Kafka：ZK+Kafka+Spark Streaming集群环境搭建安装spark2.2.1

Posted 2020-11-20 yy3b2007com

如何配置centos虚拟机请参考《Kafka：ZK+Kafka+Spark Streaming集群环境搭建（一）VMW安装四台CentOS，并实现本机与它们能交互，虚拟机内部实现可以上网。》

如何安装hadoop2.9.0请参考《Kafka：ZK+Kafka+Spark Streaming集群环境搭建（二）安装hadoop2.9.0》

安装spark的服务器：

192.168.0.120      master
192.168.0.121      slave1
192.168.0.122      slave2
192.168.0.123      slave3

从spark官网下载spark安装包：

官网地址：http://spark.apache.org/downloads.html

注意：上一篇文章中我们安装了hadoop2.9.0，但是这里没有发现待下载spark对应的hadoop版本可选项中发现hadoop2.9.0，因此也只能选择“Pre-built for Apache Hadoop 2.7 and later”。

 

这spark可选版本比较多，就选择“2.2.1(Dec 01 2017)”。

选中后，此时带下来的spark安装包版本信息为：

下载“spark-2.2.1-bin-hadoop2.7.tgz”，上传到master的/opt目录下，并解压：

[[email protected] opt]# tar -zxvf spark-2.2.1-bin-hadoop2.7.tgz 
[[email protected] opt]# ls
hadoop-2.9.0  hadoop-2.9.0.tar.gz  jdk1.8.0_171  jdk-8u171-linux-x64.tar.gz  scala-2.11.0  scala-2.11.0.tgz  spark-2.2.1-bin-hadoop2.7  spark-2.2.1-bin-hadoop2.7.tgz
[[email protected] opt]# 

配置Spark

[[email protected] opt]# ls
hadoop-2.9.0  hadoop-2.9.0.tar.gz  jdk1.8.0_171  jdk-8u171-linux-x64.tar.gz  scala-2.11.0  scala-2.11.0.tgz  spark-2.2.1-bin-hadoop2.7  spark-2.2.1-bin-hadoop2.7.tgz
[[email protected] opt]# cd spark-2.2.1-bin-hadoop2.7/conf/
[[email protected] conf]# ls
docker.properties.template  metrics.properties.template   spark-env.sh.template
fairscheduler.xml.template  slaves.template
log4j.properties.template   spark-defaults.conf.template
[[email protected] conf]# scp spark-env.sh.template spark-env.sh
[[email protected] conf]# ls
docker.properties.template  metrics.properties.template   spark-env.sh
fairscheduler.xml.template  slaves.template               spark-env.sh.template
log4j.properties.template   spark-defaults.conf.template
[[email protected] conf]# vi spark-env.sh

在spark-env.sh末尾添加以下内容（这是我的配置，你需要根据自己安装的环境情况自行修改）：

export SCALA_HOME=/opt/scala-2.11.0
export JAVA_HOME=/opt/jdk1.8.0_171
export HADOOP_HOME=/opt/hadoop-2.9.0
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
SPARK_MASTER_IP=master
SPARK_LOCAL_DIRS=/opt/spark-2.2.1-bin-hadoop2.7
SPARK_DRIVER_MEMORY=1G

注：在设置Worker进程的CPU个数和内存大小，要注意机器的实际硬件条件，如果配置的超过当前Worker节点的硬件条件，Worker进程会启动失败。

vi slaves在slaves文件下填上slave主机名：

[[email protected] conf]# scp slaves.template slaves
[[email protected] conf]# vi slaves

配置内容为：

#localhost
slave1
slave2
slave3

将配置好的spark-2.2.1-bin-hadoop2.7文件夹分发给所有slaves吧

scp -r /opt/spark-2.2.1-bin-hadoop2.7 [email protected]:/opt/
scp -r /opt/spark-2.2.1-bin-hadoop2.7 [email protected]:/opt/
scp -r /opt/spark-2.2.1-bin-hadoop2.7 [email protected]:/opt/

 注意：此时默认slave1,slave2,slave3上是没有/opt/spark-2.2.1-bin-hadoop2.7，因此直接拷贝可能会出现无权限操作 。

            解决方案，分别在slave1,slave2,slave3的/opt下创建spark-2.2.1-bin-hadoop2.7，并分配777权限。

[[email protected] opt]# mkdir spark-2.2.1-bin-hadoop2.7
[[email protected] opt]# chmod 777 spark-2.2.1-bin-hadoop2.7
[[email protected] opt]# 

之后，再次操作拷贝就有权限操作了。

启动Spark

在spark安装目录下执行下面命令才行 , 目前的master安装目录在/opt/spark-2.2.1-bin-hadoop2.7

sbin/start-all.sh

此时，我使用非root账户（spark用户名的用户）启动spark，出现master上spark无权限写日志的问题：

[[email protected] opt]$ cd /opt/spark-2.2.1-bin-hadoop2.7
[[email protected] spark-2.2.1-bin-hadoop2.7]$ sbin/start-all.sh
mkdir: cannot create directory ‘/opt/spark-2.2.1-bin-hadoop2.7/logs’: Permission denied
chown: cannot access ‘/opt/spark-2.2.1-bin-hadoop2.7/logs’: No such file or directory
starting org.apache.spark.deploy.master.Master, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out
/opt/spark-2.2.1-bin-hadoop2.7/sbin/spark-daemon.sh: line 128: /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out: No such file or directory
failed to launch: nice -n 0 /opt/spark-2.2.1-bin-hadoop2.7/bin/spark-class org.apache.spark.deploy.master.Master --host master --port 7077 --webui-port 8080
tail: cannot open ‘/opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out’ for reading: No such file or directory
full log in /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out
slave1: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave1.out
slave3: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave3.out
slave2: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave2.out
[[email protected] spark-2.2.1-bin-hadoop2.7]$ cd ..
[[email protected] opt]$ su root
Password: 
[[email protected] opt]# chmod 777 spark-2.2.1-bin-hadoop2.7
[[email protected] opt]# su spark
[[email protected] opt]$ cd spark-2.2.1-bin-hadoop2.7
[[email protected] spark-2.2.1-bin-hadoop2.7]$ sbin/start-all.sh           
starting org.apache.spark.deploy.master.Master, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out
slave2: org.apache.spark.deploy.worker.Worker running as process 3153.  Stop it first.
slave3: org.apache.spark.deploy.worker.Worker running as process 3076.  Stop it first.
slave1: org.apache.spark.deploy.worker.Worker running as process 3241.  Stop it first.
[[email protected] spark-2.2.1-bin-hadoop2.7]$ sbin/stop-all.sh 
slave1: stopping org.apache.spark.deploy.worker.Worker
slave3: stopping org.apache.spark.deploy.worker.Worker
slave2: stopping org.apache.spark.deploy.worker.Worker
stopping org.apache.spark.deploy.master.Master
[[email protected] spark-2.2.1-bin-hadoop2.7]$ sbin/start-all.sh
starting org.apache.spark.deploy.master.Master, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.master.Master-1-master.out
slave1: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave1.out
slave3: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave3.out
slave2: starting org.apache.spark.deploy.worker.Worker, logging to /opt/spark-2.2.1-bin-hadoop2.7/logs/spark-spark-org.apache.spark.deploy.worker.Worker-1-slave2.out

解决方案：给master的spark安装目录也分配777操作权限。

验证 Spark 是否安装成功

用jps检查，在 master 上应该有以下几个进程：

$ jps
7949 Jps
7328 SecondaryNameNode
7805 Master
7137 NameNode
7475 ResourceManager

在 slave 上应该有以下几个进程：

$jps
3132 DataNode
3759 Worker
3858 Jps
3231 NodeManager

进入Spark的Web管理页面： http://192.168.0.120:8080

运行示例

本地方式两线程运行测试：

[[email protected] spark-2.2.1-bin-hadoop2.7]$ cd /opt/spark-2.2.1-bin-hadoop2.7
[[email protected] spark-2.2.1-bin-hadoop2.7]$ ./bin/run-example SparkPi 10 --master local[2]

Spark Standalone 集群模式运行

[[email protected] spark-2.2.1-bin-hadoop2.7]$ cd /opt/spark-2.2.1-bin-hadoop2.7
[[email protected] spark-2.2.1-bin-hadoop2.7]$ ./bin/spark-submit > --class org.apache.spark.examples.SparkPi > --master spark://master:7077 
> examples/jars/spark-examples_2.11-2.2.1.jar > 100

此时是可以从spark监控界面查看到运行状况：

Spark on YARN 集群上 yarn-cluster 模式运行

[[email protected] spark-2.2.1-bin-hadoop2.7]$ cd /opt/spark-2.2.1-bin-hadoop2.7
[[email protected] spark-2.2.1-bin-hadoop2.7]$ ./bin/spark-submit > --class org.apache.spark.examples.SparkPi > --master yarn-cluster > /opt/spark-2.2.1-bin-hadoop2.7/examples/jars/spark-examples_2.11-2.2.1.jar > 10

注意：Spark on YARN 支持两种运行模式，分别为yarn-cluster和yarn-client，具体的区别可以看这篇博文，从广义上讲，yarn-cluster适用于生产环境；而yarn-client适用于交互和调试，也就是希望快速地看到application的输出。