hadoop format过程

private static boolean format(Configuration conf,
                                boolean isConfirmationNeeded
                                ) throws IOException {
    boolean allowFormat = conf.getBoolean("dfs.namenode.support.allowformat",    // 获取format配置文件，
                                          true);
    if (!allowFormat) {
      throw new IOException("The option dfs.namenode.support.allowformat is "
                            + "set to false for this filesystem, so it "
                            + "cannot be formatted. You will need to set "
                            + "dfs.namenode.support.allowformat parameter "
                            + "to true in order to format this filesystem");
    }

  }

为了防止生产集群误操作hadoop被format，可加入一下配置，

<property>
  <name>dfs.namenode.support.allowformat</name>
  <value>flase</value>
  <description>Does HDFS namenode allow itself to be formatted?
               You should consider setting this to false for any production
               cluster, to avoid any possibility of formatting a running DFS.
  </description>
</property>

Collection<File> dirsToFormat = FSNamesystem.getNamespaceDirs(conf); //获取namenode元数据目录， FSNamesystem.java  395行

 Collection<File> editDirsToFormat =
                 FSNamesystem.getNamespaceEditsDirs(conf);  ／／ 获取editlog目录, FSNamesystem.java 406行

 public static Collection<File> getNamespaceDirs(Configuration conf) {
    Collection<String> dirNames = conf.getStringCollection("dfs.name.dir");
    if (dirNames.isEmpty())
      dirNames.add("/tmp/hadoop/dfs/name");            ／／如果不配置目录，将使用默认目录：/tmp/hadoop/dfs/name
    Collection<File> dirs = new ArrayList<File>(dirNames.size());
    for(String name : dirNames) {
      dirs.add(new File(name));               ／／新建目录的意思
    }
    return dirs;                              ／／最后将目录返回
  }

Collection 网上看好像是接口的意思，记得以前学的时候implements是接口的意思，这个问题，有待深入，哈哈

  for(Iterator<File> it = dirsToFormat.iterator(); it.hasNext();) {   //这里是个for 循环，说明元数据目录可配置多个,直到所有目录
      File curDir = it.next();
      if (!curDir.exists())
        continue;
      if (isConfirmationNeeded) {
        System.err.print("Re-format filesystem in " + curDir +" ? (Y or N) ");     ／／如果已经被format过，将提示是否re-format
        if (!(System.in.read() == 'Y')) {
          System.err.println("Format aborted in "+ curDir);
          return true;
        }
        while(System.in.read() != '\n'); // discard the enter-key        直接回车不输入的话，应该是不格式化，

 FSNamesystem nsys = new FSNamesystem(new FSImage(dirsToFormat,
                                         editDirsToFormat), conf);    ／／先实例化FSNamesystem，传入namenode和editlog两个目录
    nsys.dir.fsImage.format();                       //执行format
    return false;
  }

//开始看format的过程：

  FSNamesystem(FSImage fsImage, Configuration conf) throws IOException {
    this.fsLock = new ReentrantReadWriteLock();
    setConfigurationParameters(conf);                               ／／ setConfigurationParameters  FSNamesystem.java   466行
    this.dir = new FSDirectory(fsImage, this, conf);
  }

private void setConfigurationParameters(Configuration conf)
                                          throws IOException {
    fsNamesystemObject = this;

    if (conf.getBoolean("hadoop.disable.shell",false)){
      conf.setStrings(UnixUserGroupInformation.UGI_PROPERTY_NAME, new String[]{"hadoop", "hadoop"});
      Shell.setDisabled(true);
    }

    try {
      fsOwner = UnixUserGroupInformation.login(conf);
    } catch (LoginException e) {
      throw new IOException(StringUtils.stringifyException(e));
    }

    LOG.info("fsOwner=" + fsOwner);
           　　　　　　　　                                                            ／／ 用户和目录权限
    this.hasRwLock = conf.getBoolean("dfs.rwlock", false);
    this.supergroup = conf.get("dfs.permissions.supergroup", "supergroup");
    this.isPermissionEnabled = conf.getBoolean("dfs.permissions", true);
    this.persistBlocks = conf.getBoolean("dfs.persist.blocks", false);
    LOG.info("supergroup=" + supergroup);
    LOG.info("isPermissionEnabled=" + isPermissionEnabled);
    short filePermission = (short)conf.getInt("dfs.upgrade.permission", 0777);
    this.defaultPermission = PermissionStatus.createImmutable(
        fsOwner.getUserName(), supergroup, new FsPermission(filePermission));

    this.maxCorruptFilesReturned = conf.getInt("dfs.corruptfilesreturned.max",
        DEFAULT_MAX_CORRUPT_FILES_RETURNED);
    this.defaultReplication = conf.getInt("dfs.replication", 3);
    this.maxReplication = conf.getInt("dfs.replication.max", 512);
    this.minReplication = conf.getInt("dfs.replication.min", 1);
    if (minReplication <= 0)                                                                    ／／ 获取备份数，最大512个，最小1个，小于1 报异常
      throw new IOException(
                            "Unexpected configuration parameters: dfs.replication.min = "
                            + minReplication
                            + " must be greater than 0");
    if (maxReplication >= (int)Short.MAX_VALUE)
      throw new IOException(
                            "Unexpected configuration parameters: dfs.replication.max = "
                            + maxReplication + " must be less than " + (Short.MAX_VALUE));
    if (maxReplication < minReplication)
      throw new IOException(
                            "Unexpected configuration parameters: dfs.replication.min = "
                            + minReplication
                            + " must be less than dfs.replication.max = "
                            + maxReplication);
    this.maxReplicationStreams = conf.getInt("dfs.max-repl-streams", 2);
    long heartbeatInterval = conf.getLong("dfs.heartbeat.interval", 3) * 1000;       ／／ 心跳时间
    this.heartbeatRecheckInterval = conf.getInt(
        "heartbeat.recheck.interval", 5 * 60 * 1000); // 5 minutes                ／／ 5分钟check
    this.heartbeatExpireInterval = 2 * heartbeatRecheckInterval +
      10 * heartbeatInterval;　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　／／ 10分钟丢弃
    this.replicationRecheckInterval =
      conf.getInt("dfs.replication.interval", 3) * 1000L;
    this.defaultBlockSize = conf.getLong("dfs.block.size", DEFAULT_BLOCK_SIZE);
    this.maxFsObjects = conf.getLong("dfs.max.objects", 0);
    this.blockInvalidateLimit = Math.max(this.blockInvalidateLimit,
                                         20*(int)(heartbeatInterval/1000));
    this.accessTimePrecision = conf.getLong("dfs.access.time.precision", 0);
    this.supportAppends = conf.getBoolean("dfs.support.append", false);

    // set soft and hard lease period
    long hardLeaseLimit = conf.getLong(FSConstants.DFS_HARD_LEASE_KEY,
             FSConstants.LEASE_HARDLIMIT_PERIOD);
    long softLeaseLimit = conf.getLong(FSConstants.DFS_SOFT_LEASE_KEY,
             FSConstants.LEASE_SOFTLIMIT_PERIOD);
    this.leaseManager.setLeasePeriod(
             Math.min(hardLeaseLimit, softLeaseLimit), hardLeaseLimit);
  }

默认快大小是64M

<property>
  <name>dfs.block.size</name>
  <value>67108864</value>
  <description>The default block size for new files.</description>
</property>

 FSDirectory(FSImage fsImage, FSNamesystem ns, Configuration conf) {
    rootDir = new INodeDirectoryWithQuota(INodeDirectory.ROOT_NAME,
        ns.createFsOwnerPermissions(new FsPermission((short)0755)),
        Integer.MAX_VALUE, -1);
    this.fsImage = fsImage;
    namesystem = ns;
    initialize(conf);
  }