• jobtracker对提交作业的初始化


    1. 通过EagerTaskInitializationListener类中的jobAdded(jobInProgress  job)方法将所提交的作业加入到要初始化的作业队列中,代码如下:
    2. public void jobAdded(JobInProgress job) {
          synchronized (jobInitQueue) {
            jobInitQueue.add(job);
            resortInitQueue();//按照时间大小递增排序
            jobInitQueue.notifyAll();
          }
        }
    3. 在初始化队列jobinitqueue中的是等待初始化的作业。通过线程池中的线程运行的jobinitmanager类代码进行初始化。代码如下:
    4. public void start() throws IOException {//EagerTaskInitializationListener类中的start方法
          this.jobInitManagerThread = new Thread(jobInitManager, "jobInitManager");
          jobInitManagerThread.setDaemon(true);//设置为守护进程
          this.jobInitManagerThread.start();
        }
    5. jobinitmanager类的具体代码如下:
    6. class JobInitManager implements Runnable {
          public void run() {
            JobInProgress job = null;
            while (true) {
              try {
                synchronized (jobInitQueue) {
                  while (jobInitQueue.isEmpty()) {
                    jobInitQueue.wait();
                  }
                  job = jobInitQueue.remove(0);//每次取出第一个
                }
                threadPool.execute(new InitJob(job));
              } catch (InterruptedException t) {
                LOG.info("JobInitManagerThread interrupted.");
                break;
              }
            }
            LOG.info("Shutting down thread pool");
            threadPool.shutdownNow();
          }
        }
    7. initjob(job)代码如下:
    8. class InitJob implements Runnable {
          private JobInProgress job;
          public InitJob(JobInProgress job) {
            this.job = job;
          }
          public void run() {
            ttm.initJob(job);//tasktrackermanager接口定义的initjob()方法,在jobtracker中有实现
          }
        }
    9. jobtracker中的initjob()方法
    10. public void initJob(JobInProgress job) {
          if (null == job) {
            LOG.info("Init on null job is not valid");
            return;
          }
          try {
            JobStatus prevStatus = (JobStatus)job.getStatus().clone();
            LOG.info("Initializing " + job.getJobID());
            job.initTasks();//调用JobInProgress中的initTasks()方法初始化job中的所有task
            // Inform the listeners if the job state has changed
            // Note : that the job will be in PREP state.
            JobStatus newStatus = (JobStatus)job.getStatus().clone();
            if (prevStatus.getRunState() != newStatus.getRunState()) {
              JobStatusChangeEvent event =
                new JobStatusChangeEvent(job, EventType.RUN_STATE_CHANGED, prevStatus,
                    newStatus);
              synchronized (JobTracker.this) {
                updateJobInProgressListeners(event);
              }
            }
          } catch (KillInterruptedException kie) {
            //   If job was killed during initialization, job state will be KILLED
            LOG.error("Job initialization interrupted: " +
                StringUtils.stringifyException(kie));
            killJob(job);
          } catch (Throwable t) {
            String failureInfo =
              "Job initialization failed: " + StringUtils.stringifyException(t);
            // If the job initialization is failed, job state will be FAILED
            LOG.error(failureInfo);
            job.getStatus().setFailureInfo(failureInfo);
            failJob(job);
          }
           }
    11. JobInProgress中的initTasks()方法中主要代码如下:
    12.  public synchronized void initTasks()
        numMapTasks = splits.length;

          // Sanity check the locations so we don't create/initialize unnecessary tasks
          for (TaskSplitMetaInfo split : splits) {
            NetUtils.verifyHostnames(split.getLocations());
          }
          
          jobtracker.getInstrumentation().addWaitingMaps(getJobID(), numMapTasks);
          jobtracker.getInstrumentation().addWaitingReduces(getJobID(), numReduceTasks);
          this.queueMetrics.addWaitingMaps(getJobID(), numMapTasks);
          this.queueMetrics.addWaitingReduces(getJobID(), numReduceTasks);

          maps = new TaskInProgress[numMapTasks];
          for(int i=0; i < numMapTasks; ++i) {
            inputLength += splits[i].getInputDataLength();
            maps[i] = new TaskInProgress(jobId, jobFile,
                                         splits[i],
                                         jobtracker, conf, this, i, numSlotsPerMap);
          }
          LOG.info("Input size for job " + jobId + " = " + inputLength
              + ". Number of splits = " + splits.length);

          // Set localityWaitFactor before creating cache
          localityWaitFactor =
            conf.getFloat(LOCALITY_WAIT_FACTOR, DEFAULT_LOCALITY_WAIT_FACTOR);
          if (numMapTasks > 0) {
            nonRunningMapCache = createCache(splits, maxLevel);
          }

          // set the launch time
          this.launchTime = jobtracker.getClock().getTime();

          //
          // Create reduce tasks
          //
          this.reduces = new TaskInProgress[numReduceTasks];
          for (int i = 0; i < numReduceTasks; i++) {
            reduces[i] = new TaskInProgress(jobId, jobFile,
                                            numMapTasks, i,
                                            jobtracker, conf, this, numSlotsPerReduce);
            nonRunningReduces.add(reduces[i]);
          }

          // Calculate the minimum number of maps to be complete before
          // we should start scheduling reduces
          completedMapsForReduceSlowstart =
            (int)Math.ceil(
                (conf.getFloat("mapred.reduce.slowstart.completed.maps",
                               DEFAULT_COMPLETED_MAPS_PERCENT_FOR_REDUCE_SLOWSTART) *
                 numMapTasks));
          
          // ... use the same for estimating the total output of all maps
          resourceEstimator.setThreshhold(completedMapsForReduceSlowstart);
          
          // create cleanup two cleanup tips, one map and one reduce.
          cleanup = new TaskInProgress[2];

          // cleanup map tip. This map doesn't use any splits. Just assign an empty
          // split.
          TaskSplitMetaInfo emptySplit = JobSplit.EMPTY_TASK_SPLIT;
          cleanup[0] = new TaskInProgress(jobId, jobFile, emptySplit,
                  jobtracker, conf, this, numMapTasks, 1);
          cleanup[0].setJobCleanupTask();

          // cleanup reduce tip.
          cleanup[1] = new TaskInProgress(jobId, jobFile, numMapTasks,
                             numReduceTasks, jobtracker, conf, this, 1);
          cleanup[1].setJobCleanupTask();

          // create two setup tips, one map and one reduce.
          setup = new TaskInProgress[2];

          // setup map tip. This map doesn't use any split. Just assign an empty
          // split.
          setup[0] = new TaskInProgress(jobId, jobFile, emptySplit,
                  jobtracker, conf, this, numMapTasks + 1, 1);
          setup[0].setJobSetupTask();

          // setup reduce tip.
          setup[1] = new TaskInProgress(jobId, jobFile, numMapTasks,
                             numReduceTasks + 1, jobtracker, conf, this, 1);
          setup[1].setJobSetupTask();
        }
  • 相关阅读:
    LCSTS自动摘要数据集
    知识图谱问答-资料
    多轮对话管理系统Opendial
    关于自动摘要
    使用tatk构建对话代理
    android http协议
    web容器,servlet容器,模型驱动
    <s:property />
    linux下部署java web项目
    @ResultPath("/WEB-INF/jsp/project/")
  • 原文地址:https://www.cnblogs.com/cxtblogs/p/5038516.html
Copyright © 2020-2023  润新知