使用shell脚本一键下载、安装、配置spark（单机版）

1. 把下面的脚本复制保存为/tmp/install_spark.sh文件

#!/bin/bash

# sprak安装版本
sprak_version="2.4.0"
# sprak安装目录
sprak_installDir="/opt/module"
# hadoop安装路径，修改为自己的
HADOOP_HOME=/opt/module/hadoop-3.1.3
# java安装路径，修改为自己的
JAVA_HOME=/opt/module/jdk1.8.0_381

install_spark() {
  local version=$1
  local installDir=$2
  # 下载地址
  local downloadUrl="https://archive.apache.org/dist/spark/spark-$version/spark-$version-bin-hadoop2.7.tgz"
  # 检查安装目录是否存在，不存在则创建
  if [ ! -d "${installDir}" ]; then
    echo "创建安装目录..."
    sudo mkdir -p "${installDir}"
  fi

  if test -f /tmp/spark-"$version"-bin-hadoop2.7.tgz; then
    echo "/tmp/spark-$version-bin-hadoop2.7.tgz文件已存在"
  else
    # 下载并解压 spark
    echo "开始下载spark..."
    wget "$downloadUrl" -P /tmp
  fi
  echo "开始解压spark..."
  tar -zxvf "/tmp/spark-$version-bin-hadoop2.7.tgz" -C "${installDir}"
  mv "${installDir}/spark-${version}-bin-hadoop2.7" "${installDir}/spark-${version}"

  if test -n "$(grep '#SPARK_HOME' ~/.bashrc)"; then
    echo "SPARK_HOME已存在"
  else
    # 设置spark用户环境变量
    echo >> ~/.bashrc
    echo '#SPARK_HOME' >> ~/.bashrc
    echo "export SPARK_HOME=${installDir}/spark-${version}" >> ~/.bashrc
    echo 'export PATH=$PATH:$SPARK_HOME/bin' >> ~/.bashrc
  fi

  # 配置 spark
  echo "配置spark运行环境"
  cp "${installDir}/spark-${version}/conf/spark-env.sh.template" "${installDir}/spark-${version}/conf/spark-env.sh"
  echo 'export SPARK_DIST_CLASSPATH=$('"$HADOOP_HOME"'/bin/hadoop classpath)' >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo "spark日志级别配置"
  cp "${installDir}/spark-${version}/conf/log4j.properties.template" "${installDir}/spark-${version}/conf/log4j.properties"
  sed -i "s|^log4j.rootCategory=.*|log4j.rootCategory=WARN, console|" "${installDir}/spark-${version}/conf/log4j.properties"
  echo "配置spark单机模式"
  echo "export SPARK_MASTER_HOST=$(hostname)    #设置主节点地址" >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo 'export SPARK_WORKER_MEMORY=1g           #设置节点内存大小' >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo 'export SPARK_WORKER_CORES=1             #设置节点参与计算的核心数' >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo 'export SPARK_WORKER_INSTANCES=1         #设置节点实例数' >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo "export JAVA_HOME=$JAVA_HOME" >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo "export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop" >> "${installDir}/spark-${version}/conf/spark-env.sh"
  echo "设置spark节点"
  echo "$(hostname)" > "${installDir}/spark-${version}/conf/slaves"
  echo "删除spark安装包"
  rm -rf /tmp/spark-"$version"-bin-hadoop2.7.tgz
  echo "spark 安装完成！"
}

install_spark "$sprak_version" "$sprak_installDir"

2. 增加执行权限

chmod a+x /tmp/install_spark.sh

3. 执行/tmp/install_spark.sh

/tmp/install_spark.sh

执行之后等待下载、安装、配置完成，如下图：

4. 加载环境变量

source ~/.bashrc

进入spark交互界面

spark-shell

如下图：

5. 启动/停止spark

启动spark

/opt/module/spark-2.4.0/sbin/start-all.sh

查看是否启动成功

jps | awk '$NF == "Master" || $NF == "Worker" { print $NF }'

启动成功如下图：

浏览器访问sparkUI界面：http://192.168.145.103:8080/
如下图：

停止spark

/opt/module/spark-2.4.0/sbin/stop-all.sh