0)停止Standalone模式下的spark集群
[atguigu@hadoop102 spark-standalone]$ sbin/stop-all.sh
[atguigu@hadoop102 spark-standalone]$ zk.sh stop
[atguigu@hadoop103 spark-standalone]$ sbin/stop-master.sh
1)为了防止和Standalone模式冲突,再单独解压一份spark
[atguigu@hadoop102 software]$ tar -zxvf spark-3.0.0-bin-hadoop3.2.tgz -C /opt/module/
2)进入到/opt/module目录,修改spark-3.0.0-bin-hadoop3.2名称为spark-yarn
[atguigu@hadoop102 module]$ mv spark-3.0.0-bin-hadoop3.2/ spark-yarn
3)修改hadoop配置文件/opt/ha/hadoop-3.1.3/etc/hadoop/yarn-site.xml,添加如下内容
因为测试环境虚拟机内存较少,防止执行过程进行被意外杀死,做如下配置
[atguigu@hadoop102 hadoop]$ vim yarn-site.xml
<!--是否启动一个线程检查每个任务正使用的物理内存量,如果任务超出分配值,则直接将其杀掉,默认是true -->
<property>
<name>yarn.nodemanager.pmem-check-enabled</name>
<value>false</value>
</property>
<!--是否启动一个线程检查每个任务正使用的虚拟内存量,如果任务超出分配值,则直接将其杀掉,默认是true -->
<property>
<name>yarn.nodemanager.vmem-check-enabled</name>
<value>false</value>
</property>
4)分发配置文件
[atguigu@hadoop102 conf]$ xsync /opt/ha/hadoop-3.1.3/etc/hadoop/yarn-site.xml
5)修改/opt/module/spark-yarn/conf/spark-env.sh,添加YARN_CONF_DIR配置,保证后续运行任务的路径都变成集群路径
[atguigu@hadoop102 conf]$ mv spark-env.sh.template spark-env.sh
[atguigu@hadoop102 conf]$ vim spark-env.sh
YARN_CONF_DIR=/opt/ha/hadoop-3.1.3/etc/hadoo
6)启动HDFS以及YARN集群
[atguigu@hadoop102 hadoop-3.1.3]$ sbin/start-dfs.sh
[atguigu@hadoop103 hadoop-3.1.3]$ sbin/start-yarn.sh
7)执行一个程序
[atguigu@hadoop102 spark-yarn]$ bin/spark-submit \
--class org.apache.spark.examples.SparkPi \
--master yarn \
./examples/jars/spark-examples_2.12-3.0.0.jar \
10
参数:--master yarn,表示Yarn方式运行;--deploy-mode表示客户端方式运行程序
9)查看hadoop103:8088页面,点击History,查看历史页面