hbase region split源码分析

hbase region split ：

split执行调用流程：

1.HbaseAdmin发起split：###

2.RSRpcServices实现类执行split（Implements the regionserver RPC services.）###

3.CompactSplitThread类与SplitRequest类用来执行region切割:###

4.splitRequest执行doSplitting操作###

4.1初始化两个子region###

4.2执行切割####

4.2.1：（创建子region。stepsBeforePONR函数）#####

4.2.2：子region执行：stepsAfterPONR函数执行(包含openDaughters函数)：#####

4.2.3：HRegionServer添加子region到meta表，加入RegionServer#####

4.3等待region切分完成，修改meta表信息，报告master####

1.HbaseAdmin发起split：

    public void split(final TableName tableName) throws IOException {
split(tableName, null);
  }
   public void split(final ServerName sn, final HRegionInfo hri,
  byte[] splitPoint) throws IOException {
if (hri.getStartKey() != null && splitPoint != null &&
 Bytes.compareTo(hri.getStartKey(), splitPoint) == 0) {
   throw new IOException("should not give a splitkey which equals to startkey!");
}
// TODO: There is no timeout on this controller. Set one!
HBaseRpcController controller = rpcControllerFactory.newController();
controller.setPriority(hri.getTable());

// TODO: this does not do retries, it should. Set priority and timeout in controller
AdminService.BlockingInterface admin = this.connection.getAdmin(sn);

//hbase中split调用请求通过protobuf（AdminProtos）实现
ProtobufUtil.split(controller, admin, hri, splitPoint);
    实现代码：
     try {
        admin.splitRegion(controller, request);
        } 

2.RSRpcServices实现类执行split（Implements the regionserver RPC services.）

      代码：
      try {
  checkOpen();
  requestCount.increment();
  Region region = getRegion(request.getRegion());
  region.startRegionOperation(Operation.SPLIT_REGION);
  if (region.getRegionInfo().getReplicaId() != HRegionInfo.DEFAULT_REPLICA_ID) {
throw new IOException("Can't split replicas directly. "
+ "Replicas are auto-split when their primary is split.");
  }
  LOG.info("Splitting " + region.getRegionInfo().getRegionNameAsString());
  region.flush(true);
  byte[] splitPoint = null;
  if (request.hasSplitPoint()) {

  //确定切割点splitpoint
splitPoint = request.getSplitPoint().toByteArray();
  }
  ((HRegion)region).forceSplit(splitPoint);

  //请求region切割
  regionServer.compactSplitThread.requestSplit(region, ((HRegion)region).checkSplit(),
RpcServer.getRequestUser());
  return SplitRegionResponse.newBuilder().build();
}

3.CompactSplitThread类与SplitRequest类用来执行region切割:

    代码：
    try {
  this.splits.execute(new SplitRequest(r, midKey, this.server, user));
  if (LOG.isDebugEnabled()) {
LOG.debug("Split requested for " + r + ".  " + this);
  }
}

try {
  //acquire a shared read lock on the table, so that table schema modifications
  //do not happen concurrently

  //获取table的读锁。。
  tableLock = server.getTableLockManager().readLock(parent.getTableDesc().getTableName()
  , "SPLIT_REGION:" + parent.getRegionInfo().getRegionNameAsString());
  try {
tableLock.acquire();
  }

4.splitRequest执行doSplitting操作

4.1###

       //初始化两个子region信息
this.hri_a = new HRegionInfo(hri.getTable(), startKey, this.splitrow, false, rid);
    this.hri_b = new HRegionInfo(hri.getTable(), this.splitrow, endKey, false, rid);

4.2###

       //执行切割
   st.execute(this.server, this.server, user);

4.2.1：（创建子region。stepsBeforePONR函数）####

            //创建两个子region
        PairOfSameType regions = createDaughters(server, services, user);

4.2.2：stepsAfterPONR函数执行（openDaughters）：####

            // 两个子region DaughterOpener线程 start
        DaughterOpener aOpener = new DaughterOpener(server, (HRegion)a);
        DaughterOpener bOpener = new DaughterOpener(server, (HRegion)b);

        //向hdfs上写入.regionInfo文件以便meta挂掉以便恢复 
        writeRegionInfoOnFilesystem(content, true);

        //初始化所有的hstore
        initializeStores(reporter, status);

        //LoadStoreFiles函数执行：
        if (files == null || files.size() == 0) {
        return new ArrayList();
        }

             // initialize the thread pool for opening store files in parallel..
            ThreadPoolExecutor storeFileOpenerThreadPool =
            this.region.getStoreFileOpenAndCloseThreadPool("StoreFileOpenerThread-" +
         this.getColumnFamilyName());
             CompletionService completionService =
            new ExecutorCompletionService(storeFileOpenerThreadPool);

            int totalValidStoreFile = 0;
            for (final StoreFileInfo storeFileInfo: files) {

            //HDFS上对应的路径和文件
            completionService.submit(new Callable() {
             @Override
            public StoreFile call() throws IOException {

            //每个文件创建一个StoreFile对象，对每个storefile对象会读取文件上的内容创建一个 
            HalfStoreFileReader读对象来操作该region的父region上的相应的文件，及该 
            region上目前存储的是引用文件，其指向的是其父region上的相应的文件，对该 
            region的所有读或写都将关联到父region上
         StoreFile storeFile = createStoreFileAndReader(storeFileInfo);
         return storeFile;
             }
            });
             totalValidStoreFile++;
             }


             //将子Region添加到rs的online region列表上，并添加到meta表上 
             {
         if (useZKForAssignment) {
          // add 2nd daughter first (see HBASE-4335)
          services.postOpenDeployTasks(b);
         } else if (!services.reportRegionStateTransition(TransitionCode.SPLIT,
           parent.getRegionInfo(), hri_a, hri_b)) {
         throw new IOException("Failed to report split region to master: "
          + parent.getRegionInfo().getShortNameToLog());
         }
         // Should add it to OnlineRegions
         services.addToOnlineRegions(b);
         if (useZKForAssignment) {
           services.postOpenDeployTasks(a);
        }
         services.addToOnlineRegions(a);
         }

参考文章：
http://blog.csdn.net/Pun_C/article/details/47173453