Hudi 源码编译安装

1.安装前准备

1.1 环境准备

hadoop-3.3.2
hive-3.1.3
spark-3.3.2
hudi-0.12.3

1.2 安装maven

apache-maven-3.8.6-bin.tar.gz

配置环境变量

# maven
export MAVEN_HOME=/data/base/maven-3.8.6
export PATH=$PATH:$MAVEN_HOME/bin

修改maven本地仓库配置和阿里云maven远程仓库

  <localRepository>/m2/repository</localRepository>
  
      <mirror>
        <id>nexus-aliyun</id>
        <mirrorOf>central</mirrorOf>
        <name>Nexus aliyun</name>
        <url>http://maven.aliyun.com/nexus/content/groups/public</url>
    </mirror>

1.3 下载源码包

wget https://dlcdn.apache.org/hudi/0.12.3/hudi-0.12.3.src.tgz --no-check-certificate

2. 编译

2.1 修改hudi pom.xml 

vim /data/lake/hudi-0.12.3/pom.xml

    <hadoop.version>3.3.2</hadoop.version>
    <hive.groupid>org.apache.hive</hive.groupid>
    <hive.version>3.1.3</hive.version>

修改 hudi pom.xml 新增阿里云maven仓库源

   <repository>
        <id>nexus-aliyun</id>
        <name>nexus-aliyun</name>
        <url>http://maven.aliyun.com/nexus/content/groups/public/</url>
        <releases>
            <enabled>true</enabled>
        </releases>
        <snapshots>
            <enabled>false</enabled>
        </snapshots>
    </repository>

使用maven编译

mvn clean package -DskipTests -Dspark3.3 -Dflink1.14 -Dscala-2.12 -Dhadoop.version=3.3.2 -Pflink-bundle-shade-hive3

2.2 编译报错解决

HoodieParquetDataBlock.java:[110,44] 对于FSDataOutputStream(java.io.ByteArrayOutputStream), 找不到合适的构造器

解决方法

3. Hudi客户端使用测试

cd /data/lake/hudi-0.12.3/hudi-cli

./hudi-cli.sh