投稿作者:赵海军
现就职于一家创业公司任职运维兼DBA,曾就职于猎豹移动,负责数据库团队,运维前线作者之一。
一、概要
由于工作需要,最近一段时间开始接触学习hadoop相关的东西,目前公司的实时任务和离线任务都跑在一个hadoop集群,离线任务的特点就是每天定时跑,任务跑完了资源就空闲了,为了合理的利用资源,我们打算在搭一个集群用于跑离线任务,计算节点和储存节点分离,计算节点结合aws的Auto Scaling(自动扩容、缩容服务)以及竞价实例,动态调整,在跑任务的时候拉起一批实例,任务跑完就自动释放掉服务器,本文记录下hadoop集群的搭建过程,方便自己日后查看,也希望能帮到初学者,本文所有软件都是通过yum安装,大家也可以下载相应的二进制文件进行安装,使用哪种方式安装,从属个人习惯。
二、环境
1、角色介绍
10.10.103.246 namenode zkfc journalNode QuorumaPeerMain datanode resourcemanager nodemanager WebAppProxyServer JobHistoryServer
10.10.103.144 namenode zkfc journalNode QuorumaPeerMain datanode resourcemanager nodemanager WebAppProxyServer
10.10.103.62 zkfc journalNode QuorumaPeerMain datanode nodemanager
10.10.20.130 client
2、基础环境说明
a、系统版本
我们用的是aws的ec2,用的aws自己定制过的系统,不过和redhat基本相同,内核版本:4.9.20-10.30.amzn1.x86_64
b、java版本
java version “1.8.0_121”
c、hadoop版本
hadoop-2.6.0
d、cdh版本
cdh5.11.0
e、关于主机名,因为我这里用的aws的ec2,默认已有主机名,并且内网可以解析,故就不单独做主机名的配置了,如果你的主机名内网不能解析,请一定要配置主机名,集群内部通讯很多组件使用的是主机名
三、配置部署
1、设置yum源
vim /etc/yum.repos.d/cloudera.repo
[cloudera-cdh5-11-0]
# Packages for Cloudera’s Distribution for Hadoop, Version 5.11.0, on RedHat or CentOS 6 x86_64
name=Cloudera’s Distribution for Hadoop, Version 5.11.0
baseurl=http://archive.cloudera.com/cdh5/redhat/6/x86_64/cdh/5.11.0/
gpgkey=http://archive.cloudera.com/cdh5/redhat/6/x86_64/cdh/RPM-GPG-KEY-cloudera
gpgcheck=1
[cloudera-gplextras5b2]
# Packages for Cloudera’s GPLExtras, Version 5.11.0, on RedHat or CentOS 6 x86_64
name=Cloudera’s GPLExtras, Version 5.11.0
baseurl=http://archive.cloudera.com/gplextras5/redhat/6/x86_64/gplextras/5.11.0/
gpgkey=http://archive.cloudera.com/gplextras5/redhat/6/x86_64/gplextras/RPM-GPG-KEY-cloudera
gpgcheck=1
PS:我这里安装的5.11.0,如果想安装低版本或者高版本,根据自己的需求修改版本号即可
2、安装配置zookeeper集群
yum -y install zookeeper zookeeper-server
vi /etc/zookeeper/conf/zoo.cfg
tickTime=2000
initLimit=10
syncLimit=5
dataDir=/data/zookeeper
clientPort=2181
maxClientCnxns=0
server.1=10.10.103.144:2888:3888
server.2=10.10.103.226:2888:3888
server.3=10.10.103.62:2888:3888
autopurge.snapRetainCount=3
autopurge.purgeInterval=1
mkdir /data/zookeeper #创建datadir目录
/etc/init.d/zookeeper-server init #所有节点先初始化
echo 1 > /data/zookeeper/myid #10.10.103.144上操作
echo 2 > /data/zookeeper/myid #10.10.103.226上操作
echo 3 > /data/zookeeper/myid #10.10.103.62上操作
/etc/init.d/zookeeper-server #启动服务
/usr/lib/zookeeper/bin/zkServer.sh status #查看所有节点状态,其中只有一个节点是Mode: leader就正常 了
3、安装
a、10.10.103.246和10.10.103.144安装
yum -y install hadoop hadoop-client hadoop-hdfs hadoop-hdfs-namenode hadoop-hdfs-zkfc hadoop-hdfs-journalnode hadoop-hdfs-datanode hadoop-mapreduce-historyserver hadoop-yarn-nodemanager hadoop-yarn-proxyserver hadoop-yarn hadoop-mapreduce hadoop-yarn-resourcemanager hadoop-lzo* impala-lzo
b、10.10.103.62上安装
yum -y install hadoop hadoop-client hadoop-hdfs hadoop-hdfs-journalnode hadoop-hdfs-datanode hadoop-lzo* impala-lzo hadoop-yarn hadoop-mapreduce hadoop-yarn-nodemanager
PS:
1、一般小公司,计算节点(ResourceManager)和储存节点(NameNode)的主节点部署在两台服务器上做HA,计算节点(NodeManager)和储存节点(DataNode)部署在多台服务器上,每台服务器上都启动NodeManager和DataNode服务。
2、如果大集群,可能需要计算资源和储存资源分离,集群的各个角色都有服务器单独部署,个人建议划分如下:
a、储存节点
NameNode:
需要安装hadoop hadoop-client hadoop-hdfs hadoop-hdfs-namenode hadoop-hdfs-zkfc hadoop-lzo* impala-lzo
DataNode:
需要安装hadoop hadoop-client hadoop-hdfs hadoop-hdfs-datanode hadoop-lzo* impala-lzo
QJM集群:
需要安装hadoop hadoop-hdfs hadoop-hdfs-journalnode zookeeper zookeeper-server
b、计算节点
ResourceManager:
需要安装hadoop hadoop-client hadoop-yarn hadoop-mapreduce hadoop-yarn-resourcemanager
WebAppProxyServer:
需要安装 hadoop hadoop-yarn hadoop-mapreduce hadoop-yarn-proxyserver
JobHistoryServer:
需要安装 hadoop hadoop-yarn hadoop-mapreduce hadoop-mapreduce-historyserver
NodeManager:
需要安装hadoop hadoop-client hadoop-yarn hadoop-mapreduce hadoop-yarn-nodemanager
4、配置
a、创建目录并设置权限
mkdir -p /data/hadoop/dfs/nn #datanode上操作
chown hdfs:hdfs /data/hadoop/dfs/nn/ -R #datanode上操作
mkdir -p /data/hadoop/dfs/dn #namenode上操作
chown hdfs:hdfs /data/hadoop/dfs/dn/ -R #namenode上操作
mkdir -p /data/hadoop/dfs/jn #journalnode上操作
chown hdfs:hdfs /data/hadoop/dfs/jn/ -R #journalnode上操作
mkdir /data/hadoop/yarn -p #nodemanager上操作
chown yarn:yarn /data/hadoop/yarn -R #nodemanager上操作
b、撰写配置文件
5、服务启动
a、启动journalnode(三台服务器上都启动)
/etc/init.d/hadoop-hdfs-journalnode start
b、格式化namenode(在其中一台namenode10.10.103.246上操作)
sudo -u hdfs hadoop namenode -format
c、初始化zk中HA的状态(在其中一台namenode10.10.103.246上操作)
sudo -u hdfs hdfs zkfc -formatZK
d、初始化共享Edits文件(在其中一台namenode10.10.103.246上操作)
sudo -u hdfs hdfs namenode -initializeSharedEdits
e、启动10.10.103.246上namenode
/etc/init.d/hadoop-hdfs-namenode start
f、同步源数据并启动10.10.103.144上namenode
sudo -u hdfs hdfs namenode -bootstrapStandby
/etc/init.d/hadoop-hdfs-namenode start
g、在两台namenode上启动zkfc
/etc/init.d/hadoop-hdfs-zkfc start
h、启动datanode(所有机器上操作)
/etc/init.d/hadoop-hdfs-journalnode start
i、在10.10.103.246上启动WebAppProxyServer、JobHistoryServer、httpfs
/etc/init.d/hadoop-yarn-proxyserver start
/etc/init.d/hadoop-mapreduce-historyserver start
/etc/init.d/hadoop-httpfs start
j、在所有机器上启动nodemanager
/etc/init.d/hadoop-yarn-nodemanager restart
四、功能验证
1、hadoop功能
a、查看hdfs根目录
[root@ip-10-10-103-246 ~]# hadoop fs -ls /
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
Found 3 items
drwxr-xr-x – hdfs hdfs 0 2017-05-11 11:40 /tmp
drwxrwx— – mapred hdfs 0 2017-05-11 11:28 /user
drwxr-xr-x – yarn hdfs 0 2017-05-11 11:28 /var
b、上传一个文件到根目录
[root@ip-10-10-103-246 ~]# hadoop fs -put /tmp/test.txt /
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
[root@ip-10-10-103-246 ~]# hadoop fs -ls /
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
Found 4 items
-rw-r–r– 2 root hdfs 22 2017-05-11 15:47 /test.txt
drwxr-xr-x – hdfs hdfs 0 2017-05-11 11:40 /tmp
drwxrwx— – mapred hdfs 0 2017-05-11 11:28 /user
drwxr-xr-x – yarn hdfs 0 2017-05-11 11:28 /var
c、直接删除文件不放回收站
[root@ip-10-10-103-246 ~]# hadoop fs -rm -skipTrash /test.txt
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
Deleted /test.txt
d、跑一个wordcount用例
[root@ip-10-10-103-246 ~]# hadoop fs -put /tmp/test.txt /user/hdfs/rand/
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
[root@ip-10-10-103-246 conf]# sudo -u hdfs hadoop jar /usr/lib/hadoop-mapreduce/hadoop-mapreduce-examples-2.6.0-cdh5.11.0.jar wordcount /user/hdfs/rand/ /tmp
OpenJDK 64-Bit Server VM warning: ignoring option MaxPermSize=128m; support was removed in 8.0
17/05/11 11:40:08 INFO client.ConfiguredRMFailoverProxyProvider: Failing over to 10.10.103.246
17/05/11 11:40:09 INFO input.FileInputFormat: Total input paths to process : 1
17/05/11 11:40:09 INFO lzo.GPLNativeCodeLoader: Loaded native gpl library
17/05/11 11:40:09 INFO lzo.LzoCodec: Successfully loaded & initialized native-lzo library [hadoop-lzo rev 674c65bbf0f779edc3e00a00c953b121f1988fe1]
17/05/11 11:40:09 INFO mapreduce.JobSubmitter: number of splits:1
17/05/11 11:40:09 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1494472050574_0003
17/05/11 11:40:09 INFO impl.YarnClientImpl: Submitted application application_1494472050574_0003
17/05/11 11:40:09 INFO mapreduce.Job: The url to track the job: http://10.10.103.246:8100/proxy/application_1494472050574_0003/
17/05/11 11:40:09 INFO mapreduce.Job: Running job: job_1494472050574_0003
17/05/11 11:40:15 INFO mapreduce.Job: Job job_1494472050574_0003 running in uber mode : false
17/05/11 11:40:15 INFO mapreduce.Job: map 0% reduce 0%
17/05/11 11:40:20 INFO mapreduce.Job: map 100% reduce 0%
17/05/11 11:40:25 INFO mapreduce.Job: map 100% reduce 100%
17/05/11 11:40:25 INFO mapreduce.Job: Job job_1494472050574_0003 completed successfully
17/05/11 11:40:25 INFO mapreduce.Job: Counters: 53
File System Counters
FILE: Number of bytes read=1897
FILE: Number of bytes written=262703
FILE: Number of read operations=0
FILE: Number of large read operations=0
FILE: Number of write operations=0
HDFS: Number of bytes read=6431
HDFS: Number of bytes written=6219
HDFS: Number of read operations=6
HDFS: Number of large read operations=0
HDFS: Number of write operations=2
Job Counters
Launched map tasks=1
Launched reduce tasks=1
Data-local map tasks=1
Total time spent by all maps in occupied slots (ms)=2592
Total time spent by all reduces in occupied slots (ms)=5360
Total time spent by all map tasks (ms)=2592
Total time spent by all reduce tasks (ms)=2680
Total vcore-milliseconds taken by all map tasks=2592
Total vcore-milliseconds taken by all reduce tasks=2680
Total megabyte-milliseconds taken by all map tasks=3981312
Total megabyte-milliseconds taken by all reduce tasks=8232960
Map-Reduce Framework
Map input records=102
Map output records=96
Map output bytes=6586
Map output materialized bytes=1893
Input split bytes=110
Combine input records=96
Combine output records=82
Reduce input groups=82
Reduce shuffle bytes=1893
Reduce input records=82
Reduce output records=82
Spilled Records=164
Shuffled Maps =1
Failed Shuffles=0
Merged Map outputs=1
GC time elapsed (ms)=120
CPU time spent (ms)=1570
Physical memory (bytes) snapshot=501379072
Virtual memory (bytes) snapshot=7842639872
Total committed heap usage (bytes)=525860864
Peak Map Physical memory (bytes)=300183552
Peak Map Virtual memory (bytes)=3244224512
Peak Reduce Physical memory (bytes)=201195520
Peak Reduce Virtual memory (bytes)=4598415360
Shuffle Errors
BAD_ID=0
CONNECTION=0
IO_ERROR=0
WRONG_LENGTH=0
WRONG_MAP=0
WRONG_REDUCE=0
File Input Format Counters
Bytes Read=6321
File Output Format Counters
Bytes Written=6219
[root@ip-10-10-103-246 conf]#
2、namenode高可用验证
查看http://10.10.103.246:50070
查看http://10.10.103.144:50070
停掉10.10.103.246节点的namenode进程,查看10.10.103.144节点是否会提升为active节点
3、resourcemanager高可用验证
查看http://10.10.103.246:8088
查看http://10.10.103.144:8088
resourcemanager进程,在浏览器输入http://10.10.103.144:8088,就不会在跳转了。
五、总结
1、hadoop集群能成本部署完成,这才是开始,后期的维护,业务方问题的解决这些经验需要一点一点积累,多出差多折腾总是好的。
2、对应上面部署的集群后期需要扩容,直接把10.10.103.62这台机器做个镜像,用镜像启动服务器即可,服务会自动启动并且加入到集群
3、云上hadoop集群的成本优化,这里只针对aws而言
a、冷数据存在在s3上,hdfs可以直接支持s3,在hdfs-site.xml里面添加s3的key参数(fs.s3n.awsAccessKeyId和fs.s3n.awsSecretAccessKey)即可,需要注意的是程序上传、下载的逻辑需要多加几个重试机制,s3有时候不稳定会导致上传或者下载不成功
b、使用Auto Scaling服务结合竞价实例,配置扩展策略,比如当cpu大于50%的时候就扩容5台服务器,当cpu小于10%的时候就缩容5台服务器,当然你可以配置更多阶梯级的扩容、缩容策略,Auto Scaling还有一个计划任务的功能,你可以向设置crontab一样设置,让Auto Scaling帮你扩容、缩容服务器。