搭建Spark的单机版集群

一、创建用户

# useradd spark

# passwd spark

二、下载软件

JDK，Scala，SBT，Maven

版本信息如下：

JDK jdk-7u79-linux-x64.gz

Scala scala-2.10.5.tgz

SBT sbt-0.13.7.zip

Maven apache-maven-3.2.5-bin.tar.gz

注意：如果只是安装Spark环境，则只需JDK和Scala即可，SBT和Maven是为了后续的源码编译。

三、解压上述文件并进行环境变量配置

# cd /usr/local/

# tar xvf /root/jdk-7u79-linux-x64.gz

# tar xvf /root/scala-2.10.5.tgz

# tar xvf /root/apache-maven-3.2.5-bin.tar.gz

# unzip /root/sbt-0.13.7.zip

修改环境变量的配置文件

# vim /etc/profile

export JAVA_HOME=/usr/local/jdk1.7.0_79export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jarexport SCALA_HOME=/usr/local/scala-2.10.5export MAVEN_HOME=/usr/local/apache-maven-3.2.5export SBT_HOME=/usr/local/sbtexport PATH=$PATH:$JAVA_HOME/bin:$SCALA_HOME/bin:$MAVEN_HOME/bin:$SBT_HOME/bin

使配置文件生效

# source /etc/profile

测试环境变量是否生效

# java –version

java version "1.7.0_79"Java(TM) SE Runtime Environment (build 1.7.0_79-b15)Java HotSpot(TM) 64-Bit Server VM (build 24.79-b02, mixed mode)

# scala –version

Scala code runner version 2.10.5 -- Copyright 2002-2013, LAMP/EPFL

# mvn –version

Apache Maven 3.2.5 (12a6b3acb947671f09b81f49094c53f426d8cea1; 2014-12-15T01:29:23+08:00)Maven home: /usr/local/apache-maven-3.2.5Java version: 1.7.0_79, vendor: Oracle CorporationJava home: /usr/local/jdk1.7.0_79/jreDefault locale: en_US, platform encoding: UTF-8OS name: "linux", version: "3.10.0-229.el7.x86_64", arch: "amd64", family: "unix"

# sbt --version

sbt launcher version 0.13.7

四、主机名绑定

[root@spark01 ~]# vim /etc/hosts

192.168.244.147 spark01

五、配置spark

切换到spark用户下

下载hadoop和spark，可使用wget命令下载

spark-1.4.0 http://d3kbcqa49mib13.cloudfront.net/spark-1.4.0-bin-hadoop2.6.tgz

Hadoop http://mirror.bit.edu.cn/apache/hadoop/common/hadoop-2.6.0/hadoop-2.6.0.tar.gz

解压上述文件并进行环境变量配置

修改spark用户环境变量的配置文件

[spark@spark01 ~]$ vim .bash_profile

export SPARK_HOME=$HOME/spark-1.4.0-bin-hadoop2.6export HADOOP_HOME=$HOME/hadoop-2.6.0export HADOOP_CONF_DIR=$HOME/hadoop-2.6.0/etc/hadoopexport PATH=$PATH:$SPARK_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin

使配置文件生效

[spark@spark01 ~]$ source .bash_profile

修改spark配置文件

[spark@spark01 ~]$ cd spark-1.4.0-bin-hadoop2.6/conf/

[spark@spark01 conf]$ cp spark-env.sh.template spark-env.sh

[spark@spark01 conf]$ vim spark-env.sh

在后面添加如下内容：

export SCALA_HOME=/usr/local/scala-2.10.5export SPARK_MASTER_IP=spark01export SPARK_WORKER_MEMORY=1500mexport JAVA_HOME=/usr/local/jdk1.7.0_79

有条件的童鞋可将SPARK_WORKER_MEMORY适当设大一点，因为我虚拟机内存是2G，所以只给了1500m。

配置slaves

[spark@spark01 conf]$ cp slaves slaves.template

[spark@spark01 conf]$ vim slaves

将localhost修改为spark01

启动master

[spark@spark01 spark-1.4.0-bin-hadoop2.6]$ sbin/start-master.sh

starting org.apache.spark.deploy.master.Master, logging to /home/spark/spark-1.4.0-bin-hadoop2.6/sbin/../logs/spark-spark-org.apache.spark.deploy.master.Master-1-spark01.out

查看上述日志的输出内容

[spark@spark01 spark-1.4.0-bin-hadoop2.6]$ cd logs/

[spark@spark01 logs]$ cat spark-spark-org.apache.spark.deploy.master.Master-1-spark01.out

Spark Command: /usr/local/jdk1.7.0_79/bin/java -cp /home/spark/spark-1.4.0-bin-hadoop2.6/sbin/../conf/:/home/spark/spark-1.4.0-bin-hadoop2.6/lib/spark-assembly-1.4.0-hadoop2.6.0.jar:/home/spark/spark-1.4.0-bin-hadoop2.6/lib/datanucleus-core-3.2.10.jar:/home/spark/spark-1.4.0-bin-hadoop2.6/lib/datanucleus-api-jdo-3.2.6.jar:/home/spark/spark-1.4.0-bin-hadoop2.6/lib/datanucleus-rdbms-3.2.9.jar:/home/spark/hadoop-2.6.0/etc/hadoop/ -Xms512m -Xmx512m -XX:MaxPermSize=128m org.apache.spark.deploy.master.Master --ip spark01 --port 7077 --webui-port 8080========================================16/01/16 15:12:30 INFO master.Master: Registered signal handlers for [TERM, HUP, INT]16/01/16 15:12:31 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable16/01/16 15:12:32 INFO spark.SecurityManager: Changing view acls to: spark16/01/16 15:12:32 INFO spark.SecurityManager: Changing modify acls to: spark16/01/16 15:12:32 INFO spark.SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(spark); users with modify permissions: Set(spark)16/01/16 15:12:33 INFO slf4j.Slf4jLogger: Slf4jLogger started16/01/16 15:12:33 INFO Remoting: Starting remoting16/01/16 15:12:33 INFO Remoting: Remoting started; listening on addresses :[akka.tcp://sparkMaster@spark01:7077]16/01/16 15:12:33 INFO util.Utils: Successfully started service 'sparkMaster' on port 7077.16/01/16 15:12:34 INFO server.Server: jetty-8.y.z-SNAPSHOT16/01/16 15:12:34 INFO server.AbstractConnector: Started SelectChannelConnector@spark01:606616/01/16 15:12:34 INFO util.Utils: Successfully started service on port 6066.16/01/16 15:12:34 INFO rest.StandaloneRestServer: Started REST server for submitting applications on port 606616/01/16 15:12:34 INFO master.Master: Starting Spark master at spark://spark01:707716/01/16 15:12:34 INFO master.Master: Running Spark version 1.4.016/01/16 15:12:34 INFO server.Server: jetty-8.y.z-SNAPSHOT16/01/16 15:12:34 INFO server.AbstractConnector: Started SelectChannelConnector@0.0.0.0:808016/01/16 15:12:34 INFO util.Utils: Successfully started service 'MasterUI' on port 8080.16/01/16 15:12:34 INFO ui.MasterWebUI: Started MasterWebUI at http://192.168.244.147:808016/01/16 15:12:34 INFO master.Master: I have been elected leader! New state: ALIVE

从日志中也可看出，master启动正常

下面来看看master的 web管理界面，默认在8080端口

启动worker

[spark@spark01 spark-1.4.0-bin-hadoop2.6]$ sbin/start-slaves.sh spark://spark01:7077

spark01: Warning: Permanently added 'spark01,192.168.244.147' (ECDSA) to the list of known hosts.spark@spark01's password:spark01: starting org.apache.spark.deploy.worker.Worker, logging to /home/spark/spark-1.4.0-bin-hadoop2.6/sbin/../logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-spark01.out

输入spark01上spark用户的密码

可通过日志的信息来确认workder是否正常启动，因信息太多，在这里就不贴出了。

[spark@spark01 spark-1.4.0-bin-hadoop2.6]$ cd logs/

[spark@spark01 logs]$ cat spark-spark-org.apache.spark.deploy.worker.Worker-1-spark01.out

启动spark shell

[spark@spark01 spark-1.4.0-bin-hadoop2.6]$ bin/spark-shell --master spark://spark01:7077

16/01/16 15:33:17 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable16/01/16 15:33:18 INFO spark.SecurityManager: Changing view acls to: spark16/01/16 15:33:18 INFO spark.SecurityManager: Changing modify acls to: spark16/01/16 15:33:18 INFO spark.SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(spark); users with modify permissions: Set(spark)16/01/16 15:33:18 INFO spark.HttpServer: Starting HTTP Server16/01/16 15:33:18 INFO server.Server: jetty-8.y.z-SNAPSHOT16/01/16 15:33:18 INFO server.AbstractConnector: Started SocketConnector@0.0.0.0:4230016/01/16 15:33:18 INFO util.Utils: Successfully started service 'HTTP class server' on port 42300.Welcome to   ____       __   / __/__ ___ _____/ /__  _\ \/ _ \/ _ `/ __/ '_/  /___/ .__/\_,_/_/ /_/\_\  version 1.4.0   /_/Using Scala version 2.10.4 (Java HotSpot(TM) 64-Bit Server VM, Java 1.7.0_79)Type in expressions to have them evaluated.Type :help for more information.16/01/16 15:33:30 INFO spark.SparkContext: Running Spark version 1.4.016/01/16 15:33:30 INFO spark.SecurityManager: Changing view acls to: spark16/01/16 15:33:30 INFO spark.SecurityManager: Changing modify acls to: spark16/01/16 15:33:30 INFO spark.SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(spark); users with modify permissions: Set(spark)16/01/16 15:33:31 INFO slf4j.Slf4jLogger: Slf4jLogger started16/01/16 15:33:31 INFO Remoting: Starting remoting16/01/16 15:33:31 INFO Remoting: Remoting started; listening on addresses :[akka.tcp://sparkDriver@192.168.244.147:43850]16/01/16 15:33:31 INFO util.Utils: Successfully started service 'sparkDriver' on port 43850.16/01/16 15:33:31 INFO spark.SparkEnv: Registering MapOutputTracker16/01/16 15:33:31 INFO spark.SparkEnv: Registering BlockManagerMaster16/01/16 15:33:31 INFO storage.DiskBlockManager: Created local directory at /tmp/spark-7b7bd4bd-ff20-4e3d-a354-61a4ca7c4b2f/blockmgr-0e855210-3609-4204-b5e3-151e0c096c1516/01/16 15:33:31 INFO storage.MemoryStore: MemoryStore started with capacity 265.4 MB16/01/16 15:33:31 INFO spark.HttpFileServer: HTTP File server directory is /tmp/spark-7b7bd4bd-ff20-4e3d-a354-61a4ca7c4b2f/httpd-56ac16d2-dd82-41cb-99d7-4d11ef36b42e16/01/16 15:33:31 INFO spark.HttpServer: Starting HTTP Server16/01/16 15:33:31 INFO server.Server: jetty-8.y.z-SNAPSHOT16/01/16 15:33:31 INFO server.AbstractConnector: Started SocketConnector@0.0.0.0:4763316/01/16 15:33:31 INFO util.Utils: Successfully started service 'HTTP file server' on port 47633.16/01/16 15:33:31 INFO spark.SparkEnv: Registering OutputCommitCoordinator16/01/16 15:33:31 INFO server.Server: jetty-8.y.z-SNAPSHOT16/01/16 15:33:31 INFO server.AbstractConnector: Started SelectChannelConnector@0.0.0.0:404016/01/16 15:33:31 INFO util.Utils: Successfully started service 'SparkUI' on port 4040.16/01/16 15:33:31 INFO ui.SparkUI: Started SparkUI at http://192.168.244.147:404016/01/16 15:33:32 INFO client.AppClient$ClientActor: Connecting to master akka.tcp://sparkMaster@spark01:7077/user/Master...16/01/16 15:33:33 INFO cluster.SparkDeploySchedulerBackend: Connected to Spark cluster with app ID app-20160116153332-000016/01/16 15:33:33 INFO client.AppClient$ClientActor: Executor added: app-20160116153332-0000/0 on worker-20160116152314-192.168.244.147-58914 (192.168.244.147:58914) with 2 cores16/01/16 15:33:33 INFO cluster.SparkDeploySchedulerBackend: Granted executor ID app-20160116153332-0000/0 on hostPort 192.168.244.147:58914 with 2 cores, 512.0 MB RAM16/01/16 15:33:33 INFO client.AppClient$ClientActor: Executor updated: app-20160116153332-0000/0 is now LOADING16/01/16 15:33:33 INFO client.AppClient$ClientActor: Executor updated: app-20160116153332-0000/0 is now RUNNING16/01/16 15:33:34 INFO util.Utils: Successfully started service 'org.apache.spark.network.netty.NettyBlockTransferService' on port 33146.16/01/16 15:33:34 INFO netty.NettyBlockTransferService: Server created on 3314616/01/16 15:33:34 INFO storage.BlockManagerMaster: Trying to register BlockManager16/01/16 15:33:34 INFO storage.BlockManagerMasterEndpoint: Registering block manager 192.168.244.147:33146 with 265.4 MB RAM, BlockManagerId(driver, 192.168.244.147, 33146)16/01/16 15:33:34 INFO storage.BlockManagerMaster: Registered BlockManager16/01/16 15:33:34 INFO cluster.SparkDeploySchedulerBackend: SchedulerBackend is ready for scheduling beginning after reached minRegisteredResourcesRatio: 0.016/01/16 15:33:34 INFO repl.SparkILoop: Created spark context..Spark context available as sc.16/01/16 15:33:38 INFO hive.HiveContext: Initializing execution hive, version 0.13.116/01/16 15:33:43 INFO metastore.HiveMetaStore: 0: Opening raw store with implemenation class:org.apache.hadoop.hive.metastore.ObjectStore16/01/16 15:33:43 INFO metastore.ObjectStore: ObjectStore, initialize called16/01/16 15:33:44 INFO DataNucleus.Persistence: Property datanucleus.cache.level2 unknown - will be ignored16/01/16 15:33:44 INFO DataNucleus.Persistence: Property hive.metastore.integral.jdo.pushdown unknown - will be ignored16/01/16 15:33:44 INFO cluster.SparkDeploySchedulerBackend: Registered executor: AkkaRpcEndpointRef(Actor[akka.tcp://sparkExecutor@192.168.244.147:46741/user/Executor#-2043358626]) with ID 016/01/16 15:33:44 WARN DataNucleus.Connection: BoneCP specified but not present in CLASSPATH (or one of dependencies)16/01/16 15:33:45 INFO storage.BlockManagerMasterEndpoint: Registering block manager 192.168.244.147:33017 with 265.4 MB RAM, BlockManagerId(0, 192.168.244.147, 33017)16/01/16 15:33:46 WARN DataNucleus.Connection: BoneCP specified but not present in CLASSPATH (or one of dependencies)16/01/16 15:33:48 INFO metastore.ObjectStore: Setting MetaStore object pin classes with hive.metastore.cache.pinobjtypes="Table,StorageDescriptor,SerDeInfo,Partition,Database,Type,FieldSchema,Order"16/01/16 15:33:48 INFO metastore.MetaStoreDirectSql: MySQL check failed, assuming we are not on mysql: Lexical error at line 1, column 5. Encountered: "@" (64), after : "".16/01/16 15:33:52 INFO DataNucleus.Datastore: The class "org.apache.hadoop.hive.metastore.model.MFieldSchema" is tagged as "embedded-only" so does not have its own datastore table.16/01/16 15:33:52 INFO DataNucleus.Datastore: The class "org.apache.hadoop.hive.metastore.model.MOrder" is tagged as "embedded-only" so does not have its own datastore table.16/01/16 15:33:54 INFO DataNucleus.Datastore: The class "org.apache.hadoop.hive.metastore.model.MFieldSchema" is tagged as "embedded-only" so does not have its own datastore table.16/01/16 15:33:54 INFO DataNucleus.Datastore: The class "org.apache.hadoop.hive.metastore.model.MOrder" is tagged as "embedded-only" so does not have its own datastore table.16/01/16 15:33:54 INFO metastore.ObjectStore: Initialized ObjectStore16/01/16 15:33:54 WARN metastore.ObjectStore: Version information not found in metastore. hive.metastore.schema.verification is not enabled so recording the schema version 0.13.1aa16/01/16 15:33:55 INFO metastore.HiveMetaStore: Added admin role in metastore16/01/16 15:33:55 INFO metastore.HiveMetaStore: Added public role in metastore16/01/16 15:33:56 INFO metastore.HiveMetaStore: No user is added in admin role, since config is empty16/01/16 15:33:56 INFO session.SessionState: No Tez session required at this point. hive.execution.engine=mr.16/01/16 15:33:56 INFO repl.SparkILoop: Created sql context (with Hive support)..SQL context available as sqlContext.scala>

打开spark shell以后，可以写一个简单的程序，say hello to the world

scala> println("helloworld")helloworld

再来看看spark的web管理界面，可以看出，多了一个Workders和Running Applications的信息

至此，Spark的伪分布式环境搭建完毕，

有以下几点需要注意：

1. 上述中的Maven和SBT是非必须的，只是为了后续的源码编译，所以，如果只是单纯的搭建Spark环境，可不用下载Maven和SBT。

2. 该Spark的伪分布式环境其实是集群的基础，只需修改极少的地方，然后copy到slave节点上即可，鉴于篇幅有限，后文再表。 

原标题：搭建Spark的单机版集群

关键词：