet(context.getJobModelProvider(),
+        new JobModelHttpServlet.Metrics(context.getMetricsRegistry()));
+    HttpServer httpServer = new HttpServer("/", clusterManagerConfig.getCoordinatorUrlPort(), null,
+        new ServletHolder(DefaultServlet.class));
+    httpServer.addServlet("/", jobModelHttpServlet);
+    return new HttpCoordinatorCommunication(httpServer);

Review comment:
       What are your thoughts/plans for the job coordinator readiness probe support? Will we leverage this embedded HTTP server?




-- 
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: commits-unsubscribe@samza.apache.org

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