环境:centos + hadoop2.5.2 +scala-2.10.5 + spark1.3.1
1、从http://spark.apache.org/downloads.html 下载编译好的spark
2、准备scala
从http://www.scala-lang.org/ 下载scala-2.10.5.rpm。不下载2.11,因为下载2.11要重新编译spark
安装scala
rpm -ivh scala-2.10.5.rpm
3、解压spark
tar -zxvf spark-1.3.1-bin-hadoop2.4.tar.gz
4、配置环境变量
在/etc/profile最后面增加
export SPARK_HOME=/usr/local/spark-1.3.1-bin-hadoop2.4
export PATH=$PATH:$SPARK_HOME/bin
# 生效
source /etc/profile
5、配置spark
vi /usr/local/spark-1.3.1-bin-hadoop2.4/conf/spark-env.sh
在最后面增加:
export JAVA_HOME=/usr/java/jdk1.7.0_76
export SPARK_MASTER_IP=192.168.1.21
export SPARK_WORKER_MEMORY=2g
export HADOOP_CONF_DIR=/usr/local/hadoop-2.5.2/etc/hadoop
6、配置slave节点
vi /usr/local/spark-1.3.1-bin-hadoop2.4/conf/slaves
master
slaver1
7、复制配置文件到slave节点
scp -r /usr/local/spark-1.3.1-bin-hadoop2.4/ root@slaver1:/usr/local/
8、启动集群
cd /usr/local/spark-1.3.1-bin-hadoop2.4/sbin
./start-all.sh
9、查看集群是否启动成功
jps
# master 查看是否有:Master,Worker
# slaver 查看是否有:Worker
参考:http://stark-summer.iteye.com/blog/2173219