hadoop分析之三org.apache.hadoop.hdfs.server.namenode各个类的功能与角色

以hadoop0.21为例。

NameNode.java: 主要维护文件系统的名字空间和文件的元数据，以下是代码中的说明。

/**********************************************************
 * NameNode serves as both directory namespace manager and
 * "inode table" for the Hadoop DFS.  There is a single NameNode
 * running in any DFS deployment.  (Well, except when there
 * is a second backup/failover NameNode.)
 *
 * The NameNode controls two critical tables:
 *   1)  filename ->blocksequence (namespace)
 *   2)  block ->machinelist ("inodes")
 *
 * The first table is stored on disk and is very precious.
 * The second table is rebuilt every time the NameNode comes
 * up.
 *
 * 'NameNode' refers to both this class as well as the 'NameNode server'.
 * The 'FSNamesystem' class actually performs most of the filesystem
 * management.  The majority of the 'NameNode' class itself is concerned
 * with exposing the IPC interface and the http server to the outside world,
 * plus some configuration management.
 *
 * NameNode implements the ClientProtocol interface, which allows
 * clients to ask for DFS services.  ClientProtocol is not
 * designed for direct use by authors of DFS client code.  End -users
 * should instead use the org.apache.nutch.hadoop.fs.FileSystem class.
 *
 * NameNode also implements the DatanodeProtocol interface, used by
 * DataNode programs that actually store DFS data blocks.  These
 * methods are invoked repeatedly and automatically by all the
 * DataNodes in a DFS deployment.
 *
 * NameNode also implements the NamenodeProtocol interface, used by
 * secondary namenodes or rebalancing processes to get partial namenode's
 * state, for example partial blocksMap etc.
 **********************************************************/

FSNamesystem.java:  主要维护几个表的信息：维护了文件名与block列表的映射关系；有效的block的集合；block与节点列表的映射关系；节点与block列表的映射关系；更新的heatbeat节点的LRU cache

/***************************************************
 * FSNamesystem does the actual bookkeeping work for the
 * DataNode.
 *
 * It tracks several important tables.
 *
 * 1)  valid fsname --> blocklist  (kept on disk, logged)
 * 2)  Set of all valid blocks (inverted #1)
 * 3)  block --> machinelist (kept in memory, rebuilt dynamically from reports)
 * 4)  machine --> blocklist (inverted #2)
 * 5)  LRU cache of updated -heartbeat machines
 ***************************************************/

INode.java： HDFS将文件和文件目录抽象成INode。

/**
 * We keep an in-memory representation of the file/block hierarchy.
 * This is a base INode class containing common fields for file and
 * directory inodes.
 */

FSImage.java： 需要将INode信息持久化到磁盘上FSImage上。

/**
 * FSImage handles checkpointing and logging of the namespace edits.
 *
 */

FSEditLog.java:写Edits文件

/**
 * FSEditLog maintains a log of the namespace modifications.
 *
 */

BlockInfo.java:INode主要是所文件和目录信息的，而对于文件的内容来说，这是用block描述的。我们假设一个文件的长度大小为Size,那么从文件的0偏移开始，按照固定大小，顺序对文件划分并编号，划分好的每一块为一个block

/**
 * Internal class for block metadata.
 */

DatanodeDescriptor.java:代表的具体的存储对象。

/**************************************************
 * DatanodeDescriptor tracks stats on a given DataNode,
 * such as available storage capacity, last update time, etc.,
 * and maintains a set of blocks stored on the datanode.
 *
 * This data structure is a data structure that is internal
 * to the namenode. It is *not* sent over- the- wire to the Client
 * or the Datnodes. Neither is it stored persistently in the
 * fsImage.

 **************************************************/

FSDirectory.java: 代表了HDFS中的所有目录和结构属性

/*************************************************
 * FSDirectory stores the filesystem directory state.
 * It handles writing/loading values to disk, and logging
 * changes as we go.
 *
 * It keeps the filename->blockset mapping always- current
 * and logged to disk.
 *
 *************************************************/

EditLogOutputStream.java:所有的日志记录都是通过EditLogOutputStream输出，在具体实例化的时候，这一组EditLogOutputStream包含多个EditLogFIleOutputStream和一个EditLogBackupOutputStream

/**
 * A generic abstract class to support journaling of edits logs into
 * a persistent storage.
 */

EditLogFileOutputStream.java: 将日志记录写到edits或edits.new中。

/**
 * An implementation of the abstract class {@link EditLogOutputStream}, which
 * stores edits in a local file.
 */

EditLogBackupOutputStream.java：将日志通过网络发送到backupnode上。

/**
 * An implementation of the abstract class {@link EditLogOutputStream},
 * which streams edits to a backup node.
 *
 * @see org.apache.hadoop.hdfs.server.protocol.NamenodeProtocol#journal
 * (org.apache.hadoop.hdfs.server.protocol.NamenodeRegistration,
 *  int, int, byte[])
 */

BackupNode.java：name Node的backup：升级阶段：Secondary Name Node -》Checkpoint Node(定期保存元数据，定期checkpoint) -》Backup Node(在内存中保持一份和Name Node完全一致的镜像，当元数据发生变化时，其元数据进行更新，可以利用自身的镜像来checkpoint，无需从nameNode下载)-》Standby Node（可以进行热备）

/**
 * BackupNode.
 * <p>
 * Backup node can play two roles.
 * <ol>
 * <li>{@link NamenodeRole#CHECKPOINT} node periodically creates checkpoints,
 * that is downloads image and edits from the active node, merges them, and
 * uploads the new image back to the active. </li>
 * <li>{@link NamenodeRole#BACKUP} node keeps its namespace in sync with the
 * active node, and periodically creates checkpoints by simply saving the
 * namespace image to local disk(s).</li>
 * </ol>
 */

BackupStorage.java：在Backup Node备份目录下创建jspool，并创建edits.new,将输出流指向edits.new

  /**
   * Load checkpoint from local files only if the memory state is empty.<br>
   * Set new checkpoint time received from the name -node. <br>
   * Move <code>lastcheckpoint.tmp </code> to <code>previous.checkpoint</code> .
   * @throws IOException
   */

TransferFsImage.java:负责从name Node去文件。

/**
 * This class provides fetching a specified file from the NameNode.
 */

GetImageServlet.java：是httpServlet的子类，处理doGet请求。

/**
 * This class is used in Namesystem's jetty to retrieve a file.
 * Typically used by the Secondary NameNode to retrieve image and
 * edit file for periodic checkpointing.
 */