Apache Atlas源码编译部署和配置运行

点击关注上方“知了小巷”，

设为“置顶或星标”，第一时间送达干货。

Apache Atlas源码编译部署和配置运行

Apache Atlas源码地址：
https://github.com/apache/atlas.git
官方文档：
https://atlas.apache.org/#/
接上篇【Apache Atlas元数据管理入门】

内容提要：
1.大数据组件准备
2.Apache Atlas源码编译和打包
3.Apache Atlas集成外部组件

1. 大数据组件准备

Hadoop集群（本地单机）

$ hadoop version
Hadoop 3.1.1
$ start-dfs.sh
$ start-yarn.sh
$ jps -l
62000 org.apache.hadoop.hdfs.server.namenode.NameNode
62305 org.apache.hadoop.yarn.server.resourcemanager.ResourceManager
62194 org.apache.hadoop.hdfs.server.namenode.SecondaryNameNode
62389 org.apache.hadoop.yarn.server.nodemanager.NodeManager
27495 org.apache.hadoop.util.RunJar
62092 org.apache.hadoop.hdfs.server.datanode.DataNode

Zookeeper集群（本地单机）

$ zkServer.sh start
$ jps -l|grep zookeeper
62593 org.apache.zookeeper.server.quorum.QuorumPeerMain

Kafka集群（本地单机）

$ kafka-server-start.sh -daemon kafka_2.11-0.11.0.2/config/server.properties
$ jps -l|grep kafka
63383 kafka.Kafka

HBase集群（本地单机）

$ start-hbase.sh
$ jps -l|grep hbase
63608 org.apache.hadoop.hbase.regionserver.HRegionServer
63512 org.apache.hadoop.hbase.master.HMaster

Solr（本地单机）
这里是Atlas2.1源码pom.xml中的solr版本7.5.0，其他组件版本没有全部保持一致：

<solr.version>7.5.0</solr.version>

solr安装包地址：
http://archive.apache.org/dist/lucene/solr/
直接下载编译好的包：

$ wget http://archive.apache.org/dist/lucene/solr/7.5.0/solr-7.5.0.zip
$ unzip solr-7.5.0.zip
$ cd solr-7.5.0
$ ./bin/solr start -p 8983
$ lsof -i:8983
COMMAND   PID        USER   FD   TYPE             DEVICE SIZE/OFF NODE NAME
java    80795 shaozhipeng  129u  IPv6 0x7ecef7d3412b64e9      0t0  TCP *:smc-https (LISTEN)

页面访问：
http://127.0.0.1:8983

Hive
基于MySQL的元数据启动起来

$ hive --version
Hive 3.1.1
$ hive --service metastore &
$ lsof -i:9083
COMMAND   PID        USER   FD   TYPE             DEVICE SIZE/OFF NODE NAME
java    27495 shaozhipeng  491u  IPv6 0x7ecef7d340adef29      0t0  TCP *:9083 (LISTEN)
# 27495 org.apache.hadoop.util.RunJar

2. Apache Atlas源码编译和打包

有关apache-atlas的热点问题列表
https://stackoverflow.com/questions/tagged/apache-atlas?sort=votes

下载源码包：
https://mirrors.tuna.tsinghua.edu.cn/apache/atlas/

$ wget https://mirrors.tuna.tsinghua.edu.cn/apache/atlas/2.1.0/apache-atlas-2.1.0-sources.tar.gz
$ tar zxvf apache-atlas-2.1.0-sources.tar.gz
$ ls
apache-atlas-2.1.0-sources.tar.gz apache-atlas-sources-2.1.0
$ cd apache-atlas-sources-2.1.0/
$ mvn --version
Apache Maven 3.6.2 (40f52333136460af0dc0d7232c0dc0bcf0d9e117; 2019-08-27T23:06:16+08:00)
Maven home: /Users/shaozhipeng/Development/apache-maven-3.6.2
Java version: 1.8.0_181, vendor: Oracle Corporation, runtime: /Library/Java/JavaVirtualMachines/jdk1.8.0_181.jdk/Contents/Home/jre
Default locale: zh_CN, platform encoding: UTF-8
OS name: "mac os x", version: "10.13.6", arch: "x86_64", family: "mac"

maven conf/settings.xml
添加阿里云的仓库镜像

<mirror>
  <id>aliyunmaven</id>
  <mirrorOf>*</mirrorOf>
  <name>ali</name>
  <url>https://maven.aliyun.com/repository/public</url>
</mirror>

可能会由于Caused by: java.lang.ClassNotFoundException: org.apache.commons.collections.ExtendedProperties
报错导致编译失败，解决方案：
https://stackoverflow.com/questions/51957221/process-resource-bundles-maven-error-calssnotfoundexception-org-apache-commons-c

类似报错org.apache.maven.plugins:maven-site-plugin:3.7:site: org/apache/commons/lang/StringUtils
：

[ERROR] Failed to execute goal org.apache.maven.plugins:maven-site-plugin:3.7:site (default) on project hdfs-model: Execution default of goal org.apache.maven.plugins:maven-site-plugin:3.7:site failed: A required class was missing while executing org.apache.maven.plugins:maven-site-plugin:3.7:site: org/apache/commons/lang/StringUtils

阿里云仓库找不到sqoop的jar包Could not find artifact org.apache.sqoop:sqoop:jar:1.4.6.2.3.99.0-195 in aliyunmaven
：

[ERROR] Failed to execute goal on project sqoop-bridge-shim: Could not resolve dependencies for project org.apache.atlas:sqoop-bridge-shim:jar:2.1.0: Could not find artifact org.apache.sqoop:sqoop:jar:1.4.6.2.3.99.0-195 in aliyunmaven (https://maven.aliyun.com/repository/public) -> [Help 1]

修改pom.xml文件，分别找到maven-remote-resources-plugin
和maven-site-plugin
两个插件的位置：

<plugin>
    <groupId>org.apache.maven.plugins</groupId>
    <artifactId>maven-remote-resources-plugin</artifactId>
    <version>1.6.0</version>
    <dependencies>
        <dependency>
            <groupId>commons-collections</groupId>
            <artifactId>commons-collections</artifactId>
            <version>3.2.2</version>
        </dependency>
        <!-- https://mvnrepository.com/artifact/commons-lang/commons-lang -->
        <dependency>
            <groupId>commons-lang</groupId>
            <artifactId>commons-lang</artifactId>
            <version>2.6</version>
        </dependency>
    </dependencies>
    <executions>
        <execution>
          <goals>
            <goal>bundle</goal>
          </goals>
        </execution>
    </executions>
    <configuration>
        <excludeGroupIds>org.restlet.jee</excludeGroupIds>
    </configuration>
</plugin>

<plugin>
    <groupId>org.apache.maven.plugins</groupId>
    <artifactId>maven-site-plugin</artifactId>
    <version>${maven-site-plugin.version}</version>
    <dependencies>
        <!-- https://mvnrepository.com/artifact/commons-lang/commons-lang -->
        <dependency>
            <groupId>commons-lang</groupId>
            <artifactId>commons-lang</artifactId>
            <version>2.6</version>
        </dependency>
    </dependencies>
</plugin>

sqoop依赖jar包下载失败的问题，下载地址，百度网盘：
链接: https://pan.baidu.com/s/1FWHK7AGPQQI0OdSFbTE7tg 密码: gjfn

<sqoop.version>1.4.6.2.3.99.0-195</sqoop.version>

安装到本地maven仓库

$ mvn install:install-file -Dfile=sqoop-1.4.6.2.3.99.0-195.jar -DgroupId=org.apache.sqoop -DartifactId=sqoop -Dversion=1.4.6.2.3.99.0-195 -Dpackaging=jar
...
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time:  0.644 s
...

源码编译

$ mvn clean -DskipTests install

打包(可以直接打包)

$ mvn clean -DskipTests package -Pdist

不到8分钟就可以打包完成，查看打包后的文件

$ cd distro/
$ ls
pom.xml src target
$ ls target/
META-INF          apache-atlas-2.1.0-hive-hook.tar.gz   apache-atlas-2.1.0-storm-hook
apache-atlas-2.1.0-bin            apache-atlas-2.1.0-impala-hook      apache-atlas-2.1.0-storm-hook.tar.gz
apache-atlas-2.1.0-bin.tar.gz     apache-atlas-2.1.0-impala-hook.tar.gz   archive-tmp
apache-atlas-2.1.0-classification-updater apache-atlas-2.1.0-kafka-hook     atlas-distro-2.1.0.jar
apache-atlas-2.1.0-classification-updater.zip apache-atlas-2.1.0-kafka-hook.tar.gz    bin
apache-atlas-2.1.0-falcon-hook          apache-atlas-2.1.0-server     conf
apache-atlas-2.1.0-falcon-hook.tar.gz   apache-atlas-2.1.0-server.tar.gz    maven-archiver
apache-atlas-2.1.0-hbase-hook           apache-atlas-2.1.0-sources.tar.gz   maven-shared-archive-resources
apache-atlas-2.1.0-hbase-hook.tar.gz    apache-atlas-2.1.0-sqoop-hook     rat.txt
apache-atlas-2.1.0-hive-hook            apache-atlas-2.1.0-sqoop-hook.tar.gz    test-classes

3. Apache Atlas集成外部组件

Apache Atlas依赖Solr（或ES）
、HBase（或Cassandra）
和Kafka
。

将压缩包apache-atlas-2.1.0-server.tar.gz
拷贝出来并解压：

$ ls
apache-atlas-2.1.0
$ ls apache-atlas-2.1.0
DISCLAIMER.txt  LICENSE   NOTICE    bin   conf    models    server    tools
$ cd apache-atlas-2.1.0/conf/
$ ls
atlas-application.properties  atlas-log4j.xml     cassandra.yml.template    solr        zookeeper
atlas-env.sh      atlas-simple-authz-policy.json  hbase       users-credentials.properties

3.1 Apache Atlas集成HBase

1. 修改atlas存储数据所在的机器主机atlas.graph.storage.hostname
   ，是ZK地址

$ vi atlas-application.properties
atlas.graph.storage.backend=hbase2
atlas.graph.storage.hbase.table=apache_atlas_janus

#Hbase
#For standalone mode , specify localhost
#for distributed mode, specify zookeeper quorum here
atlas.graph.storage.hostname=localhost:2181
atlas.graph.storage.hbase.regions-per-server=1
atlas.graph.storage.lock.wait-time=10000

1. 拷贝hbase的配置文件hbase-site.xml
   到atlas的conf/hbase
   下面：

$ cp ~/Development/hbase-2.1.0/conf/hbase-site.xml hbase/
$ ls hbase/
hbase-site.xml    hbase-site.xml.template

1. 修改atlas的环境变量配置信息，添加hbase的配置文件目录

$ vi atlas-env.sh
export HBASE_CONF_DIR=/Users/shaozhipeng/Development/project/java/atlas-download/apache-atlas-2.1.0/conf/hbase

3.2 Apache Atlas集成Solr

1. 默认是配置solr cloud，由于实验是单机环境部署，改为http模式：

# Graph Search Index
atlas.graph.index.search.backend=solr

#Solr 这里的注释掉
#Solr cloud mode properties
#atlas.graph.index.search.solr.mode=cloud
#atlas.graph.index.search.solr.zookeeper-url=
#atlas.graph.index.search.solr.zookeeper-connect-timeout=60000
#atlas.graph.index.search.solr.zookeeper-session-timeout=60000
#atlas.graph.index.search.solr.wait-searcher=true

#Solr http mode properties
atlas.graph.index.search.solr.mode=http
atlas.graph.index.search.solr.http-urls=http://localhost:8983/solr

1. 将Apache Atlas自带的Solr文件夹拷贝到外部Solr相应目录下，并重命名为atlas_conf
   

$ ls solr/
currency.xml lang  protwords.txt schema.xml solrconfig.xml stopwords.txt synonyms.txt
$ cp -rf solr ~/Development/solr-7.5.0/atlas_conf
$ ls ~/Development/solr-7.5.0/atlas_conf/
currency.xml lang  protwords.txt schema.xml solrconfig.xml stopwords.txt synonyms.txt

1. 创建collection

$ ~/Development/solr-7.5.0/bin/solr create -c vertex_index -d ~/Development/solr-7.5.0/atlas_conf 
INFO  - 2020-12-16 23:36:52.930; org.apache.solr.util.configuration.SSLCredentialProviderFactory; Processing SSL Credential Provider chain: env;sysprop

Created new core 'vertex_index' 

1. Solr页面验证
   
   

   

3.3 Apache Atlas集成Kafka

1. 修改Notification Configs
   相关配置：

$ vi atlas-application.properties
#########  Notification Configs  #########
atlas.notification.embedded=false
atlas.kafka.data=/Users/shaozhipeng/Development/pseudo/kafka/kafka-logs
atlas.kafka.zookeeper.connect=localhost:2181
atlas.kafka.bootstrap.servers=localhost:9092
atlas.kafka.zookeeper.session.timeout.ms=4000

1. 创建Kafka Topic

$ kafka-topics.sh --zookeeper localhost:2181 --create --replication-factor 1 --partitions 3 --topic _HOATLASOK
Created topic "_HOATLASOK".
$ kafka-topics.sh --zookeeper localhost:2181 --create --replication-factor 1 --partitions 3 --topic ATLAS_ENTITIES
Created topic "ATLAS_ENTITIES".
$ kafka-topics.sh --zookeeper localhost:2181 --list
ATLAS_ENTITIES
_HOATLASOK
__consumer_offsets

3.4 Apache Atlas Server其它配置

修改Server Properties

$ vi atlas-application.properties
#########  Server Properties  #########
atlas.rest.address=http://localhost:21000
# If enabled and set to true, this will run setup steps when the server starts
atlas.server.run.setup.on.start=false

#########  Entity Audit Configs  #########
atlas.audit.hbase.tablename=apache_atlas_entity_audit
atlas.audit.zookeeper.session.timeout.ms=1000
atlas.audit.hbase.zookeeper.quorum=localhost:2181

修改atlas-log4j.xml
日志输出，找到org.apache.atlas.perf
，去掉注释

$ vi atlas-log4j.xml 
    <appender name="perf_appender" class="org.apache.log4j.DailyRollingFileAppender">
        <param name="file" value="${atlas.log.dir}/atlas_perf.log" />
        <param name="datePattern" value="'.'yyyy-MM-dd" />
        <param name="append" value="true" />
        <layout class="org.apache.log4j.PatternLayout">
            <param name="ConversionPattern" value="%d|%t|%m%n" />
        </layout>
    </appender>

    <logger name="org.apache.atlas.perf" additivity="false">
        <level value="debug" />
        <appender-ref ref="perf_appender" />
    </logger>

3.5 启动Atlas服务

atlas_start.py
是Python2的语法，Mac环境直接运行即可。

$ ./bin/atlas_start.py 
starting atlas on host localhost
starting atlas on port 21000
..............................................................................................................................................................................................................................................................................................................
Apache Atlas Server started!!!
$ jps -l |grep atlas
27180 org.apache.atlas.Atlas

页面访问：
http://localhost:21000
用户名密码：admin/admin

点一下，代码无 Bug