本文为《Flink大数据项目实战》学习笔记,想通过视频系统学习Flink这个最火爆的大数据计算框架的同学,推荐学习课程:
Flink大数据项目实战:http://t.cn/EJtKhaz
1. 创建Flink项目及依赖管理
1.1创建Flink项目
官网创建Flink项目有两种方式:
https://ci.apache.org/projects/flink/flink-docs-release-1.6/quickstart/java_api_quickstart.html
方式一:
mvn archetype:generate \
-DarchetypeGroupId=org.apache.flink \
-DarchetypeArtifactId=flink-quickstart-java \
-DarchetypeVersion=1.6.2
方式二
$ curl https://flink.apache.org/q/quickstart.sh | bash -s 1.6.2
这里我们仍然使用第一种方式创建Flink项目。
打开终端,切换到对应的目录,通过maven创建flink项目
mvn archetype:generate -DarchetypeGroupId=org.apache.flink -DarchetypeArtifactId=flink-quickstart-java -DarchetypeVersion=1.6.2
项目构建过程中需要输入groupId,artifactId,version和package
Flink项目创建成功
打开IDEA工具,点击open。
选择刚刚创建的flink项目
Flink项目已经成功导入IDEA开发工具
通过maven打包测试运行
mvn clean package
刷新target目录可以看到刚刚打包的flink项目
1.2. Flink依赖
Core Dependencies(核心依赖):
1.核心依赖打包在flink-dist*.jar里
2.包含coordination, networking, checkpoints, failover, APIs, operations (such as windowing), resource management等必须的依赖
注意:核心依赖不会随着应用打包(<scope>provided</scope>)
3.核心依赖项尽可能小,并避免依赖项冲突
Pom文件中添加核心依赖
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.6.2</version>
<scope>provided</scope>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.11</artifactId>
<version>1.6.2</version>
<scope>provided</scope>
</dependency>
注意:不会随着应用打包。
User Application Dependencies(应用依赖):
connectors, formats, or libraries(CEP, SQL, ML)、
注意:应用依赖会随着应用打包(scope保持默认值就好)
Pom文件中添加应用依赖
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-connector-kafka-0.10_2.11</artifactId>
<version>1.6.2</version>
</dependency>
注意:应用依赖按需选择,会随着应用打包,可以通过Maven Shade插件进行打包。
1.3. 关于Scala版本
Scala各版本之间是不兼容的(你基于Scala2.12开发Flink应用就不能依赖Scala2.11的依赖包)。
只使用Java的开发人员可以选择任何Scala版本,Scala开发人员需要选择与他们的应用程序的Scala版本匹配的Scala版本。
1.4. Hadoop依赖
不要把Hadoop依赖直接添加到Flink application,而是:
export HADOOP_CLASSPATH=`hadoop classpath`
Flink组件启动时会使用该环境变量的
特殊情况:如果在Flink application中需要用到Hadoop的input-/output format,只需引入Hadoop兼容包即可(Hadoop compatibility wrappers)
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-hadoop-compatibility_2.11</artifactId>
<version>1.6.2</version>
</dependency>
1.5 Flink项目打包
Flink 可以使用maven-shade-plugin对Flink maven项目进行打包,具体打包命令为mvn clean package。
2. 自己编译Flink
2.1安装maven
1.下载
到maven官网下载安装包,这里我们可以选择使用apache-maven-3.3.9-bin.tar.gz。
2.解压
将apache-maven-3.3.9-bin.tar.gz安装包上传至主节点的,然后使用tar命令进行解压
tar -zxvf apache-maven-3.3.9-bin.tar.gz
3.创建软连接
ln -s apache-maven-3.3.9 maven
4.配置环境变量
vi ~/.bashrc
export MAVEN_HOME=/home/hadoop/app/maven
export PATH=$MAVEN_HOME/bin:$PATH
5.生效环境变量
source ~/.bashrc
6.查看maven版本
mvn –version
7. settings.xml配置阿里镜像
添加阿里镜像
<mirror>
<id>nexus-osc</id>
<mirrorOf>*</mirrorOf>
<name>Nexus osc</name>
<url>http://maven.aliyun.com/nexus/content/repositories/central</url>
</mirror>
2.2安装jdk
编译flink要求jdk8或者以上版本,这里已经提前安装好jdk1.8,具体安装配置不再赘叙,查看版本如下:
[hadoop@cdh01 conf]$ java -version
java version "1.8.0_51"
Java(TM) SE Runtime Environment (build 1.8.0_51-b16)
Java HotSpot(TM) 64-Bit Server VM (build 25.51-b03, mixed mode)
2.3下载源码
登录github:https://github.com/apache/flink,获取flink下载地址:https://github.com/apache/flink.git
打开Flink主节点终端,进入/home/hadoop/opensource目录,通过git clone下载flink源码:
git clone https://github.com/apache/flink.git
错误1:如果Linux没有安装git,会报如下错误:
bash: git: command not found
解决:git安装步骤如下所示:
1.安装编译git时需要的包(注意需要在root用户下安装)
yum install curl-devel expat-devel gettext-devel openssl-devel zlib-devel
yum install gcc perl-ExtUtils-MakeMaker
2.删除已有的git
yum remove git
3.下载git源码
先安装wget
yum -y install wget
使用wget下载git源码
wget https://www.kernel.org/pub/software/scm/git/git-2.0.5.tar.gz
解压git
tar xzf git-2.0.5.tar.gz
编译安装git
cd git-2.0.5
make prefix=/usr/local/git all
sudo make prefix=/usr/local/git install
echo "export PATH=$PATH:/usr/local/git/bin" >> ~/.bashrc
source ~/.bashrc
查看git版本
git –version
错误2:git clone https://github.com/apache/flink.git
Cloning into 'flink'...
fatal: unable to access 'https://github.com/apache/flink.git/': SSL connect error
解决:
升级 nss 版本:yum update nss
2.4切换对应flink版本
使用如下命令查看flink版本分支
git tag
切换到flink对应版本(这里我们使用flink1.6.2)
git checkout release-1.6.2
2.5编译flink
进入flink 源码根目录:/home/hadoop/opensource/flink,通过maven编译flink
mvn clean install -DskipTests -Dhadoop.version=2.6.0
报错:
[INFO] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 06:58 min
[INFO] Finished at: 2019-01-18T22:11:54-05:00
[INFO] Final Memory: 106M/454M
[INFO] ------------------------------------------------------------------------
[ERROR] Failed to execute goal on project flink-mapr-fs: Could not resolve dependencies for project org.apache.flink:flink-mapr-fs:jar:1.6.2: Could not find artifact com.mapr.hadoop:maprfs:jar:5.2.1-mapr in nexus-osc (http://maven.aliyun.com/nexus/content/repositories/central) -> [Help 1]
[ERROR]
[ERROR] To see the full stack trace of the errors, re-run Maven with the -e switch.
[ERROR] Re-run Maven using the -X switch to enable full debug logging.
[ERROR]
[ERROR] For more information about the errors and possible solutions, please read the following articles:
[ERROR] [Help 1] http://cwiki.apache.org/confluence/display/MAVEN/DependencyResolutionException
[ERROR]
[ERROR] After correcting the problems, you can resume the build with the command
[ERROR] mvn <goals> -rf :flink-mapr-fs
报错缺失flink-mapr-fs,需要手动下载安装。
解决:
1.下载maprfs jar包
通过手动下载maprfs-5.2.1-mapr.jar包,下载地址地址:https://repository.mapr.com/nexus/content/groups/mapr-public/com/mapr/hadoop/maprfs/5.2.1-mapr/
2.上传至主节点
将下载的maprfs-5.2.1-mapr.jar包上传至主节点的/home/hadoop/downloads目录下。
3.手动安装
手动安装缺少的包到本地仓库
mvn install:install-file -DgroupId=com.mapr.hadoop -DartifactId=maprfs -Dversion=5.2.1-mapr -Dpackaging=jar -Dfile=/home/hadoop/downloads/maprfs-5.2.1-mapr.jar
4.继续编译
使用maven继续编译flink(可以排除刚刚已经安装的包)
mvn clean install -Dmaven.test.skip=true -Dhadoop.version=2.7.3 -rf :flink-mapr-fs
报错:
[INFO] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 05:51 min
[INFO] Finished at: 2019-01-18T22:39:20-05:00
[INFO] Final Memory: 108M/480M
[INFO] ------------------------------------------------------------------------
[ERROR] Failed to execute goal org.apache.maven.plugins:maven-compiler-plugin:3.1:compile (default-compile) on project flink-mapr-fs: Compilation failure: Compilation failure:
[ERROR] /home/hadoop/opensource/flink/flink-filesystems/flink-mapr-fs/src/main/java/org/apache/flink/runtime/fs/maprfs/MapRFileSystem.java:[70,44] package org.apache.hadoop.fs does not exist
[ERROR] /home/hadoop/opensource/flink/flink-filesystems/flink-mapr-fs/src/main/java/org/apache/flink/runtime/fs/maprfs/MapRFileSystem.java:[73,45] cannot find symbol
[ERROR] symbol: class Configuration
[ERROR] location: package org.apache.hadoop.conf
[ERROR] /home/hadoop/opensource/flink/flink-filesystems/flink-mapr-fs/src/main/java/org/apache/flink/runtime/fs/maprfs/MapRFileSystem.java:[73,93] cannot find symbol
[ERROR] symbol: class Configuration
缺失org.apache.hadoop.fs包,报错找不到。
解决:
flink-mapr-fs模块的pom文件中添加如下依赖:
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-common</artifactId>
<version>${hadoop.version}</version>
</dependency>
继续往后编译:
mvn clean install -Dmaven.test.skip=true -Dhadoop.version=2.7.3 -rf :flink-mapr-fs
又报错:
[ERROR] Failed to execute goal on project flink-avro-confluent-registry: Could not resolve dependencies for project org.apache.flink:flink-avro-confluent-registry:jar:1.6.2: Could not find artifact io.confluent:kafka-schema-registry-client:jar:3.3.1 in nexus-osc (http://maven.aliyun.com/nexus/content/repositories/central) -> [Help 1]
[ERROR]
报错缺少kafka-schema-registry-client-3.3.1.jar 包
解决:
手动下载kafka-schema-registry-client-3.3.1.jar包,下载地址如下:
将下载的kafka-schema-registry-client-3.3.1.jar上传至主节点的目录下/home/hadoop/downloads
手动安装缺少的kafka-schema-registry-client-3.3.1.jar包
mvn install:install-file -DgroupId=io.confluent -DartifactId=kafka-schema-registry-client -Dversion=3.3.1 -Dpackaging=jar -Dfile=/home/hadoop/downloads/kafka-schema-registry-client-3.3.1.jar
继续往后编译
mvn clean install -Dmaven.test.skip=true -Dhadoop.version=2.7.3 -rf :flink-mapr-fs