Hadoop源码解析、向NN发送心跳

Redamancy_06
49 阅读3分钟

开启掘金成长之旅！这是我参与「掘金日新计划 · 12 月更文挑战」的第31天，点击查看活动详情

点击register BPServiceActor.java

void register(NamespaceInfo nsInfo) throws IOException {
  // 创建注册信息
  DatanodeRegistration newBpRegistration = bpos.createRegistration();

  LOG.info(this + " beginning handshake with NN");

  while (shouldRun()) {
    try {
      // Use returned registration from namenode with updated fields
	  // 把注册信息发送给NN（DN调用接口方法，执行在NN）
      newBpRegistration = bpNamenode.registerDatanode(newBpRegistration);
      newBpRegistration.setNamespaceInfo(nsInfo);
      bpRegistration = newBpRegistration;
      break;
    } catch(EOFException e) {  // namenode might have just restarted
      LOG.info("Problem connecting to server: " + nnAddr + " :"
          + e.getLocalizedMessage());
      sleepAndLogInterrupts(1000, "connecting to server");
    } catch(SocketTimeoutException e) {  // namenode is busy
      LOG.info("Problem connecting to server: " + nnAddr);
      sleepAndLogInterrupts(1000, "connecting to server");
    }
  }
  … …
}

ctrl + n 搜索NameNodeRpcServer NameNodeRpcServer.java ctrl + f 在NameNodeRpcServer.java中搜索registerDatanode

public DatanodeRegistration registerDatanode(DatanodeRegistration nodeReg)
    throws IOException {
  checkNNStartup();
  verifySoftwareVersion(nodeReg);
  // 注册DN
  namesystem.registerDatanode(nodeReg);

  return nodeReg;
}

FSNamesystem.java

void registerDatanode(DatanodeRegistration nodeReg) throws IOException {
  writeLock();
  try {
    blockManager.registerDatanode(nodeReg);
  } finally {
    writeUnlock("registerDatanode");
  }
}

BlockManager.java

public void registerDatanode(DatanodeRegistration nodeReg)
    throws IOException {
  assert namesystem.hasWriteLock();
  datanodeManager.registerDatanode(nodeReg);
  bmSafeMode.checkSafeMode();
}

public void registerDatanode(DatanodeRegistration nodeReg)
    throws DisallowedDatanodeException, UnresolvedTopologyException {
	... ...
	// register new datanode 注册DN
    addDatanode(nodeDescr);
    blockManager.getBlockReportLeaseManager().register(nodeDescr);
    // also treat the registration message as a heartbeat
    // no need to update its timestamp
    // because its is done when the descriptor is created
	// 将DN添加到心跳管理
    heartbeatManager.addDatanode(nodeDescr);
    heartbeatManager.updateDnStat(nodeDescr);
    incrementVersionCount(nodeReg.getSoftwareVersion());
    startAdminOperationIfNecessary(nodeDescr);
    success = true;
	... ...
}

void addDatanode(final DatanodeDescriptor node) {
  // To keep host2DatanodeMap consistent with datanodeMap,
  // remove  from host2DatanodeMap the datanodeDescriptor removed
  // from datanodeMap before adding node to host2DatanodeMap.
  synchronized(this) {
    host2DatanodeMap.remove(datanodeMap.put(node.getDatanodeUuid(), node));
  }

  networktopology.add(node); // may throw InvalidTopologyException
  host2DatanodeMap.add(node);
  checkIfClusterIsNowMultiRack(node);
  resolveUpgradeDomain(node);

  … …
}

3.7 向NN发送心跳

点击BPServiceActor.java中的run方法中的offerService方法 BPServiceActor.java

private void offerService() throws Exception {

	while (shouldRun()) {
        ... ...
        HeartbeatResponse resp = null;
        if (sendHeartbeat) {

          boolean requestBlockReportLease = (fullBlockReportLeaseId == 0) &&
                  scheduler.isBlockReportDue(startTime);
          if (!dn.areHeartbeatsDisabledForTests()) {
		    // 发送心跳信息
            resp = sendHeartBeat(requestBlockReportLease);
            assert resp != null;
            if (resp.getFullBlockReportLeaseId() != 0) {
              if (fullBlockReportLeaseId != 0) {
				... ...
              }
              fullBlockReportLeaseId = resp.getFullBlockReportLeaseId();
            }
            ... ...
          }
        }
		... ...
	}
}

HeartbeatResponse sendHeartBeat(boolean requestBlockReportLease)
    throws IOException {
	... ...
	// 通过NN的RPC客户端发送给NN
	HeartbeatResponse response = bpNamenode.sendHeartbeat(bpRegistration,
        reports,
        dn.getFSDataset().getCacheCapacity(),
        dn.getFSDataset().getCacheUsed(),
        dn.getXmitsInProgress(),
        dn.getXceiverCount(),
        numFailedVolumes,
        volumeFailureSummary,
        requestBlockReportLease,
        slowPeers,
        slowDisks);
	... ...
}

ctrl + n 搜索NameNodeRpcServer NameNodeRpcServer.java ctrl + f 在NameNodeRpcServer.java中搜索sendHeartbeat

public HeartbeatResponse sendHeartbeat(DatanodeRegistration nodeReg,
    StorageReport[] report, long dnCacheCapacity, long dnCacheUsed,
    int xmitsInProgress, int xceiverCount,
    int failedVolumes, VolumeFailureSummary volumeFailureSummary,
    boolean requestFullBlockReportLease,
    @Nonnull SlowPeerReports slowPeers,
@Nonnull SlowDiskReports slowDisks) throws IOException {

  checkNNStartup();
  verifyRequest(nodeReg);

  // 处理DN发送的心跳
  return namesystem.handleHeartbeat(nodeReg, report,
      dnCacheCapacity, dnCacheUsed, xceiverCount, xmitsInProgress,
      failedVolumes, volumeFailureSummary, requestFullBlockReportLease,
      slowPeers, slowDisks);
}

HeartbeatResponse handleHeartbeat(DatanodeRegistration nodeReg,
    StorageReport[] reports, long cacheCapacity, long cacheUsed,
    int xceiverCount, int xmitsInProgress, int failedVolumes,
    VolumeFailureSummary volumeFailureSummary,
    boolean requestFullBlockReportLease,
    @Nonnull SlowPeerReports slowPeers,
    @Nonnull SlowDiskReports slowDisks) throws IOException {
  readLock();
  try {
    //get datanode commands
    final int maxTransfer = blockManager.getMaxReplicationStreams()
        - xmitsInProgress;
	// 处理DN发送过来的心跳
    DatanodeCommand[] cmds = blockManager.getDatanodeManager().handleHeartbeat(
        nodeReg, reports, getBlockPoolId(), cacheCapacity, cacheUsed,
        xceiverCount, maxTransfer, failedVolumes, volumeFailureSummary,
        slowPeers, slowDisks);

    long blockReportLeaseId = 0;
    if (requestFullBlockReportLease) {
      blockReportLeaseId =  blockManager.requestBlockReportLeaseId(nodeReg);
    }
    //create ha status
    final NNHAStatusHeartbeat haState = new NNHAStatusHeartbeat(
        haContext.getState().getServiceState(),
        getFSImage().getCorrectLastAppliedOrWrittenTxId());

	// 响应DN的心跳
    return new HeartbeatResponse(cmds, haState, rollingUpgradeInfo,
        blockReportLeaseId);
  } finally {
    readUnlock("handleHeartbeat");
  }
}

点击handleHeartbeat DatanodeManager.java

public DatanodeCommand[] handleHeartbeat(DatanodeRegistration nodeReg,
    StorageReport[] reports, final String blockPoolId,
    long cacheCapacity, long cacheUsed, int xceiverCount, 
    int maxTransfers, int failedVolumes,
    VolumeFailureSummary volumeFailureSummary,
    @Nonnull SlowPeerReports slowPeers,
    @Nonnull SlowDiskReports slowDisks) throws IOException {
  ... ...
  heartbeatManager.updateHeartbeat(nodeinfo, reports, cacheCapacity,
      cacheUsed, xceiverCount, failedVolumes, volumeFailureSummary);
  ... ...  
}

HeartbeatManager.java

synchronized void updateHeartbeat(final DatanodeDescriptor node,
    StorageReport[] reports, long cacheCapacity, long cacheUsed,
    int xceiverCount, int failedVolumes,
    VolumeFailureSummary volumeFailureSummary) {
  stats.subtract(node);
  blockManager.updateHeartbeat(node, reports, cacheCapacity, cacheUsed,
      xceiverCount, failedVolumes, volumeFailureSummary);
  stats.add(node);
}

BlockManager.java

void updateHeartbeat(DatanodeDescriptor node, StorageReport[] reports,
    long cacheCapacity, long cacheUsed, int xceiverCount, int failedVolumes,
    VolumeFailureSummary volumeFailureSummary) {

  for (StorageReport report: reports) {
    providedStorageMap.updateStorage(node, report.getStorage());
  }
  node.updateHeartbeat(reports, cacheCapacity, cacheUsed, xceiverCount,
      failedVolumes, volumeFailureSummary);
}

DatanodeDescriptor.java

void updateHeartbeat(StorageReport[] reports, long cacheCapacity,
    long cacheUsed, int xceiverCount, int volFailures,
    VolumeFailureSummary volumeFailureSummary) {
  updateHeartbeatState(reports, cacheCapacity, cacheUsed, xceiverCount,
      volFailures, volumeFailureSummary);
  heartbeatedSinceRegistration = true;
}

void updateHeartbeatState(StorageReport[] reports, long cacheCapacity,
    long cacheUsed, int xceiverCount, int volFailures,
    VolumeFailureSummary volumeFailureSummary) {
  // 更新存储
  updateStorageStats(reports, cacheCapacity, cacheUsed, xceiverCount,
      volFailures, volumeFailureSummary);
  // 更新心跳时间
  setLastUpdate(Time.now());
  setLastUpdateMonotonic(Time.monotonicNow());
  rollBlocksScheduled(getLastUpdateMonotonic());
}

private void updateStorageStats(StorageReport[] reports, long cacheCapacity,
    long cacheUsed, int xceiverCount, int volFailures,
    VolumeFailureSummary volumeFailureSummary) {
  long totalCapacity = 0;
  long totalRemaining = 0;
  long totalBlockPoolUsed = 0;
  long totalDfsUsed = 0;
  long totalNonDfsUsed = 0;
  … …

  setCacheCapacity(cacheCapacity);
  setCacheUsed(cacheUsed);
  setXceiverCount(xceiverCount);
  this.volumeFailures = volFailures;
  this.volumeFailureSummary = volumeFailureSummary;
  for (StorageReport report : reports) {

    DatanodeStorageInfo storage =
        storageMap.get(report.getStorage().getStorageID());
    if (checkFailedStorages) {
      failedStorageInfos.remove(storage);
    }

    storage.receivedHeartbeat(report);
    // skip accounting for capacity of PROVIDED storages!
    if (StorageType.PROVIDED.equals(storage.getStorageType())) {
      continue;
    }

    totalCapacity += report.getCapacity();
    totalRemaining += report.getRemaining();
    totalBlockPoolUsed += report.getBlockPoolUsed();
    totalDfsUsed += report.getDfsUsed();
    totalNonDfsUsed += report.getNonDfsUsed();
  }

  // Update total metrics for the node.
  // 更新存储相关信息
  setCapacity(totalCapacity);
  setRemaining(totalRemaining);
  setBlockPoolUsed(totalBlockPoolUsed);
  setDfsUsed(totalDfsUsed);
  setNonDfsUsed(totalNonDfsUsed);
  if (checkFailedStorages) {
    updateFailedStorage(failedStorageInfos);
  }
  long storageMapSize;
  synchronized (storageMap) {
    storageMapSize = storageMap.size();
  }
  if (storageMapSize != reports.length) {
    pruneStorageMap(reports);
  }
}

第四章 HDFS上传源码解析

4.1 HDFS的写数据流程

在这里插入图片描述

4.2 HDFS上传源码解析

在这里插入图片描述

4.3create创建过程

添加依赖

<dependencies>
	<dependency>
		<groupId>org.apache.hadoop</groupId>
		<artifactId>hadoop-client</artifactId>
		<version>3.1.3</version>
	</dependency>

	<dependency>
		<groupId>org.apache.hadoop</groupId>
		<artifactId>hadoop-hdfs</artifactId>
		<version>3.1.3</version>
	</dependency>

	<dependency>
		<groupId>org.apache.hadoop</groupId>
		<artifactId>hadoop-hdfs-client</artifactId>
		<version>3.1.3</version>
		<scope>provided</scope>
	</dependency>

	<dependency>
		<groupId>junit</groupId>
		<artifactId>junit</artifactId>
		<version>4.12</version>
	</dependency>
	<dependency>
		<groupId>org.slf4j</groupId>
		<artifactId>slf4j-log4j12</artifactId>
		<version>1.7.30</version>
	</dependency>
</dependencies>![在这里插入图片描述](https://img-blog.csdnimg.cn/9f616258d18a443c95b86d95490e5779.png)

4.3.1 DN向NN发起创建请求

用户自己写的代码

@Test
public void testPut2() throws IOException {
	FSDataOutputStream fos = fs.create(new Path("/input"));

	fos.write("hello world".getBytes());
}

FileSystem.java

public FSDataOutputStream create(Path f) throws IOException {
	return create(f, true);
}

public FSDataOutputStream create(Path f, boolean overwrite)
  throws IOException {
  
	return create(f, overwrite,
			  getConf().getInt(IO_FILE_BUFFER_SIZE_KEY,
				  IO_FILE_BUFFER_SIZE_DEFAULT),
			  getDefaultReplication(f),
			  getDefaultBlockSize(f));
}

public FSDataOutputStream create(Path f,
	boolean overwrite,
	int bufferSize,
	short replication,
	long blockSize) throws IOException {
	
	return create(f, overwrite, bufferSize, replication, blockSize, null);
}

public FSDataOutputStream create(Path f,
	boolean overwrite,
	int bufferSize,
	short replication,
	long blockSize,
	Progressable progress
	) throws IOException {
										
	return this.create(f, FsCreateModes.applyUMask(
	FsPermission.getFileDefault(), FsPermission.getUMask(getConf())),
	overwrite, bufferSize, replication, blockSize, progress);
}

public abstract FSDataOutputStream create(Path f,
	FsPermission permission,
	boolean overwrite,
	int bufferSize,
	short replication,
	long blockSize,
	Progressable progress) throws IOException;
avatar
文章
阅读
粉丝