Hadoop HA-HDFS

Hadoop HA高可用

概述

  • 所谓 HA(High Availablity),即高可用（7*24 小时不中断服务）
  • 实现高可用最关键的策略是消除单点故障。HA 严格来说应该分成各个组件的 HA 机制：HDFS 的 HA 和 YARN 的 HA。
  • NameNode 主要在以下两个方面影响 HDFS 集群
    • NameNode 机器发生意外，如宕机，集群将无法使用，直到管理员重启
    • NameNode 机器需要升级，包括软件、硬件升级，此时集群也将无法使用

HDFS HA 功能通过配置多个 NameNodes(Active/Standby)实现在集群中NameNode 的热备来解决上述问题。如果出现故障，如机器崩溃或机器需要升级维护，这时可通过此种方式将 NameNode 很快的切换到另外一台机器。

HDFS-HA 集群搭建

当前 HDFS 集群的规划

hadoop102 hadoop103 hadoop104
NameNode Secondarynamenode
DataNode DataNode DataNode

HA 的主要目的是消除 namenode 的单点故障,需要将 hdfs 集群规划成以下模样

hadoop102 hadoop103 hadoop104
NameNode NameNode NameNode
DataNode DataNode DataNode

HDFS-HA核心问题

  • 怎么保证三台namenode的数据一致

    • Fsimage：让一台nn生成数据，让其他机器nn同步
    • Edits：需要引进新的模块JournalNode来保证edits的文件的数据一致性

  • 怎么让同时只有一台nn是active，其他所有是standby的

    • 手动分配
    • 自动分配

  • 2nn在ha架构中并不存在，定期合并fsimage和edits的活谁来干

    由standby的nn来干

  • 如果nn真的发生了问题，怎么让其他的nn上位干活

    • 手动故障转移
    • 自动故障转移

HDFS-HA手动模式

环境准备

  • 修改IP
  • 修改主机名及主机名和IP地址的映射
  • 关闭防火墙
  • ssh免密登录
  • 安装JDK，配置环境变量等

规划集群

hadoop102 hadoop103 hadoop104
NameNode NameNode NameNode
JournalNode JournalNode JournalNode
DataNode DataNode DataNode

配置HDFS-HA集群

  • 官方地址：Apache Hadoop

  • 在opt目录下创建一个ha文件夹

    [atguigu@hadoop102 ~]$ cd /opt
    [atguigu@hadoop102 opt]$ sudo mkdir ha
    [atguigu@hadoop102 opt]$ sudo chown atguigu:atguigu /opt/ha

  • 将/opt/module/下的 hadoop-3.1.3 拷贝到/opt/ha 目录下（记得删除 data 和 log 目录）

    [atguigu@hadoop102 opt]$ cp -r /opt/module/hadoop-3.1.3 /opt/ha/

  • 配置core-site.xml

    <configuration>
    <!-- 把多个 NameNode 的地址组装成一个集群 mycluster -->
     <property>
     <name>fs.defaultFS</name>
     <value>hdfs://mycluster</value>
     </property>
    <!-- 指定 hadoop 运行时产生文件的存储目录 -->
     <property>
     <name>hadoop.tmp.dir</name>
     <value>/opt/ha/hadoop-3.1.3/data</value>
     </property>
    </configuration>

  • 配置hdfs-site.xml

    <configuration>
    <!-- NameNode 数据存储目录 -->
     <property>
     <name>dfs.namenode.name.dir</name>
     <value>file://${hadoop.tmp.dir}/name</value>
     </property>
    <!-- DataNode 数据存储目录 -->
     <property>
     <name>dfs.datanode.data.dir</name>
     <value>file://${hadoop.tmp.dir}/data</value>
     </property>
    <!-- JournalNode 数据存储目录 -->
     <property>
     <name>dfs.journalnode.edits.dir</name>
     <value>${hadoop.tmp.dir}/jn</value>
     </property>
    <!-- 完全分布式集群名称 -->
     <property>
     <name>dfs.nameservices</name>
     <value>mycluster</value>
     </property>
    <!-- 集群中 NameNode 节点都有哪些 -->
     <property>
     <name>dfs.ha.namenodes.mycluster</name>
     <value>nn1,nn2,nn3</value>
     </property>
    <!-- NameNode 的 RPC 通信地址 -->
     <property>
     <name>dfs.namenode.rpc-address.mycluster.nn1</name>
     <value>hadoop102:8020</value>
     </property>
     <property>
     <name>dfs.namenode.rpc-address.mycluster.nn2</name>
     <value>hadoop103:8020</value>
     </property>
     <property>
     <name>dfs.namenode.rpc-address.mycluster.nn3</name>
     <value>hadoop104:8020</value>
     </property>
    <!-- NameNode 的 http 通信地址 -->
     <property>
     <name>dfs.namenode.http-address.mycluster.nn1</name>
     <value>hadoop102:9870</value>
     </property>
     <property>
     <name>dfs.namenode.http-address.mycluster.nn2</name>
     <value>hadoop103:9870</value>
     </property>
     <property>
     <name>dfs.namenode.http-address.mycluster.nn3</name>
     <value>hadoop104:9870</value>
     </property>
    <!-- 指定 NameNode 元数据在 JournalNode 上的存放位置 -->
     <property>
    <name>dfs.namenode.shared.edits.dir</name>
    <value>qjournal://hadoop102:8485;hadoop103:8485;hadoop104:8485/myclus
    ter</value>
     </property>
    <!-- 访问代理类：client 用于确定哪个 NameNode 为 Active -->
     <property>
     <name>dfs.client.failover.proxy.provider.mycluster</name>
     
    <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyP
    rovider</value>
     </property>
    <!-- 配置隔离机制，即同一时刻只能有一台服务器对外响应 -->
     <property>
     <name>dfs.ha.fencing.methods</name>
     <value>sshfence</value>
     </property>
    <!-- 使用隔离机制时需要 ssh 秘钥登录-->
     <property>
     <name>dfs.ha.fencing.ssh.private-key-files</name>
     <value>/home/atguigu/.ssh/id_rsa</value>
     </property>
    </configuration>

  • 分发配置好的hadoop环境到其他节点

启动HDFS-HA集群

  • 将HADOOP_HOME环境变量更改到HA目录（三台机器）

    [atguigu@hadoop102 ~]$ sudo vim /etc/profile.d/my_env.sh

  • 将HADOOP_HOME部分改为如下部分

    #HADOOP_HOME
    export HADOOP_HOME=/opt/ha/hadoop-3.1.3
    export PATH=$PATH:$HADOOP_HOME/bin
    export PATH=$PATH:$HADOOP_HOME/sbin

  • 去三台机器上source环境变量

    [atguigu@hadoop102 ~]$source /etc/profile

  • 在各个 JournalNode 节点上，输入以下命令启动 journalnode 服务

    [atguigu@hadoop102 ~]$ hdfs --daemon start journalnode
    [atguigu@hadoop103 ~]$ hdfs --daemon start journalnode
    [atguigu@hadoop104 ~]$ hdfs --daemon start journalnode

  • 在[nn1]上，对其进行格式化，并启动

    [atguigu@hadoop102 ~]$ hdfs namenode -format
    [atguigu@hadoop102 ~]$ hdfs --daemon start namenode

  • 在[nn2]和[nn3]上，同步 nn1 的元数据信息

    [atguigu@hadoop103 ~]$ hdfs namenode -bootstrapStandby
    [atguigu@hadoop104 ~]$ hdfs namenode -bootstrapStandby

  • 启动[nn2]和[nn3]

    [atguigu@hadoop103 ~]$ hdfs --daemon start namenode
    [atguigu@hadoop104 ~]$ hdfs --daemon start namenode

  • 查看web页面显示

  • 在所有节点上，启动datanode

    [atguigu@hadoop102 ~]$ hdfs --daemon start datanode
    [atguigu@hadoop103 ~]$ hdfs --daemon start datanode
    [atguigu@hadoop104 ~]$ hdfs --daemon start datanode

  • 将[nn1]切换为 Active

    [atguigu@hadoop102 ~]$ hdfs haadmin -transitionToActive nn1

  • 查看是否 Active

    [atguigu@hadoop102 ~]$ hdfs haadmin -getServiceState nn1

HDFS-HA自动模式

自动故障转移为 HDFS 部署增加了两个新组件：ZooKeeper 和 ZKFailoverController（ZKFC）进程，如图所示。ZooKeeper 是维护少量协调数据，通知客户端这些数据的改变 和监视客户端故障的高可用服务。

HDFS-HA自动故障转移的集群规划

hadoop102 hadoop103 hadoop104
NameNode NameNode NameNode
JournalNode JournalNode JournalNode
DataNode DataNode DataNode
Zookeeper Zookeeper Zookeeper
ZKFC ZKFC ZKFC

具体配置HDFS-HA自动故障转移

  • 具体配置

    • 在hdfs-site.xml中增加

      <!-- 启用 nn 故障自动转移 -->
      <property>
      <name>dfs.ha.automatic-failover.enabled</name>
      <value>true</value>
      </property>

    • 在 core-site.xml 文件中增加

      <!-- 指定 zkfc 要连接的 zkServer 地址 -->
      <property>
      <name>ha.zookeeper.quorum</name>
      <value>hadoop102:2181,hadoop103:2181,hadoop104:2181</value>
      </property>

  • 启动

    • 关闭所有HDFS服务

      [atguigu@hadoop102 ~]$ stop-dfs.sh

    • 启动 Zookeeper 集群\

      [atguigu@hadoop102 ~]$ zkServer.sh start
      [atguigu@hadoop103 ~]$ zkServer.sh start
      [atguigu@hadoop104 ~]$ zkServer.sh start

    • 启动Zookeeper以后，然后再初始化HA在Zookeeper中状态

      [atguigu@hadoop102 ~]$ hdfs zkfc -formatZK

    • 启动 HDFS 服务

      [atguigu@hadoop102 ~]$ start-dfs.sh

    • 可以去 zkCli.sh 客户端查看 Namenode 选举锁节点内容

      [zk: localhost:2181(CONNECTED) 7] get -s 
      /hadoop-ha/mycluster/ActiveStandbyElectorLock
      myclusternn2 hadoop103
      cZxid = 0x10000000b
      ctime = Tue Jul 14 17:00:13 CST 2020
      mZxid = 0x10000000b
      mtime = Tue Jul 14 17:00:13 CST 2020
      pZxid = 0x10000000b
      cversion = 0
      dataVersion = 0
      aclVersion = 0
      ephemeralOwner = 0x40000da2eb70000
      dataLength = 33
      numChildren = 0

  • 验证

    • 将 Active NameNode 进程 kill，查看网页端三台 Namenode 的状态变化

      [atguigu@hadoop102 ~]$ kill -9 namenode 的进程 id

解决 NN 连接不上 JN 的问题

自动故障转移配置好以后，然后使用 start-dfs.sh 群起脚本启动 hdfs 集群，有可能 会遇到 NameNode 起来一会后，进程自动关闭的问题。查看 NameNode 日志，报错信 息如下：

2020-08-17 10:11:40,658 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:40,659 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:40,659 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:41,660 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 1 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:41,660 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 1 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:41,665 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 1 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:42,661 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 2 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:42,661 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 2 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:42,667 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 2 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:43,662 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 3 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:43,662 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 3 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:43,668 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 3 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:44,663 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 4 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:44,663 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 4 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:44,670 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 4 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:45,467 INFO org.apache.hadoop.hdfs.qjournal.client.QuorumJournalManager: Waited 6001 ms (timeout=20000 ms) for a response for selectStreamingInputStreams. No responses yet. 2020-08-17 10:11:45,664 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 5 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:45,664 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 5 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:45,672 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 5 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:46,469 INFO org.apache.hadoop.hdfs.qjournal.client.QuorumJournalManager: Waited 7003 ms (timeout=20000 ms) for a response for selectStreamingInputStreams. No responses yet. 2020-08-17 10:11:46,665 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 6 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:46,665 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 6 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:46,673 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 6 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:47,470 INFO org.apache.hadoop.hdfs.qjournal.client.QuorumJournalManager: Waited 8004 ms (timeout=20000 ms) for a response for selectStreamingInputStreams. No responses yet. 2020-08-17 10:11:47,666 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 7 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:47,667 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 7 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:47,674 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 7 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:48,471 INFO org.apache.hadoop.hdfs.qjournal.client.QuorumJournalManager: Waited 9005 ms (timeout=20000 ms) for a response for selectStreamingInputStreams. No responses yet. 2020-08-17 10:11:48,668 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 8 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:48,668 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 8 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:48,675 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 8 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:49,669 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop102/192.168.6.102:8485. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:49,673 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop104/192.168.6.104:8485. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:49,676 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: hadoop103/192.168.6.103:8485. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2020-08-17 10:11:49,678 WARN org.apache.hadoop.hdfs.server.namenode.FSEditLog: Unable to determine input streams from QJM to [192.168.6.102:8485, 192.168.6.103:8485, 192.168.6.104:8485]. Skipping. org.apache.hadoop.hdfs.qjournal.client.QuorumException: Got too many exceptions to achieve quorum size 2/3. 3 exceptions thrown: 192.168.6.103:8485: Call From hadoop102/192.168.6.102 to hadoop103:8485 failed on connection exception: java.net.ConnectException: 拒绝连接; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused 192.168.6.102:8485: Call From hadoop102/192.168.6.102 to hadoop102:8485 failed on connection exception: java.net.ConnectException: 拒绝连接; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused 192.168.6.104:8485: Call From hadoop102/192.168.6.102 to hadoop104:8485 failed on connection exception: java.net.ConnectException: 拒绝连接; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused

查看报错日志，可分析出报错原因是因为 NameNode 连接不上 JournalNode，而利 用 jps 命令查看到三台 JN 都已经正常启动，为什么 NN 还是无法正常连接到 JN 呢？这 是因为 start-dfs.sh 群起脚本默认的启动顺序是先启动 NN，再启动 DN，然后再启动 JN， 并且默认的 rpc 连接参数是重试次数为 10，每次重试的间隔是 1s，也就是说启动完 NN 以后的 10s 中内，JN 还启动不起来，NN 就会报错了。

core-default.xml 里面有两个参数如下：

<!-- NN 连接 JN 重试次数，默认是 10 次 -->
<property>
 <name>ipc.client.connect.max.retries</name>
 <value>10</value>
</property>
<!-- 重试时间间隔，默认 1s -->
<property>
 <name>ipc.client.connect.retry.interval</name>
 <value>1000</value>
</property>

解决方案：遇到上述问题后，可以稍等片刻，等 JN 成功启动后，手动启动下三台NN

[atguigu@hadoop102 ~]$ hdfs --daemon start namenode
[atguigu@hadoop103 ~]$ hdfs --daemon start namenode
[atguigu@hadoop104 ~]$ hdfs --daemon start namenode

也可以在 core-site.xml 里面适当调大上面的两个参数：

<!-- NN 连接 JN 重试次数，默认是 10 次 -->
<property>
 <name>ipc.client.connect.max.retries</name>
 <value>20</value>
</property>
<!-- 重试时间间隔，默认 1s -->
<property>
 <name>ipc.client.connect.retry.interval</name>
 <value>5000</value>
</property>
posted @ 2022-02-04 11:13  逆十字  阅读(85)  评论(0)    收藏  举报