Atlas 是什么?

Atlas是一组可扩展和可扩展的核心基础治理服务,使企业能够有效地满足Hadoop中的合规性要求,并允许与整个企业数据生态系统集成。

Apache Atlas为组织提供了开放的元数据管理和治理功能,以建立其数据资产的目录,对这些资产进行分类和治理,并为数据科学家,分析师和数据治理团队提供围绕这些数据资产的协作功能。

如果没有Atlas

大数据表依赖问题不好解决,元数据管理需要自行开发,如:hive血缘依赖图

对于表依赖问题,没有一个可以查询的工具,不方便错误定位,即业务sql开发

官网:http://atlas.apache.org
表与表之间的血缘依赖
img

字段与字段之间的血缘依赖
img

1 Atlas 架构原理
img

2 Atlas 安装及使用
安装需要组件,HDFS、Yarn、Zookeeper、Kafka、Hbase、Solr、Hive,Python2.7环境

需要Maven3.5.0以上,jdk_151以上,python2.7。

2.1 下载源码包2.0.0,IDEA打开
因与CDH集成,修改pom文件
在repositories标签中增加CDH仓库

cloudera
https://repository.cloudera.com/artifactory/cloudera-repos
true


false


2.2 修改相关版本与CDH版本对应

<lucene-solr.version>7.4.0</lucene-solr.version>
 <hadoop.version>3.0.0-cdh6.3.2</hadoop.version>
 <hbase.version>2.1.0-cdh6.3.2</hbase.version>
 <solr.version>7.4.0-cdh6.3.2</solr.version>
 <hive.version>2.1.1-cdh6.3.2</hive.version>
 <kafka.version>2.2.1-cdh6.3.2</kafka.version>
 <kafka.scala.binary.version>2.11</kafka.scala.binary.version>
 <zookeeper.version>3.4.5-cdh6.3.2</zookeeper.version>
 <sqoop.version>1.4.7-cdh6.3.2</sqoop.version>
 2.3 兼容Hive2.1.1
 所需修改的项目位置:apache-atlas-sources-2.1.0\addons\hive-bridge
 ①.org/apache/atlas/hive/bridge/HiveMetaStoreBridge.java 577行String catalogName = hiveDB.getCatalogName() != null ? hiveDB.getCatalogName().toLowerCase() : null;
 改为:String catalogName = null;
 ②.org/apache/atlas/hive/hook/AtlasHiveHookContext.java 81行this.metastoreHandler = (listenerEvent != null) ? metastoreEvent.getIHMSHandler() : null;
 改为:C:\Users\Heaton\Desktop\apache-atlas-2.1.0-sources\apache-atlas-sources-2.1.0\addonsthis.metastoreHandler = null;


2.3 编译
mvn clean -DskipTests package -Pdist -X -T 8
编译完成的文件在此目录apache-atlas-sources-2.1.0\distro\target

2.5 安装
mkdir /usr/local/src/atlas
cd /usr/local/src/atlas
#复制apache-atlas-2.1.0-bin.tar.gz到安装目录
tar -zxvf apache-atlas-2.1.0-bin.tar.gz
cd apache-atlas-2.1.0/
2.6 修改配置文件
vim conf\atlas-application.properties

#集成修改hbase配置
atlas.graph.storage.hostname=cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181
#集成修改solr配置
atlas.graph.index.search.solr.zookeeper-url=cdh01.cm:2181/solr,cdh02.cm:2181/solr,cdh03.cm:2181/solr
#集成修改kafka配置
atlas.notification.embedded=false #false外置的kafka
atlas.kafka.zookeeper.connect=cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181
atlas.kafka.bootstrap.servers=cdh01.cm:9092,cdh02.cm:9092,cdh03.cm:9092
atlas.kafka.zookeeper.session.timeout.ms=60000
atlas.kafka.zookeeper.connection.timeout.ms=30000
atlas.kafka.enable.auto.commit=true
#集成修改其他配置
atlas.rest.address=http://cdh01.cm:21000 #访问地址端口,此值修改不生效,默认本地21000端口,此端口和impala冲突
atlas.server.run.setup.on.start=false #如果启用并设置为true,则在服务器启动时将运行安装步骤
atlas.audit.hbase.zookeeper.quorum=cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181
#集成添加hive钩子配置(文件最下面即可)
#在hive中做任何操作,都会被钩子所感应到,并生成相应的事件发往atlas所订阅的kafka-topic,再由atlas进行元数据生成和存储管理
######### Hive Hook Configs #######
atlas.hook.hive.synchronous=false
atlas.hook.hive.numRetries=3
atlas.hook.hive.queueSize=10000
atlas.cluster.name=primary

#配置用户名密码(选做)
#开启或关闭三种验证方法
atlas.authentication.method.kerberos=true|false
atlas.authentication.method.ldap=true|false
atlas.authentication.method.file=true
#vim users-credentials.properties(修改该文件)
#>>>源文件
#username=group::sha256-password
admin=ADMIN::8c6976e5b5410415bde908bd4dee15dfb167a9c873fc4bb8a81f6f2ab448a918
rangertagsync=RANGER_TAG_SYNC::e3f67240f5117d1753c940dae9eea772d36ed5fe9bd9c94a300e40413f1afb9d
#<<<
#>>>修改成用户名bigdata123,密码bigdata123
#username=group::sha256-password
bigdata123=ADMIN::aa0336d976ba6db36f33f75a20f68dd9035b1e0e2315c331c95c2dc19b2aac13
rangertagsync=RANGER_TAG_SYNC::e3f67240f5117d1753c940dae9eea772d36ed5fe9bd9c94a300e40413f1afb9d
#<<<
#计算sha256:echo -n “bigdata123”|sha256sum
vim conf/atlas-env.sh

#集成添加hbase配置->下面的目录为atlas下的hbase配置目录,需要后面加入集群hbase配置
export HBASE_CONF_DIR=/usr/local/src/atlas/apache-atlas-2.1.0/conf/hbase/conf
#export HBASE_CONF_DIR=/etc/hbase/conf------------------------------
export MANAGE_LOCAL_HBASE=false (false外置的zk和hbase)
export MANAGE_LOCAL_SOLR=false (false外置的solr)

#修改内存指标(根据线上机器配置)
export ATLAS_SERVER_OPTS="-server -XX:SoftRefLRUPolicyMSPerMB=0
-XX:+CMSClassUnloadingEnabled -XX:+UseConcMarkSweepGC
-XX:+CMSParallelRemarkEnabled -XX:+PrintTenuringDistribution
-XX:+HeapDumpOnOutOfMemoryError
-XX:HeapDumpPath=dumps/atlas_server.hprof
-Xloggc:logs/gc-worker.log -verbose:gc
-XX:+UseGCLogFileRotation -XX:NumberOfGCLogFiles=10
-XX:GCLogFileSize=1m -XX:+PrintGCDetails -XX:+PrintHeapAtGC
-XX:+PrintGCTimeStamps"

#优化 JDK1.8(以下需要16G内存)
export ATLAS_SERVER_HEAP="-Xms15360m -Xmx15360m
-XX:MaxNewSize=5120m -XX:MetaspaceSize=100M
-XX:MaxMetaspaceSize=512m"
vim conf/atlas-log4j.xml

#去掉如下代码的注释(开启如下代码)





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

2.7 集成Hbase
添加hbase集群配置文件到apache-atlas-2.0.0/conf/hbase下(这里连接的路径需要和上面atlas-env.sh配置中一样)
ln -s /etc/hbase/conf/ /usr/local/src/atlas/apache-atlas-2.1.0/conf/hbase/
2.8 集成Solr
将apache-atlas-2.1.0/conf/solr文件拷贝到solr所有节点的安装目录下,更名为atlas-solr
scp -r /usr/local/src/atlas/apache-atlas-2.1.0/conf/solr root@cdh01.cm:/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/
scp -r /usr/local/src/atlas/apache-atlas-2.1.0/conf/solr root@cdh02.cm:/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/
scp -r /usr/local/src/atlas/apache-atlas-2.1.0/conf/solr root@cdh03.cm:/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/

#在solr节点
cd /opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/
mv solr/ atlas-solr
#在任意solr节点修改solr对应的bash
vi /etc/passwd
/sbin/nologin 修改为 /bin/bash
#切换solr用户执行
su solr
/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/bin/solr create -c vertex_index -d /opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/atlas-solr -shards 3 -replicationFactor 2

/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/bin/solr create -c edge_index -d /opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/atlas-solr -shards 3 -replicationFactor 2

/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/bin/solr create -c fulltext_index -d /opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/solr/atlas-solr -shards 3 -replicationFactor 2

#如果创建错误,可使用 /opt/cloudera/parcels/CDH/lib/solr/bin/solr delete -c ${collection_name} 删除
#切换root用户继续配置其他
su root
solr web控制台: http://cdh01.cm:8983 验证是否启动成功
img

2.9 集成kafka
创建kafka-topic
kafka-topics --zookeeper cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181 --create --replication-factor 3 --partitions 3 --topic _HOATLASOK
kafka-topics --zookeeper cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181 --create --replication-factor 3 --partitions 3 --topic ATLAS_ENTITIES
kafka-topics --zookeeper cdh01.cm:2181,cdh02.cm:2181,cdh03.cm:2181 --create --replication-factor 3 --partitions 3 --topic ATLAS_HOOK
2.10 启动测试
cd /usr/local/src/atlas/apache-atlas-2.1.0/

./bin/atlas_start.py

#停止:./bin/atlas_stop.py
http://cdh01.cm:21000

默认用户名和密码为:admin

2.11 集成Hive
将 atlas-application.properties 配置文件,压缩加入到 atlas-plugin-classloader-2.0.0.jar 中
#必须在此路径打包,才能打到第一级目录下
cd /usr/local/src/atlas/apache-atlas-2.1.0/conf

zip -u /usr/local/src/atlas/apache-atlas-2.1.0/hook/hive/atlas-plugin-classloader-2.1.0.jar atlas-application.properties
修改 hive-site.xml
img

hive.exec.post.hooks org.apache.atlas.hive.hook.HiveHook 修改 hive-env.sh 的 Gateway 客户端环境高级配置代码段(安全阀) img

HIVE_AUX_JARS_PATH=/usr/local/src/atlas/apache-atlas-2.1.0/hook/hive
修改 HIVE_AUX_JARS_PATH
img

修改 hive-site.xml 的 HiveServer2 高级配置代码段(安全阀)
img

hive.exec.post.hooks org.apache.atlas.hive.hook.HiveHook hive.reloadable.aux.jars.path /usr/local/src/atlas/apache-atlas-2.1.0/hook/hive 修改 HiveServer2 环境高级配置代码段 img

HIVE_AUX_JARS_PATH=/usr/local/src/atlas/apache-atlas-2.1.0/hook/hive
将配置好的Atlas包发往各个hive节点后重启集群
scp -r /usr/local/src/atlas/apache-atlas-2.1.0 root@cdh02.cm:/usr/local/src/atlas/

scp -r /usr/local/src/atlas/apache-atlas-2.1.0 root@cdh03.cm:/usr/local/src/atlas/
更新配置重启集群

将atlas配置文件copy到/etc/hive/conf下(集群各个节点)
scp /usr/local/src/atlas/apache-atlas-2.1.0/conf/atlas-application.properties root@cdh01.cm:/etc/hive/conf

scp /usr/local/src/atlas/apache-atlas-2.1.0/conf/atlas-application.properties root@cdh02.cm:/etc/hive/conf

scp /usr/local/src/atlas/apache-atlas-2.1.0/conf/atlas-application.properties root@cdh03.cm:/etc/hive/conf
2.12 再次启动 Atlas
#启动
./bin/atlas_start.py
#停止:./bin/atlas_stop.py
img

img

注意监控日志,看是否报错。主要日志application.log

img

2.13 将 Hive 元数据导入 Atlas
atlas节点添加hive环境变量
vim /etc/profile
#>>>
#hive
export HIVE_HOME=/opt/cloudera/parcels/CDH-6.3.2-1.cdh6.3.2.p0.1605554/lib/hive
export HIVE_CONF_DIR=/etc/hive/conf
export PATH=数据血缘 java工具 数据血缘 atlas_数据血缘 java工具PATH
#<<<

source /etc/profile
执行atlas脚本
./bin/import-hive.sh
#输入用户名:admin;输入密码:admin(如修改请使用修改的)
img

img