HBase1.2.3版本HRegion的负载均衡实现源码分析

一、HRegion的负载均衡

      负载均衡主要用来解决热点问题，使请求更均匀的发送给不同的应用服务器，Hbase是一个典型的主从架构，HMaster负载整个集群的资源调度，任务分配，而数据的IO操作是在HRegionServer上，每个HRegionServer会有N个HRegion，存的是每张表的一部分数据，而HRegionServer里的HRegion之间的负载影响着整个Hbase集群的读写性能。

二、负载均衡实现

    HRegion的负载调度由HMaster负责，所以负载的实现类在HMaster类里面。先看HMaster里面用到的涉及负载的几个变量

 /**触发balacer的线程 */
private BalancerChore balancerChore;
// Tracker for load balancer state
//负载均衡的追踪类
LoadBalancerTracker loadBalancerTracker;
//负载均衡实现类
LoadBalancer balancer;

     用来实现HRegion在各个HReginServer之间的迁移的相关操作类在HMaster构造函数中进行了相关初始化工作。

 public HMaster(final Configuration conf, CoordinatedStateManager csm)
      throws IOException, KeeperException, InterruptedException {
    super(conf, csm);
    this.rsFatals = new MemoryBoundedLogMessageBuffer(
      conf.getLong("hbase.master.buffer.for.rs.fatals", 1*1024*1024));

    LOG.info("hbase.rootdir=" + FSUtils.getRootDir(this.conf) +
      ", hbase.cluster.distributed=" + this.conf.getBoolean(HConstants.CLUSTER_DISTRIBUTED, false));
    // Disable usage of meta replicas in the master
    this.conf.setBoolean(HConstants.USE_META_REPLICAS, false);
    Replication.decorateMasterConfiguration(this.conf);
    // Hack! Maps DFSClient => Master for logs.  HDFS made this
    // config param for task trackers, but we can piggyback off of it.
    if (this.conf.get("mapreduce.task.attempt.id") == null) {
      this.conf.set("mapreduce.task.attempt.id", "hb_m_" + this.serverName.toString());
    }
    // should we check the compression codec type at master side, default true, HBASE-6370
    this.masterCheckCompression = conf.getBoolean("hbase.master.check.compression", true);
    // should we check encryption settings at master side, default true
    this.masterCheckEncryption = conf.getBoolean("hbase.master.check.encryption", true);
    this.metricsMaster = new MetricsMaster(new MetricsMasterWrapperImpl(this));
    // preload table descriptor at startup
    this.preLoadTableDescriptors = conf.getBoolean("hbase.master.preload.tabledescriptors", true);
    // Do we publish the status?
    boolean shouldPublish = conf.getBoolean(HConstants.STATUS_PUBLISHED,
        HConstants.STATUS_PUBLISHED_DEFAULT);
    Class<? extends ClusterStatusPublisher.Publisher> publisherClass =
        conf.getClass(ClusterStatusPublisher.STATUS_PUBLISHER_CLASS,
            ClusterStatusPublisher.DEFAULT_STATUS_PUBLISHER_CLASS,
            ClusterStatusPublisher.Publisher.class);

    if (shouldPublish) {
      if (publisherClass == null) {
        LOG.warn(HConstants.STATUS_PUBLISHED + " is true, but " +
            ClusterStatusPublisher.DEFAULT_STATUS_PUBLISHER_CLASS +
            " is not set - not publishing status");
      } else {
        clusterStatusPublisherChore = new ClusterStatusPublisher(this, conf, publisherClass);
        getChoreService().scheduleChore(clusterStatusPublisherChore);
      }
    }
    // Some unit tests don't need a cluster, so no zookeeper at all
    if (!conf.getBoolean("hbase.testing.nocluster", false)) {
      activeMasterManager = new ActiveMasterManager(zooKeeper, this.serverName, this);
      int infoPort = putUpJettyServer();
      /** 开启状态为Active的master的各种管理器 */
      startActiveMasterManager(infoPort);
    } else {
      activeMasterManager = null;
    }
  }

    然后会调到 finishActiveMasterInitialization，这里面最重要的是对后台负责触发负载执行的线程的初始化(其它分支挺多，我们只看主干)

 this.balancerChore = new BalancerChore(this);

    BalancerChore类继承了ScheduledChore，ScheduledChore类里面的run方法调用了chore()，然后会调到真正的balance方法，也就是说会周期性的去判断是否需要执行HRegion在HRegionServer间的负载迁移操作。

  

public class BalancerChore extends ScheduledChore {
  private static final Log LOG = LogFactory.getLog(BalancerChore.class);
  private final HMaster master;
//默认每隔5分钟触发一次balance方法，去判断此时集群是否需要进行HRegion的负载
  public BalancerChore(HMaster master) {
    super(master.getServerName() + "-BalancerChore", master, master.getConfiguration().getInt(
      "hbase.balancer.period", 300000));
    this.master = master;
  }