Hadoop TaskTracker 自我检测机制

Hadoop TaskTracker 自我检测机制
机制：通过TaskTracker在启动时循环检测，设计时通过捕获异常来完成状态检测
|-->TaskTracker.run() 
   |-->initialize();
     |-->if (shouldStartHealthMonitor(this.fConf))
        |-->startHealthMonitor(this.fConf);
          |-->healthChecker = new NodeHealthCheckerService(conf);
               |-->initialize(conf);
                    |-->this.nodeHealthScript = conf.get(HEALTH_CHECK_SCRIPT_PROPERTY); |mapred.healthChecker.script.path
                    |-->timer = new NodeHealthMonitorExecutor(args);  |初始化执行器
          |-->healthChecker.start();
               |-->if (!shouldRun(conf))  |判断是否存在自我检测
               |-->nodeHealthScriptScheduler = new Timer("NodeHealthMonitor-Timer", true);
               |-->nodeHealthScriptScheduler.scheduleAtFixedRate(timer, 0, intervalTime);  |执行如下

|-->NodeHealthMonitorExecutor
     |-->construct() |构造器
          |-->execScript.add(nodeHealthScript);
          |-->shexec = new ShellCommandExecutor((String[]) execScript.toArray(new String[execScript.size()]), null, null, scriptTimeout);
     |-->run() |Executor线程执行
          |-->HealthCheckerExitStatus status = HealthCheckerExitStatus.SUCCESS; |如无异常则表示状态正常
          |-->shexec.execute();  |定时器执行健康检查脚本
          |-->Exception |设计时以Exception的处理确定检查状态
          |-->reportHealthStatus(status);