1. Hadoop的HA机制
前言：正式引入HA机制是从hadoop2.0开始，之前的版本中没有HA机制
1.1. HA的运作机制
（1）hadoop-HA集群运作机制介绍
所谓HA，即高可用（7*24小时不中断服务）
实现高可用最关键的是消除单点故障
hadoop-ha严格来说应该分成各个组件的HA机制——HDFS的HA、YARN的HA
（2）HDFS的HA机制详解
通过双namenode消除单点故障
双namenode协调工作的要点：
A、元数据管理方式需要改变：
内存中各自保存一份元数据
Edits日志只能有一份，只有Active状态的namenode节点可以做写操作
两个namenode都可以读取edits
共享的edits放在一个共享存储中管理（qjournal和NFS两个主流实现）
B、需要一个状态管理功能模块
实现了一个zkfailover，常驻在每一个namenode所在的节点
每一个zkfailover负责监控自己所在namenode节点，利用zk进行状态标识
当需要进行状态切换时，由zkfailover来负责切换
切换时需要防止brain split现象的发生
1.2. HDFS-HA图解
2. 主机规划
|
主机名称 |
外网IP |
内网IP |
操作系统 |
备注 |
安装软件 |
运行进程 |
|
mini01 |
10.0.0.111 |
172.16.1.111 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop |
NameNode、DFSZKFailoverController(zkfc) |
|
mini02 |
10.0.0.112 |
172.16.1.112 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop |
NameNode、DFSZKFailoverController(zkfc) |
|
mini03 |
10.0.0.113 |
172.16.1.113 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop、zookeeper |
ResourceManager |
|
mini04 |
10.0.0.114 |
172.16.1.114 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop、zookeeper |
ResourceManager |
|
mini05 |
10.0.0.115 |
172.16.1.115 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop、zookeeper |
DataNode、NodeManager、JournalNode、QuorumPeerMain |
|
mini06 |
10.0.0.116 |
172.16.1.116 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop、zookeeper |
DataNode、NodeManager、JournalNode、QuorumPeerMain |
|
mini07 |
10.0.0.117 |
172.16.1.117 |
CentOS 7.4 |
ssh port:22 |
jdk、hadoop、zookeeper |
DataNode、NodeManager、JournalNode、QuorumPeerMain |
|
1
|
注意：针对HA模式，就不需要SecondaryNameNode了，因为STANDBY状态的namenode会负责做checkpoint |
Linux添加hosts信息，保证每台都可以相互ping通
1 [root@mini01 ~]# cat /etc/hosts
2 127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
3 ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
4
5 10.0.0.111 mini01
6 10.0.0.112 mini02
7 10.0.0.113 mini03
8 10.0.0.114 mini04
9 10.0.0.115 mini05
10 10.0.0.116 mini06
11 10.0.0.117 mini07
Windows的hosts文件修改
1 # 文件位置C:\Windows\System32\drivers\etc 在hosts中追加如下内容
2 …………………………………………
3 10.0.0.111 mini01
4 10.0.0.112 mini02
5 10.0.0.113 mini03
6 10.0.0.114 mini04
7 10.0.0.115 mini05
8 10.0.0.116 mini06
9 10.0.0.117 mini07
3. 添加用户账号
1 # 使用一个专门的用户，避免直接使用root用户
2 # 添加用户、指定家目录并指定用户密码
3 useradd -d /app yun && echo '123456' | /usr/bin/passwd --stdin yun
4 # sudo提权
5 echo "yun ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers
6 # 让其它普通用户可以进入该目录查看信息
7 chmod 755 /app/
4. 实现yun用户免秘钥登录
1 要求：根据规划实现 mini01 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
2 实现 mini02 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
3 实现 mini03 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
4 实现 mini04 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
5 实现 mini05 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
6 实现 mini06 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
7 实现 mini07 到 mini01、mini02、mini03、mini04、mini05、mini06、mini07 免秘钥登录
8
9 # 可以使用ip也可以是hostname 但是由于我们计划使用的是 hostname 方式交互，所以使用hostname
10 # 同时hostname方式分发，可以通过hostname远程登录，也可以IP远程登录
具体过程就不多说了，请参见 Hadoop2.7.6_01_部署
5. Jdk【java8】
具体过程就不多说了，请参见 Hadoop2.7.6_01_部署
6. Zookeeper部署
根据规划zookeeper部署在mini03、mini04、mini05、mini06、mini07上
6.1. 配置信息
1 [yun@mini03 conf]$ pwd
2 /app/zookeeper/conf
3 [yun@mini03 conf]$ vim zoo.cfg
4 #单个客户端与单台服务器之间的连接数的限制，是ip级别的，默认是60，如果设置为0，那么表明不作任何限制。
5 maxClientCnxns=1500
6 # The number of milliseconds of each tick
7 tickTime=2000
8 # The number of ticks that the initial
9 # synchronization phase can take
10 initLimit=10
11 # The number of ticks that can pass between
12 # sending a request and getting an acknowledgement
13 syncLimit=5
14 # the directory where the snapshot is stored.
15 # do not use /tmp for storage, /tmp here is just
16 # example sakes.
17 # dataDir=/tmp/zookeeper
18 dataDir=/app/bigdata/zookeeper/data
19 # the port at which the clients will connect
20 clientPort=2181
21 #
22 # Be sure to read the maintenance section of the
23 # administrator guide before turning on autopurge.
24 #
25 # http://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_maintenance
26 #
27 # The number of snapshots to retain in dataDir
28 #autopurge.snapRetainCount=3
29 # Purge task interval in hours
30 # Set to "0" to disable auto purge feature
31 #autopurge.purgeInterval=1
32
33 # leader和follow通信端口和投票选举端口
34 server.3=mini03:2888:3888
35 server.4=mini04:2888:3888
36 server.5=mini05:2888:3888
37 server.6=mini06:2888:3888
38 server.7=mini07:2888:3888
6.2. 添加myid文件
1 [yun@mini03 data]$ pwd
2 /app/bigdata/zookeeper/data
3 [yun@mini03 data]$ vim myid # 其中mini03的myid 为3；mini04的myid 为4；mini05的myid 为5；mini06的myid 为6；mini07的myid 为7
4 3
6.3. 启动zk服务
1 # 依次在启动mini03、mini04、mini05、mini06、mini07 zk服务
2 [yun@mini03 ~]$ cd zookeeper/bin/
3 [yun@mini03 bin]$ pwd
4 /app/zookeeper/bin
5 [yun@mini03 bin]$ ll
6 total 56
7 -rwxr-xr-x 1 yun yun 238 Oct 1 2012 README.txt
8 -rwxr-xr-x 1 yun yun 1909 Oct 1 2012 zkCleanup.sh
9 -rwxr-xr-x 1 yun yun 1049 Oct 1 2012 zkCli.cmd
10 -rwxr-xr-x 1 yun yun 1512 Oct 1 2012 zkCli.sh
11 -rwxr-xr-x 1 yun yun 1333 Oct 1 2012 zkEnv.cmd
12 -rwxr-xr-x 1 yun yun 2599 Oct 1 2012 zkEnv.sh
13 -rwxr-xr-x 1 yun yun 1084 Oct 1 2012 zkServer.cmd
14 -rwxr-xr-x 1 yun yun 5467 Oct 1 2012 zkServer.sh
15 -rw-rw-r-- 1 yun yun 17522 Jun 28 21:01 zookeeper.out
16 [yun@mini03 bin]$ ./zkServer.sh start
17 JMX enabled by default
18 Using config: /app/zookeeper/bin/../conf/zoo.cfg
19 Starting zookeeper ... STARTED
6.4. 查询运行状态
1 # 其中mini03、mini04、mini06、mini07状态如下
2 [yun@mini03 bin]$ ./zkServer.sh status
3 JMX enabled by default
4 Using config: /app/zookeeper/bin/../conf/zoo.cfg
5 Mode: follower
6
7 # 其中mini05 状态如下
8 [yun@mini05 bin]$ ./zkServer.sh status
9 JMX enabled by default
10 Using config: /app/zookeeper/bin/../conf/zoo.cfg
11 Mode: leader
PS：4个follower 1个leader
7. Hadoop部署与配置修改
注意：每台机器的Hadoop以及配置相同
7.1. 部署
1 [yun@mini01 software]$ pwd
2 /app/software
3 [yun@mini01 software]$ ll
4 total 194152
5 -rw-r--r-- 1 yun yun 198811365 Jun 8 16:36 CentOS-7.4_hadoop-2.7.6.tar.gz
6 [yun@mini01 software]$ tar xf CentOS-7.4_hadoop-2.7.6.tar.gz
7 [yun@mini01 software]$ mv hadoop-2.7.6/ /app/
8 [yun@mini01 software]$ cd
9 [yun@mini01 ~]$ ln -s hadoop-2.7.6/ hadoop
10 [yun@mini01 ~]$ ll
11 total 4
12 lrwxrwxrwx 1 yun yun 13 Jun 9 16:21 hadoop -> hadoop-2.7.6/
13 drwxr-xr-x 9 yun yun 149 Jun 8 16:36 hadoop-2.7.6
14 lrwxrwxrwx 1 yun yun 12 May 26 11:18 jdk -> jdk1.8.0_112
15 drwxr-xr-x 8 yun yun 255 Sep 23 2016 jdk1.8.0_112
7.2. 环境变量
1 [root@mini01 profile.d]# pwd
2 /etc/profile.d
3 [root@mini01 profile.d]# vim hadoop.sh
4 export HADOOP_HOME="/app/hadoop"
5 export PATH=$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$PATH
6
7 [root@mini01 profile.d]# source /etc/profile # 生效
7.3. core-site.xml
1 [yun@mini01 hadoop]$ pwd
2 /app/hadoop/etc/hadoop
3 [yun@mini01 hadoop]$ vim core-site.xml
4 <?xml version="1.0" encoding="UTF-8"?>
5 <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
6 <!--
7 ……………………
8 -->
9
10 <!-- Put site-specific property overrides in this file. -->
11
12 <configuration>
13 <!-- 指定hdfs的nameservice为bi -->
14 <property>
15 <name>fs.defaultFS</name>
16 <value>hdfs://bi/</value>
17 </property>
18
19 <!-- 指定hadoop临时目录 -->
20 <property>
21 <name>hadoop.tmp.dir</name>
22 <value>/app/hadoop/tmp</value>
23 </property>
24
25 <!-- 指定zookeeper地址 -->
26 <property>
27 <name>ha.zookeeper.quorum</name>
28 <value>mini03:2181,mini04:2181,mini05:2181,mini06:2181,mini07:2181</value>
29 </property>
30
31 </configuration>
7.4. hdfs-site.xml
1 [yun@mini01 hadoop]$ pwd
2 /app/hadoop/etc/hadoop
3 [yun@mini01 hadoop]$ vim hdfs-site.xml
4 <?xml version="1.0" encoding="UTF-8"?>
5 <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
6 <!--
7 ……………………
8 -->
9
10 <!-- Put site-specific property overrides in this file. -->
11
12 <configuration>
13 <!--指定hdfs的nameservice为bi，需要和core-site.xml中的保持一致 -->
14 <property>
15 <name>dfs.nameservices</name>
16 <value>bi</value>
17 </property>
18
19 <!-- bi下面有两个NameNode，分别是nn1，nn2 -->
20 <property>
21 <name>dfs.ha.namenodes.bi</name>
22 <value>nn1,nn2</value>
23 </property>
24
25 <!-- nn1的RPC通信地址 -->
26 <property>
27 <name>dfs.namenode.rpc-address.bi.nn1</name>
28 <value>mini01:9000</value>
29 </property>
30 <!-- nn1的http通信地址 -->
31 <property>
32 <name>dfs.namenode.http-address.bi.nn1</name>
33 <value>mini01:50070</value>
34 </property>
35
36 <!-- nn2的RPC通信地址 -->
37 <property>
38 <name>dfs.namenode.rpc-address.bi.nn2</name>
39 <value>mini02:9000</value>
40 </property>
41 <!-- nn2的http通信地址 -->
42 <property>
43 <name>dfs.namenode.http-address.bi.nn2</name>
44 <value>mini02:50070</value>
45 </property>
46
47 <!-- 指定NameNode的edits元数据在JournalNode上的存放位置 -->
48 <property>
49 <name>dfs.namenode.shared.edits.dir</name>
50 <value>qjournal://mini05:8485;mini06:8485;mini07:8485/bi</value>
51 </property>
52
53 <!-- 指定JournalNode在本地磁盘存放数据的位置 -->
54 <property>
55 <name>dfs.journalnode.edits.dir</name>
56 <value>/app/hadoop/journaldata</value>
57 </property>
58
59 <!-- 开启NameNode失败自动切换 -->
60 <property>
61 <name>dfs.ha.automatic-failover.enabled</name>
62 <value>true</value>
63 </property>
64
65 <!-- 配置失败自动切换实现方式 -->
66 <property>
67 <name>dfs.client.failover.proxy.provider.bi</name>
68 <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
69 </property>
70
71 <!-- 配置隔离机制方法，多个机制用换行分割，即每个机制暂用一行-->
72 <!-- 其中shell(/bin/true) 表示可执行一个脚本 比如 shell(/app/yunwei/hadoop_fence.sh) -->
73 <property>
74 <name>dfs.ha.fencing.methods</name>
75 <value>
76 sshfence
77 shell(/bin/true)
78 </value>
79 </property>
80
81 <!-- 使用sshfence隔离机制时需要ssh免登陆 -->
82 <property>
83 <name>dfs.ha.fencing.ssh.private-key-files</name>
84 <value>/app/.ssh/id_rsa</value>
85 </property>
86
87 <!-- 配置sshfence隔离机制超时时间 单位：毫秒 -->
88 <property>
89 <name>dfs.ha.fencing.ssh.connect-timeout</name>
90 <value>30000</value>
91 </property>
92
93
94 </configuration>
7.5. mapred-site.xml
1 [yun@mini01 hadoop]$ pwd
2 /app/hadoop/etc/hadoop
3 [yun@mini01 hadoop]$ cp -a mapred-site.xml.template mapred-site.xml
4 [yun@mini01 hadoop]$ vim mapred-site.xml
5 <?xml version="1.0"?>
6 <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
7 <!--
8 ……………………
9 -->
10
11 <!-- Put site-specific property overrides in this file. -->
12
13 <configuration>
14 <!-- 指定mr框架为yarn方式 -->
15 <property>
16 <name>mapreduce.framework.name</name>
17 <value>yarn</value>
18 </property>
19
20 </configuration>
7.6. yarn-site.xml
1 [yun@mini01 hadoop]$ pwd
2 /app/hadoop/etc/hadoop
3 [yun@mini01 hadoop]$ vim yarn-site.xml
4 <?xml version="1.0"?>
5 <!--
6 ……………………
7 -->
8 <configuration>
9
10 <!-- Site specific YARN configuration properties -->
11 <!-- 开启RM高可用 -->
12 <property>
13 <name>yarn.resourcemanager.ha.enabled</name>
14 <value>true</value>
15 </property>
16
17 <!-- 指定RM的cluster id -->
18 <property>
19 <name>yarn.resourcemanager.cluster-id</name>
20 <value>yrc</value>
21 </property>
22
23 <!-- 指定RM的名字 -->
24 <property>
25 <name>yarn.resourcemanager.ha.rm-ids</name>
26 <value>rm1,rm2</value>
27 </property>
28
29 <!-- 分别指定RM的地址 -->
30 <property>
31 <name>yarn.resourcemanager.hostname.rm1</name>
32 <value>mini03</value>
33 </property>
34 <property>
35 <name>yarn.resourcemanager.hostname.rm2</name>
36 <value>mini04</value>
37 </property>
38
39 <!-- 指定zk集群地址 -->
40 <property>
41 <name>yarn.resourcemanager.zk-address</name>
42 <value>mini03:2181,mini04:2181,mini05:2181,mini06:2181,mini07:2181</value>
43 </property>
44
45 <!-- reduce 获取数据的方式 -->
46 <property>
47 <name>yarn.nodemanager.aux-services</name>
48 <value>mapreduce_shuffle</value>
49 </property>
50
51
52 </configuration>
7.7. 修改slaves
slaves是指定子节点的位置，因为要在mini01上启动HDFS、在mini03启动yarn，所以mini01上的slaves文件指定的是datanode的位置，mini03上的slaves文件指定的是nodemanager的位置
1 [yun@mini01 hadoop]$ pwd
2 /app/hadoop/etc/hadoop
3 [yun@mini01 hadoop]$ vim slaves
4 mini05
5 mini06
6 mini07
PS：改后配置后，将这些配置拷到其他Hadoop机器
8. 启动相关服务
注意：第一次启动时严格按照下面的步骤！！！！！！！
8.1. 启动zookeeper集群
前面已经启动了，这里就不说了
8.2. 启动journalnode
1 # 根据规划在mini05、mini06、mini07 启动
2 # 在第一次格式化的时候需要先启动journalnode 之后就不必了
3 [yun@mini05 ~]$ hadoop-daemon.sh start journalnode # 已经配置环境变量，所以不用进入到响应的目录
4 starting journalnode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-journalnode-mini05.out
5 [yun@mini05 ~]$ jps
6 1281 QuorumPeerMain
7 1817 Jps
8 1759 JournalNode
8.3. 格式化HDFS
1 # 在mini01上执行命令
2 [yun@mini01 ~]$ hdfs namenode -format
3 18/06/30 18:29:12 INFO namenode.NameNode: STARTUP_MSG:
4 /************************************************************
5 STARTUP_MSG: Starting NameNode
6 STARTUP_MSG: host = mini01/10.0.0.111
7 STARTUP_MSG: args = [-format]
8 STARTUP_MSG: version = 2.7.6
9 STARTUP_MSG: classpath = ………………
10 STARTUP_MSG: build = Unknown -r Unknown; compiled by 'root' on 2018-06-08T08:30Z
11 STARTUP_MSG: java = 1.8.0_112
12 ************************************************************/
13 18/06/30 18:29:12 INFO namenode.NameNode: registered UNIX signal handlers for [TERM, HUP, INT]
14 18/06/30 18:29:12 INFO namenode.NameNode: createNameNode [-format]
15 Formatting using clusterid: CID-2385f26e-72e6-4935-aa09-47848b5ba4be
16 18/06/30 18:29:13 INFO namenode.FSNamesystem: No KeyProvider found.
17 18/06/30 18:29:13 INFO namenode.FSNamesystem: fsLock is fair: true
18 18/06/30 18:29:13 INFO namenode.FSNamesystem: Detailed lock hold time metrics enabled: false
19 18/06/30 18:29:13 INFO blockmanagement.DatanodeManager: dfs.block.invalidate.limit=1000
20 18/06/30 18:29:13 INFO blockmanagement.DatanodeManager: dfs.namenode.datanode.registration.ip-hostname-check=true
21 18/06/30 18:29:13 INFO blockmanagement.BlockManager: dfs.namenode.startup.delay.block.deletion.sec is set to 000:00:00:00.000
22 18/06/30 18:29:13 INFO blockmanagement.BlockManager: The block deletion will start around 2018 Jun 30 18:29:13
23 18/06/30 18:29:13 INFO util.GSet: Computing capacity for map BlocksMap
24 18/06/30 18:29:13 INFO util.GSet: VM type = 64-bit
25 18/06/30 18:29:13 INFO util.GSet: 2.0% max memory 966.7 MB = 19.3 MB
26 18/06/30 18:29:13 INFO util.GSet: capacity = 2^21 = 2097152 entries
27 18/06/30 18:29:13 INFO blockmanagement.BlockManager: dfs.block.access.token.enable=false
28 18/06/30 18:29:13 INFO blockmanagement.BlockManager: defaultReplication = 3
29 18/06/30 18:29:13 INFO blockmanagement.BlockManager: maxReplication = 512
30 18/06/30 18:29:13 INFO blockmanagement.BlockManager: minReplication = 1
31 18/06/30 18:29:13 INFO blockmanagement.BlockManager: maxReplicationStreams = 2
32 18/06/30 18:29:13 INFO blockmanagement.BlockManager: replicationRecheckInterval = 3000
33 18/06/30 18:29:13 INFO blockmanagement.BlockManager: encryptDataTransfer = false
34 18/06/30 18:29:13 INFO blockmanagement.BlockManager: maxNumBlocksToLog = 1000
35 18/06/30 18:29:13 INFO namenode.FSNamesystem: fsOwner = yun (auth:SIMPLE)
36 18/06/30 18:29:13 INFO namenode.FSNamesystem: supergroup = supergroup
37 18/06/30 18:29:13 INFO namenode.FSNamesystem: isPermissionEnabled = true
38 18/06/30 18:29:13 INFO namenode.FSNamesystem: Determined nameservice ID: bi
39 18/06/30 18:29:13 INFO namenode.FSNamesystem: HA Enabled: true
40 18/06/30 18:29:13 INFO namenode.FSNamesystem: Append Enabled: true
41 18/06/30 18:29:13 INFO util.GSet: Computing capacity for map INodeMap
42 18/06/30 18:29:13 INFO util.GSet: VM type = 64-bit
43 18/06/30 18:29:13 INFO util.GSet: 1.0% max memory 966.7 MB = 9.7 MB
44 18/06/30 18:29:13 INFO util.GSet: capacity = 2^20 = 1048576 entries
45 18/06/30 18:29:13 INFO namenode.FSDirectory: ACLs enabled? false
46 18/06/30 18:29:13 INFO namenode.FSDirectory: XAttrs enabled? true
47 18/06/30 18:29:13 INFO namenode.FSDirectory: Maximum size of an xattr: 16384
48 18/06/30 18:29:13 INFO namenode.NameNode: Caching file names occuring more than 10 times
49 18/06/30 18:29:13 INFO util.GSet: Computing capacity for map cachedBlocks
50 18/06/30 18:29:13 INFO util.GSet: VM type = 64-bit
51 18/06/30 18:29:13 INFO util.GSet: 0.25% max memory 966.7 MB = 2.4 MB
52 18/06/30 18:29:13 INFO util.GSet: capacity = 2^18 = 262144 entries
53 18/06/30 18:29:13 INFO namenode.FSNamesystem: dfs.namenode.safemode.threshold-pct = 0.9990000128746033
54 18/06/30 18:29:13 INFO namenode.FSNamesystem: dfs.namenode.safemode.min.datanodes = 0
55 18/06/30 18:29:13 INFO namenode.FSNamesystem: dfs.namenode.safemode.extension = 30000
56 18/06/30 18:29:13 INFO metrics.TopMetrics: NNTop conf: dfs.namenode.top.window.num.buckets = 10
57 18/06/30 18:29:13 INFO metrics.TopMetrics: NNTop conf: dfs.namenode.top.num.users = 10
58 18/06/30 18:29:13 INFO metrics.TopMetrics: NNTop conf: dfs.namenode.top.windows.minutes = 1,5,25
59 18/06/30 18:29:13 INFO namenode.FSNamesystem: Retry cache on namenode is enabled
60 18/06/30 18:29:13 INFO namenode.FSNamesystem: Retry cache will use 0.03 of total heap and retry cache entry expiry time is 600000 millis
61 18/06/30 18:29:13 INFO util.GSet: Computing capacity for map NameNodeRetryCache
62 18/06/30 18:29:13 INFO util.GSet: VM type = 64-bit
63 18/06/30 18:29:13 INFO util.GSet: 0.029999999329447746% max memory 966.7 MB = 297.0 KB
64 18/06/30 18:29:13 INFO util.GSet: capacity = 2^15 = 32768 entries
65 18/06/30 18:29:14 INFO namenode.FSImage: Allocated new BlockPoolId: BP-1178102935-10.0.0.111-1530354554626
66 18/06/30 18:29:14 INFO common.Storage: Storage directory /app/hadoop/tmp/dfs/name has been successfully formatted.
67 18/06/30 18:29:14 INFO namenode.FSImageFormatProtobuf: Saving image file /app/hadoop/tmp/dfs/name/current/fsimage.ckpt_0000000000000000000 using no compression
68 18/06/30 18:29:14 INFO namenode.FSImageFormatProtobuf: Image file /app/hadoop/tmp/dfs/name/current/fsimage.ckpt_0000000000000000000 of size 320 bytes saved in 0 seconds.
69 18/06/30 18:29:15 INFO namenode.NNStorageRetentionManager: Going to retain 1 images with txid >= 0
70 18/06/30 18:29:15 INFO util.ExitUtil: Exiting with status 0
71 18/06/30 18:29:15 INFO namenode.NameNode: SHUTDOWN_MSG:
72 /************************************************************
73 SHUTDOWN_MSG: Shutting down NameNode at mini01/10.0.0.111
74 ************************************************************/
拷贝到mini02
1 #格式化后会在根据core-site.xml中的hadoop.tmp.dir配置生成个文件，这里我配置的是/app/hadoop/tmp，然后将/app/hadoop/tmp拷贝到mini02的/app/hadoop/下。
2 # 方法1：
3 [yun@mini01 hadoop]$ pwd
4 /app/hadoop
5 [yun@mini01 hadoop]$ scp -r tmp/ yun@mini02:/app/hadoop
6 VERSION 100% 202 189.4KB/s 00:00
7 seen_txid 100% 2 1.0KB/s 00:00
8 fsimage_0000000000000000000.md5 100% 62 39.7KB/s 00:00
9 fsimage_0000000000000000000 100% 320 156.1KB/s 00:00
10
11 ##########################3
12 # 方法2：##也可以这样，建议hdfs namenode -bootstrapStandby # 不过需要mini02的Hadoop起来才行
8.4. 格式化ZKFC
1 #在mini01上执行一次即可
2 [yun@mini01 ~]$ hdfs zkfc -formatZK
3 18/06/30 18:54:30 INFO tools.DFSZKFailoverController: Failover controller configured for NameNode NameNode at mini01/10.0.0.111:9000
4 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:zookeeper.version=3.4.6-1569965, built on 02/20/2014 09:09 GMT
5 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:host.name=mini01
6 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.version=1.8.0_112
7 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.vendor=Oracle Corporation
8 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.home=/app/jdk1.8.0_112/jre
9 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.class.path=……………………
10 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.library.path=/app/hadoop-2.7.6/lib/native
11 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.io.tmpdir=/tmp
12 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:java.compiler=<NA>
13 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:os.name=Linux
14 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:os.arch=amd64
15 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:os.version=3.10.0-693.el7.x86_64
16 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:user.name=yun
17 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:user.home=/app
18 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Client environment:user.dir=/app/hadoop-2.7.6
19 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Initiating client connection, connectString=mini03:2181,mini04:2181,mini05:2181,mini06:2181,mini07:2181 sessionTimeout=5000 watcher=org.apache.hadoop.ha.ActiveStandbyElector$WatcherWithClientRef@7f3b84b8
20 18/06/30 18:54:30 INFO zookeeper.ClientCnxn: Opening socket connection to server mini04/10.0.0.114:2181. Will not attempt to authenticate using SASL (unknown error)
21 18/06/30 18:54:30 INFO zookeeper.ClientCnxn: Socket connection established to mini04/10.0.0.114:2181, initiating session
22 18/06/30 18:54:30 INFO zookeeper.ClientCnxn: Session establishment complete on server mini04/10.0.0.114:2181, sessionid = 0x4644fff9cb80000, negotiated timeout = 5000
23 18/06/30 18:54:30 INFO ha.ActiveStandbyElector: Session connected.
24 18/06/30 18:54:30 INFO ha.ActiveStandbyElector: Successfully created /hadoop-ha/bi in ZK.
25 18/06/30 18:54:30 INFO zookeeper.ZooKeeper: Session: 0x4644fff9cb80000 closed
26 18/06/30 18:54:30 INFO zookeeper.ClientCnxn: EventThread shut down
8.5. 启动HDFS
1 # 在mini01上执行
2 [yun@mini01 ~]$ start-dfs.sh
3 Starting namenodes on [mini01 mini02]
4 mini01: starting namenode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-namenode-mini01.out
5 mini02: starting namenode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-namenode-mini02.out
6 mini07: starting datanode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-datanode-mini07.out
7 mini06: starting datanode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-datanode-mini06.out
8 mini05: starting datanode, logging to /app/hadoop-2.7.6/logs/hadoop-yun-datanode-mini05.out
9 Starting journal nodes [mini05 mini06 mini07]
10 mini07: journalnode running as process 1691. Stop it first.
11 mini06: journalnode running as process 1665. Stop it first.
12 mini05: journalnode running as process 1759. Stop it first.
13 Starting ZK Failover Controllers on NN hosts [mini01 mini02]
14 mini01: starting zkfc, logging to /app/hadoop-2.7.6/logs/hadoop-yun-zkfc-mini01.out
15 mini02: starting zkfc, logging to /app/hadoop-2.7.6/logs/hadoop-yun-zkfc-mini02.out
8.6. 启动YARN
1 #####注意#####：是在mini03上执行start-yarn.sh，把namenode和resourcemanager分开是因为性能问题
2 # 因为他们都要占用大量资源，所以把他们分开了，他们分开了就要分别在不同的机器上启动
3 [yun@mini03 ~]$ start-yarn.sh
4 starting yarn daemons
5 starting resourcemanager, logging to /app/hadoop-2.7.6/logs/yarn-yun-resourcemanager-mini03.out
6 mini06: starting nodemanager, logging to /app/hadoop-2.7.6/logs/yarn-yun-nodemanager-mini06.out
7 mini07: starting nodemanager, logging to /app/hadoop-2.7.6/logs/yarn-yun-nodemanager-mini07.out
8 mini05: starting nodemanager, logging to /app/hadoop-2.7.6/logs/yarn-yun-nodemanager-mini05.out
9
10
11 ################################
12 # 在mini04启动 resourcemanager
13 [yun@mini04 ~]$ yarn-daemon.sh start resourcemanager # 也可用start-yarn.sh
14 starting resourcemanager, logging to /app/hadoop-2.7.6/logs/yarn-yun-resourcemanager-mini04.out
8.7. 启动说明
1 # 第一次启动的时候请严格按照上面的步骤【第一次涉及格式化问题】
2 # 第二次以及之后，步骤为： 启动zookeeper、HDFS、YARN
9. 浏览访问
9.1. Hdfs访问
9.1.1. 正常情况访问
1 http://mini01:50070
1 http://mini02:50070
9.1.2. mini01挂了Active自动切换
1 # mini01操作
2 [yun@mini01 ~]$ jps
3 3584 DFSZKFailoverController
4 3283 NameNode
5 5831 Jps
6 [yun@mini01 ~]$ kill 3283
7 [yun@mini01 ~]$ jps
8 3584 DFSZKFailoverController
9 5893 Jps
Namenode挂了所以mini01不能访问
1 http://mini02:50070
可见Hadoop已经切换过去了，之后mini01即使起来了，状态也只能为standby 。
9.2. Yarn访问
1 http://mini03:8088
1 http://mini04:8088
2 会直接跳转到http://mini03:8088/
# 该图从其他地方截取，所以不怎么匹配
1 # Linux下访问
2 [yun@mini01 ~]$ curl mini04:8088
3 This is standby RM. The redirect url is: http://mini03:8088/
HA完毕
10. 集群运维测试
10.1. Haadmin与状态切换管理
1 [yun@mini01 ~]$ hdfs haadmin
2 Usage: haadmin
3 [-transitionToActive [--forceactive] <serviceId>]
4 [-transitionToStandby <serviceId>]
5 [-failover [--forcefence] [--forceactive] <serviceId> <serviceId>]
6 [-getServiceState <serviceId>]
7 [-checkHealth <serviceId>]
8 [-help <command>]
9
10 Generic options supported are
11 -conf <configuration file> specify an application configuration file
12 -D <property=value> use value for given property
13 -fs <local|namenode:port> specify a namenode
14 -jt <local|resourcemanager:port> specify a ResourceManager
15 -files <comma separated list of files> specify comma separated files to be copied to the map reduce cluster
16 -libjars <comma separated list of jars> specify comma separated jar files to include in the classpath.
17 -archives <comma separated list of archives> specify comma separated archives to be unarchived on the compute machines.
18
19 The general command line syntax is
20 bin/hadoop command [genericOptions] [commandOptions]
可以看到，状态操作的命令示例：
1 # 查看namenode工作状态
2 hdfs haadmin -getServiceState nn1
3
4 # 将standby状态namenode切换到active
5 hdfs haadmin -transitionToActive nn1
6
7 # 将active状态namenode切换到standby
8 hdfs haadmin -transitionToStandby nn2
10.2. 测试集群工作状态的一些指令
1 测试集群工作状态的一些指令 ：
2 hdfs dfsadmin -report 查看hdfs的各节点状态信息
3 hdfs haadmin -getServiceState nn1 # hdfs haadmin -getServiceState nn2 获取一个namenode节点的HA状态
4 hadoop-daemon.sh start namenode 单独启动一个namenode进程
5 hadoop-daemon.sh start zkfc 单独启动一个zkfc进程
10.3. Datanode动态上下线
Datanode动态上下线很简单，步骤如下：
a) 准备一台服务器，设置好环境
b) 部署hadoop的安装包，并同步集群配置
c) 联网上线，新datanode会自动加入集群
d) 如果是一次增加大批datanode，还应该做集群负载重均衡
10.4. 数据块的balance
启动balancer的命令：
1 start-balancer.sh -threshold 8
运行之后，会有Balancer进程出现：
上述命令设置了Threshold为8%，那么执行balancer命令的时候，首先统计所有DataNode的磁盘利用率的均值，然后判断如果某一个DataNode的磁盘利用率超过这个均值Threshold，那么将会把这个DataNode的block转移到磁盘利用率低的DataNode，这对于新节点的加入来说十分有用。Threshold的值为1到100之间，不显示的进行参数设置的话，默认是10。