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

Examples of org.apache.hadoop.yarn.server.api.protocolrecords.NodeHeartbeatResponse

151
152
153
154
155
156
157
158
159
          nodeId, getContainerStatuses(containers));
    nodeStatus.setResponseId(responseID);
    NodeHeartbeatRequest request = recordFactory
        .newRecordInstance(NodeHeartbeatRequest.class);
    request.setNodeStatus(nodeStatus);
    NodeHeartbeatResponse response = resourceTrackerService
        .nodeHeartbeat(request);
    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
115
116
117
118
      contReceived += conts.size();
    }
    Assert.assertEquals(request, contReceived);
   
    am.unregisterAppAttempt();
    NodeHeartbeatResponse resp = nm1.nodeHeartbeat(attempt.getAppAttemptId(), 1,
        ContainerState.COMPLETE);
    am.waitForState(RMAppAttemptState.FINISHED);

    //currently only containers are cleaned via this
    //AM container is cleaned via container launcher
    resp = nm1.nodeHeartbeat(true);
    List<ContainerId> contsToClean = resp.getContainersToCleanup();
    List<ApplicationId> apps = resp.getApplicationsToCleanup();
    int cleanedConts = contsToClean.size();
    int cleanedApps = apps.size();
    waitCount = 0;
    while ((cleanedConts < 2 || cleanedApps < 1) && waitCount++ < 200) {
      LOG.info("Waiting to get cleanup events.. cleanedConts: "
          + cleanedConts + " cleanedApps: " + cleanedApps);
      Thread.sleep(100);
      resp = nm1.nodeHeartbeat(true);
      contsToClean = resp.getContainersToCleanup();
      apps = resp.getApplicationsToCleanup();
      cleanedConts += contsToClean.size();
      cleanedApps += apps.size();
    }
   
    Assert.assertEquals(1, apps.size());
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
243
244
        new ArrayList<ContainerStatus>();
    containerStatusList.add(BuilderUtils.newContainerStatus(conts.get(0)
      .getId(), ContainerState.RUNNING, "nothing", 0));
    containerStatuses.put(app.getApplicationId(), containerStatusList);

    NodeHeartbeatResponse resp = nm1.nodeHeartbeat(containerStatuses, true);
    dispatcher.await();
    List<ContainerId> contsToClean = resp.getContainersToCleanup();
    int cleanedConts = contsToClean.size();
    waitCount = 0;
    while (cleanedConts < 1 && waitCount++ < 200) {
      LOG.info("Waiting to get cleanup events.. cleanedConts: " + cleanedConts);
      Thread.sleep(100);
      resp = nm1.nodeHeartbeat(true);
      dispatcher.await();
      contsToClean = resp.getContainersToCleanup();
      cleanedConts += contsToClean.size();
    }
    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(0)
      .getId(), ContainerState.RUNNING, "nothing", 0));
    containerStatuses.put(app.getApplicationId(), containerStatusList);

    resp = nm1.nodeHeartbeat(containerStatuses, true);
    dispatcher.await();
    contsToClean = resp.getContainersToCleanup();
    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++ < 200) {
      LOG.info("Waiting to get cleanup events.. cleanedConts: " + cleanedConts);
      Thread.sleep(100);
      resp = nm1.nodeHeartbeat(true);
      dispatcher.await();
      contsToClean = resp.getContainersToCleanup();
      cleanedConts += contsToClean.size();
    }
    LOG.info("Got cleanup for " + contsToClean.get(0));
    Assert.assertEquals(1, cleanedConts);
View Full Code Here
273
274
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
302
303
304
          new RMNodeEvent(nodeId, RMNodeEventType.DECOMMISSION));
      return shutDown;
    }
   
    // 3. Check if it's a 'fresh' heartbeat i.e. not duplicate heartbeat
    NodeHeartbeatResponse lastNodeHeartbeatResponse = rmNode.getLastNodeHeartBeatResponse();
    if (remoteNodeStatus.getResponseId() + 1 == lastNodeHeartbeatResponse
        .getResponseId()) {
      LOG.info("Received duplicate heartbeat from node "
          + rmNode.getNodeAddress());
      return lastNodeHeartbeatResponse;
    } else if (remoteNodeStatus.getResponseId() + 1 < lastNodeHeartbeatResponse
        .getResponseId()) {
      String message =
          "Too far behind rm response id:"
              + lastNodeHeartbeatResponse.getResponseId() + " nm response id:"
              + remoteNodeStatus.getResponseId();
      LOG.info(message);
      resync.setDiagnosticsMessage(message);
      // TODO: Just sending reboot is not enough. Think more.
      this.rmContext.getDispatcher().getEventHandler().handle(
          new RMNodeEvent(nodeId, RMNodeEventType.REBOOTING));
      return resync;
    }

    // Heartbeat response
    NodeHeartbeatResponse nodeHeartBeatResponse = YarnServerBuilderUtils
        .newNodeHeartbeatResponse(lastNodeHeartbeatResponse.
            getResponseId() + 1, NodeAction.NORMAL, null, null, null, null,
            nextHeartBeatInterval);
    rmNode.updateNodeHeartbeatResponseForCleanup(nodeHeartBeatResponse);
View Full Code Here
116
117
118
119
120
121
122
123
124
125
126
  @After
  public void tearDown() throws Exception {
  }
 
  private RMNodeStatusEvent getMockRMNodeStatusEvent() {
    NodeHeartbeatResponse response = mock(NodeHeartbeatResponse.class);

    NodeHealthStatus healthStatus = mock(NodeHealthStatus.class);
    Boolean yes = new Boolean(true);
    doReturn(yes).when(healthStatus).getIsNodeHealthy();
   
View Full Code Here
323
324
325
326
327
328
329
330
331
332
333
334
335
336
    // but updating heartbeat response for cleanup does
    RMNodeStatusEvent statusEvent = getMockRMNodeStatusEvent();
    node.handle(statusEvent);
    Assert.assertEquals(1, node.getContainersToCleanUp().size());
    Assert.assertEquals(1, node.getAppsToCleanup().size());
    NodeHeartbeatResponse hbrsp = Records.newRecord(NodeHeartbeatResponse.class);
    node.updateNodeHeartbeatResponseForCleanup(hbrsp);
    Assert.assertEquals(0, node.getContainersToCleanUp().size());
    Assert.assertEquals(0, node.getAppsToCleanup().size());
    Assert.assertEquals(1, hbrsp.getContainersToCleanup().size());
    Assert.assertEquals(completedContainerId, hbrsp.getContainersToCleanup().get(0));
    Assert.assertEquals(1, hbrsp.getApplicationsToCleanup().size());
    Assert.assertEquals(finishedAppId, hbrsp.getApplicationsToCleanup().get(0));
  }
View Full Code Here
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
      this.rmContext.getDispatcher().getEventHandler().handle(
          new RMNodeEvent(nodeId, RMNodeEventType.DECOMMISSION));
      return shutDown;
    }

    NodeHeartbeatResponse nodeHeartBeatResponse = recordFactory
        .newRecordInstance(NodeHeartbeatResponse.class);

    // 3. Check if it's a 'fresh' heartbeat i.e. not duplicate heartbeat
    HeartbeatResponse lastHeartbeatResponse = rmNode.getLastHeartBeatResponse();
    if (remoteNodeStatus.getResponseId() + 1 == lastHeartbeatResponse
        .getResponseId()) {
      LOG.info("Received duplicate heartbeat from node "
          + rmNode.getNodeAddress());
      nodeHeartBeatResponse.setHeartbeatResponse(lastHeartbeatResponse);
      return nodeHeartBeatResponse;
    } else if (remoteNodeStatus.getResponseId() + 1 < lastHeartbeatResponse
        .getResponseId()) {
      LOG.info("Too far behind rm response id:"
          + lastHeartbeatResponse.getResponseId() + " nm response id:"
          + remoteNodeStatus.getResponseId());
      // TODO: Just sending reboot is not enough. Think more.
      this.rmContext.getDispatcher().getEventHandler().handle(
          new RMNodeEvent(nodeId, RMNodeEventType.REBOOTING));
      return reboot;
    }

    // Heartbeat response
    HeartbeatResponse latestResponse = recordFactory
        .newRecordInstance(HeartbeatResponse.class);
    latestResponse.setResponseId(lastHeartbeatResponse.getResponseId() + 1);
    latestResponse.addAllContainersToCleanup(rmNode.getContainersToCleanUp());
    latestResponse.addAllApplicationsToCleanup(rmNode.getAppsToCleanup());
    latestResponse.setNodeAction(NodeAction.NORMAL);

    // 4. Send status to RMNode, saving the latest response.
    this.rmContext.getDispatcher().getEventHandler().handle(
        new RMNodeStatusEvent(nodeId, remoteNodeStatus.getNodeHealthStatus(),
            remoteNodeStatus.getContainersStatuses(),
            remoteNodeStatus.getKeepAliveApplications(), latestResponse));

    nodeHeartBeatResponse.setHeartbeatResponse(latestResponse);
    return nodeHeartBeatResponse;
  }
View Full Code Here
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
          return new ResourceTracker() {

            @Override
            public NodeHeartbeatResponse nodeHeartbeat(
                NodeHeartbeatRequest request) throws YarnRemoteException {
              NodeHeartbeatResponse response = recordFactory.newRecordInstance(
                  NodeHeartbeatResponse.class);
              try {
                response.setHeartbeatResponse(rt.nodeHeartbeat(request)
                    .getHeartbeatResponse());
              } catch (IOException ioe) {
                LOG.info("Exception in heartbeat from node " +
                    request.getNodeStatus().getNodeId(), ioe);
                throw RPCUtil.getRemoteException(ioe);
              }
              return response;
            }

            @Override
            public RegisterNodeManagerResponse registerNodeManager(
                RegisterNodeManagerRequest request)
                throws YarnRemoteException {
              RegisterNodeManagerResponse response = recordFactory.
                  newRecordInstance(RegisterNodeManagerResponse.class);
              try {
                response.setRegistrationResponse(rt
                    .registerNodeManager(request)
                    .getRegistrationResponse());
              } catch (IOException ioe) {
                LOG.info("Exception in node registration from "
                    + request.getNodeId().toString(), ioe);
View Full Code Here
216
217
218
219
220
221
222
223
224
225
      }
      HeartbeatResponse response = recordFactory
          .newRecordInstance(HeartbeatResponse.class);
      response.setResponseId(heartBeatID);

      NodeHeartbeatResponse nhResponse = recordFactory
          .newRecordInstance(NodeHeartbeatResponse.class);
      nhResponse.setHeartbeatResponse(response);
      return nhResponse;
    }
View Full Code Here
310
311
312
313
314
315
316
317
318
319
      HeartbeatResponse response = recordFactory
          .newRecordInstance(HeartbeatResponse.class);
      response.setResponseId(heartBeatID);
      response.setNodeAction(heartBeatNodeAction);
     
      NodeHeartbeatResponse nhResponse = recordFactory
      .newRecordInstance(NodeHeartbeatResponse.class);
      nhResponse.setHeartbeatResponse(response);
      return nhResponse;
    }
View Full Code Here
TOP

Related Classes of org.apache.hadoop.yarn.server.api.protocolrecords.NodeHeartbeatResponse

  • org.apache.hadoop.yarn.server.api.impl.pb.service.ResourceTrackerPBServiceImpl
  • org.apache.hadoop.yarn.server.MiniYARNCluster$CustomNodeManager
  • org.apache.hadoop.yarn.server.MiniYARNCluster$ShortCircuitedNodeManager
  • org.apache.hadoop.yarn.server.nodemanager.LocalRMInterface
  • 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.nodemanager.TestNodeStatusUpdater$MyResourceTracker4
    • 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.