部署Spark-assembly Jar包
下载spark-1.6.3-bin-hadoop2.6.tgz，下载地址如下：
https://www.apache.org/dyn/closer.lua/spark/spark-1.6.3/spark-1.6.3-bin-hadoop2.6.tgz
将下载的spark-1.6.3-bin-hadoop2.6.tgz上传至集群的任意节点并解压，这里以cdh02节点为例
[root@cdh02 ~]# tar -zxvf spark-1.6.3-bin-hadoop2.6.tgz
将解压出来的spark-assembly-1.6.3-hadoop2.6.0.jar拷贝至CDH的jars目录
<pre style="margin: 0px; padding: 0px; max-width: 100%; box-sizing: border-box !important; overflow-wrap: break-word !important; color: rgb(51, 51, 51); font-size: 17px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: 0.544px; orphans: 2; text-align: justify; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; text-decoration-style: initial; text-decoration-color: initial; background-color: rgb(202, 230, 202);">[root@cdh02 spark-1.6.3-bin-hadoop2.6]# scp root/spark-1.6.3-bin-hadoop2.6/lib/spark-assembly-1.6.3-hadoop2.6.0.jar opt/cloudera/parcels/CDH/jars/</pre>
替换CDH中spark默认的spark-assembly jar包
将spark-assembly-1.6.3-hadoop2.6.0.jar包上传至HDFS目录
在CM上对Spark进行配置，保存配置并重启Spark服务。
修改/etc/spark/conf/ classpath.txt文件在末尾增加如下内容
由于CDH5.11以后版本，Navigator2.10增加了Spark的血缘分析，所以这里需要添加spark-lineage_2.10-1.6.0-cdh5.13.0.jar包，否则连接Spark会报错找不到com.cloudera.spark.lineage.ClouderaNavigatorListener类。
<pre style="margin: 0px; padding: 0px; max-width: 100%; box-sizing: border-box !important; overflow-wrap: break-word !important; color: rgb(51, 51, 51); font-size: 17px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: 0.544px; orphans: 2; text-align: justify; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; text-decoration-style: initial; text-decoration-color: initial; background-color: rgb(202, 230, 202);">/opt/cloudera/parcels/CDH-5.13.0-1.cdh5.13.0.p0.29/jars/spark-lineage_2.10-1.6.0-cdh5.13.0.jar</pre>
部署Spark ThriftServer启动和停止脚本
拷贝Spark ThriftServer启动和停止脚本
将 spark-1.6.3-bin-hadoop2.6/sbin/目录下的 start-thriftserver.sh 和 stop-thriftserver.sh 脚本拷贝到/opt/cloudera/parcels/CDH/lib/spark/sbin目录下，并设置执行权限。
[root@cdh02 sbin]# scp start-thriftserver.sh stop-thriftserver.sh opt/cloudera/parcels/CDH/lib/spark/sbin/
[root@cdh02 sbin]# chmod +x opt/cloudera/parcels/CDH/lib/spark/sbin/*thriftserver.sh
[root@cdh02 sbin]# ll opt/cloudera/parcels/CDH/lib/spark/sbin/
修改 load-spark-env.sh 脚本，这个脚本是启动 spark 相关服务时加载环境变量信息的
[root@ip-172-31-5-190 sbin]# cd opt/cloudera/parcels/CDH/lib/spark/bin
[root@ip-172-31-5-190 bin]# pwd
/opt/cloudera/parcels/CDH/lib/spark/bin
[root@ip-172-31-5-190 bin]#
## 将注释掉exec "$SPARK_HOME/bin/$SCRIPT""$@"，因为在start-thriftserver.sh脚本中会执行这个命令
测试Spark Thrift
使用beeline通过JDBC连接Spark，可以发现连接的是Spark SQL
[root@ip-172-31-5-190 ~]# beeline
beeline> !connect jdbc:hive2://ip-172-31-5-190:10000
Enter username for jdbc:hive2://ip-172-31-5-190:10000: hive
Enter password for jdbc:hive2://ip-172-31-5-190:10000:
运行SQL测试
<pre style="margin: 0px; padding: 0px; max-width: 100%; box-sizing: border-box !important; overflow-wrap: break-word !important; color: rgb(51, 51, 51); font-size: 17px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: 0.544px; orphans: 2; text-align: justify; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; text-decoration-style: initial; text-decoration-color: initial; background-color: rgb(202, 230, 202);">*0:* jdbc:hive2://ip-172-31-5-190:10000*> show* tables;
0: jdbc:hive2://ip-172-31-5-190:10000*>* **select** count(*) *from* test_table;
0: jdbc:hive2://ip-172-31-5-190:10000*>* **select** * from test;
0: jdbc:hive2://ip-172-31-5-190:10000*>* **select** * from test_table;
0: jdbc:hive2://ip-172-31-5-190:10000*>* **select** count(test_table.s2) *from* test_table join test on test_table.s1=test.s1;</pre>
在Yarn的8088中查看Spark任务，可以发现都是通过Spark执行的。
总结
CDH默认不支持Spark Thrift，需要使用Spark原生的spark-assembly jar包替换CDH自带的jar包
CDH5.11版本以后，Navigator2.10增加了Spark的血缘分析，所以需要将spark-lineage的jar加载的Spark的运行环境。
否则连接Spark会报错找不到com.cloudera.spark.lineage.ClouderaNavigatorListener类。
关注公众号：领取精彩视频课程&海量免费语音课程
