Package org.apache.hadoop.yarn.server.api.records

Examples of org.apache.hadoop.yarn.server.api.records.HeartbeatResponse

99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
  @Test(timeout = 500)
  public void testNodeHeartbeatResponsePBImpl() {
    NodeHeartbeatResponsePBImpl original = new NodeHeartbeatResponsePBImpl();

    HeartbeatResponse response = recordFactory
        .newRecordInstance(HeartbeatResponse.class);
    response.setMasterKey(getMasterKey());
    response.setNodeAction(NodeAction.REBOOT);
    response.setResponseId(1);
    original.setHeartbeatResponse(response);

    NodeHeartbeatResponsePBImpl copy = new NodeHeartbeatResponsePBImpl(
        original.getProto());
    assertEquals(NodeAction.REBOOT, copy.getHeartbeatResponse().getNodeAction());
View Full Code Here
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
   
    ClusterMetrics metrics = ClusterMetrics.getMetrics();
    assert(metrics != null);
    int metricCount = metrics.getNumDecommisionedNMs();

    HeartbeatResponse nodeHeartbeat = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    nodeHeartbeat = nm2.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    nodeHeartbeat = nm3.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));

    // To test that IPs also work
    String ip = NetUtils.normalizeHostName("localhost");
    writeToHostsFile("host1", ip);

    rm.getNodesListManager().refreshNodes();

    nodeHeartbeat = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    Assert
        .assertEquals(0, ClusterMetrics.getMetrics().getNumDecommisionedNMs());

    nodeHeartbeat = nm2.nodeHeartbeat(true);
    Assert.assertTrue("Node is not decommisioned.", NodeAction.SHUTDOWN
        .equals(nodeHeartbeat.getNodeAction()));

    checkDecommissionedNMCount(rm, ++metricCount);

    nodeHeartbeat = nm3.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    Assert.assertEquals(metricCount, ClusterMetrics.getMetrics()
      .getNumDecommisionedNMs());
  }
View Full Code Here
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
            request.setNodeStatus(nodeStatus);
            if (isSecurityEnabled()) {
              request.setLastKnownMasterKey(NodeStatusUpdaterImpl.this.context
                .getContainerTokenSecretManager().getCurrentKey());
            }
            HeartbeatResponse response =
              resourceTracker.nodeHeartbeat(request).getHeartbeatResponse();

            // See if the master-key has rolled over
            if (isSecurityEnabled()) {
              MasterKey updatedMasterKey = response.getMasterKey();
              if (updatedMasterKey != null) {
                // Will be non-null only on roll-over on RM side
                context.getContainerTokenSecretManager().setMasterKey(
                  updatedMasterKey);
              }
            }

            if (response.getNodeAction() == NodeAction.SHUTDOWN) {
              LOG
                  .info("Recieved SHUTDOWN signal from Resourcemanager as part of heartbeat," +
                      " hence shutting down.");
              NodeStatusUpdaterImpl.this.stop();
              break;
            }
            if (response.getNodeAction() == NodeAction.REBOOT) {
              LOG.info("Node is out of sync with ResourceManager,"
                  + " hence rebooting.");
              NodeStatusUpdaterImpl.this.reboot();
              break;
            }

            lastHeartBeatID = response.getResponseId();
            List<ContainerId> containersToCleanup = response
                .getContainersToCleanupList();
            if (containersToCleanup.size() != 0) {
              dispatcher.getEventHandler().handle(
                  new CMgrCompletedContainersEvent(containersToCleanup));
            }
            List<ApplicationId> appsToCleanup =
                response.getApplicationsToCleanupList();
            //Only start tracking for keepAlive on FINISH_APP
            trackAppsForKeepAlive(appsToCleanup);
            if (appsToCleanup.size() != 0) {
              dispatcher.getEventHandler().handle(
                  new CMgrCompletedAppsEvent(appsToCleanup));
View Full Code Here
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
    MockNM nm2 = rm.registerNode("host2:5678", 10240);
    MockNM nm3 = rm.registerNode("localhost:4433", 1024);

    int metricCount = ClusterMetrics.getMetrics().getNumDecommisionedNMs();

    HeartbeatResponse nodeHeartbeat = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    nodeHeartbeat = nm2.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));

    // To test that IPs also work
    String ip = NetUtils.normalizeHostName("localhost");
    writeToHostsFile("host2", ip);

    rm.getNodesListManager().refreshNodes();

    nodeHeartbeat = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));
    nodeHeartbeat = nm2.nodeHeartbeat(true);
    Assert.assertTrue("The decommisioned metrics are not updated",
        NodeAction.SHUTDOWN.equals(nodeHeartbeat.getNodeAction()));
    checkDecommissionedNMCount(rm, ++metricCount);

    nodeHeartbeat = nm3.nodeHeartbeat(true);
    Assert.assertTrue("The decommisioned metrics are not updated",
        NodeAction.SHUTDOWN.equals(nodeHeartbeat.getNodeAction()));
    checkDecommissionedNMCount(rm, ++metricCount);
  }
View Full Code Here
178
179
180
181
182
183
184
185
186
187
188
189
190
    MockNM nm1 = rm.registerNode("host1:1234", 5120);
    MockNM nm2 = rm.registerNode("host2:1234", 2048);

    int initialMetricCount = ClusterMetrics.getMetrics().getNumRebootedNMs();
    HeartbeatResponse nodeHeartbeat = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(nodeHeartbeat.getNodeAction()));

    nodeHeartbeat = nm2.nodeHeartbeat(
      new HashMap<ApplicationId, List<ContainerStatus>>(), true, -100);
    Assert.assertTrue(NodeAction.REBOOT.equals(nodeHeartbeat.getNodeAction()));
    checkRebootedNMCount(rm, ++initialMetricCount);
  }
View Full Code Here
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
    // TODO Metrics incorrect in case of the FifoScheduler
    Assert.assertEquals(5120, metrics.getAvailableMB());

    // reconnect of healthy node
    nm1 = rm.registerNode("host1:1234", 5120);
    HeartbeatResponse response = nm1.nodeHeartbeat(true);
    Assert.assertTrue(NodeAction.NORMAL.equals(response.getNodeAction()));
    dispatcher.await();
    Assert.assertEquals(expectedNMs, ClusterMetrics.getMetrics().getNumActiveNMs());
    checkUnealthyNMCount(rm, nm2, true, 1);

    // reconnect of unhealthy node
    nm2 = rm.registerNode("host2:5678", 5120);
    response = nm2.nodeHeartbeat(false);
    Assert.assertTrue(NodeAction.NORMAL.equals(response.getNodeAction()));
    dispatcher.await();
    Assert.assertEquals(expectedNMs, ClusterMetrics.getMetrics().getNumActiveNMs());
    checkUnealthyNMCount(rm, nm2, true, 1);

    // reconnect of node with changed capability
    nm1 = rm.registerNode("host2:5678", 10240);
    dispatcher.await();
    response = nm2.nodeHeartbeat(true);
    dispatcher.await();
    Assert.assertTrue(NodeAction.NORMAL.equals(response.getNodeAction()));
    Assert.assertEquals(5120 + 10240, metrics.getAvailableMB());
  }
View Full Code Here
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
    NodeHeartbeatRequest nodeHeartBeatRequest = recordFactory
        .newRecordInstance(NodeHeartbeatRequest.class);
    nodeHeartBeatRequest.setNodeStatus(nodeStatus);

    nodeStatus.setResponseId(0);
    HeartbeatResponse response = resourceTrackerService.nodeHeartbeat(
        nodeHeartBeatRequest).getHeartbeatResponse();
    Assert.assertTrue(response.getResponseId() == 1);

    nodeStatus.setResponseId(response.getResponseId());
    response = resourceTrackerService.nodeHeartbeat(nodeHeartBeatRequest)
        .getHeartbeatResponse();
    Assert.assertTrue(response.getResponseId() == 2);  

    /* try calling with less response id */
    response = resourceTrackerService.nodeHeartbeat(nodeHeartBeatRequest)
        .getHeartbeatResponse();
    Assert.assertTrue(response.getResponseId() == 2);

    nodeStatus.setResponseId(0);
    response = resourceTrackerService.nodeHeartbeat(nodeHeartBeatRequest)
        .getHeartbeatResponse();
    Assert.assertTrue(NodeAction.REBOOT.equals(response.getNodeAction()));
  }
View Full Code Here
147
148
149
150
151
152
153
154
155
          nodeId, getContainerStatuses(containers));
    nodeStatus.setResponseId(responseID);
    NodeHeartbeatRequest request = recordFactory
        .newRecordInstance(NodeHeartbeatRequest.class);
    request.setNodeStatus(nodeStatus);
    HeartbeatResponse response = resourceTrackerService
        .nodeHeartbeat(request).getHeartbeatResponse();
    responseID = response.getResponseId();
  }
View Full Code Here
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
      Thread.sleep(2000);
    }
    Assert.assertEquals(request, conts.size());
   
    am.unregisterAppAttempt();
    HeartbeatResponse resp = nm1.nodeHeartbeat(attempt.getAppAttemptId(), 1,
        ContainerState.COMPLETE);
    am.waitForState(RMAppAttemptState.FINISHED);

    int cleanedConts = 0;
    int cleanedApps = 0;
    List<ContainerId> contsToClean = null;
    List<ApplicationId> apps = null;
   
    //currently only containers are cleaned via this
    //AM container is cleaned via container launcher
    waitCount = 0;
    while ((cleanedConts < 2 || cleanedApps < 1) && waitCount++ < 20) {
      contsToClean = resp.getContainersToCleanupList();
      apps = resp.getApplicationsToCleanupList();
      LOG.info("Waiting to get cleanup events.. cleanedConts: "
          + cleanedConts + " cleanedApps: " + cleanedApps);
      cleanedConts += contsToClean.size();
      cleanedApps += apps.size();
      Thread.sleep(1000);
View Full Code Here
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
        new ArrayList<ContainerStatus>();
    containerStatusList.add(BuilderUtils.newContainerStatus(conts.get(1)
      .getId(), ContainerState.RUNNING, "nothing", 0));
    containerStatuses.put(app.getApplicationId(), containerStatusList);

    HeartbeatResponse resp = nm1.nodeHeartbeat(containerStatuses, true);
    dispatcher.await();
    List<ContainerId> contsToClean = resp.getContainersToCleanupList();
    int cleanedConts = contsToClean.size();
    waitCount = 0;
    while (cleanedConts < 1 && waitCount++ < 20) {
      resp = nm1.nodeHeartbeat(true);
      dispatcher.await();
      contsToClean = resp.getContainersToCleanupList();
      LOG.info("Waiting to get cleanup events.. cleanedConts: " + cleanedConts);
      cleanedConts += contsToClean.size();
      Thread.sleep(1000);
    }
    LOG.info("Got cleanup for " + contsToClean.get(0));
    Assert.assertEquals(1, cleanedConts);

    // Now to test the case when RM already gave cleanup, and NM suddenly
    // realizes that the container is running.
    LOG.info("Testing container launch much after release and "
        + "NM getting cleanup");
    containerStatuses.clear();
    containerStatusList.clear();
    containerStatusList.add(BuilderUtils.newContainerStatus(conts.get(1)
      .getId(), ContainerState.RUNNING, "nothing", 0));
    containerStatuses.put(app.getApplicationId(), containerStatusList);

    resp = nm1.nodeHeartbeat(containerStatuses, true);
    dispatcher.await();
    contsToClean = resp.getContainersToCleanupList();
    cleanedConts = contsToClean.size();
    // The cleanup list won't be instantaneous as it is given out by scheduler
    // and not RMNodeImpl.
    waitCount = 0;
    while (cleanedConts < 1 && waitCount++ < 20) {
      resp = nm1.nodeHeartbeat(true);
      dispatcher.await();
      contsToClean = resp.getContainersToCleanupList();
      LOG.info("Waiting to get cleanup events.. cleanedConts: " + cleanedConts);
      cleanedConts += contsToClean.size();
      Thread.sleep(1000);
    }
    LOG.info("Got cleanup for " + contsToClean.get(0));
View Full Code Here
TOP

Related Classes of org.apache.hadoop.yarn.server.api.records.HeartbeatResponse

  • org.apache.hadoop.yarn.server.nodemanager.MockNodeStatusUpdater$MockResourceTracker
  • org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl
  • org.apache.hadoop.yarn.server.nodemanager.TestNodeStatusUpdater$MyResourceTracker
  • org.apache.hadoop.yarn.server.nodemanager.TestNodeStatusUpdater$MyResourceTracker2
  • org.apache.hadoop.yarn.server.nodemanager.TestNodeStatusUpdater$MyResourceTracker3
  • org.apache.hadoop.yarn.server.resourcemanager.MockNM
  • org.apache.hadoop.yarn.server.resourcemanager.NodeManager
  • org.apache.hadoop.yarn.server.resourcemanager.resourcetracker.TestRMNMRPCResponseId
  • org.apache.hadoop.yarn.server.resourcemanager.ResourceTrackerService
  • org.apache.hadoop.yarn.server.resourcemanager.TestApplicationCleanup
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.