一、环境说明
1.1 基于CDH5.11.1的配置安装
hadoop 2.6.0
spark 2.2.0
hbase 1.2.0
zookeeper 3.4.5
janusgraph 0.3.1
Scala 2.11.8
Java 1.8.0
1.2 服务分布
| zookeeper主机 | Hbase主机 | janusgraph |
|---|---|---|
| Hbasetest1.com | Hbasetest1.com | Hbasetest3.com |
| Hbasetest2.com | Hbasetest2.com | |
| Hbasetest3.com | Hbasetest3.com |
二、系统配置
# 配置环境变量：
export JAVA_HOME=/usr/java/jdk1.8/
export SPARK_HOME=/opt/cloudera/parcels/SPARK2/lib/spark2/
export JANUSGRAPH_HOME=/home/we-op/janusgraph
export JANUSGRAPH_CONF_DIR=$JANUSGRAPH_HOME/conf/
export HBASE_CONF_DIR=/etc/hbase/conf/
export SPARK_CONF_DIR=$SPARK_HOME/conf/
export HADOOP_CONF_DIR=/etc/hadoop/conf/
export HADOOP_HOME=/opt/cloudera/parcels/CDH/lib/hadoop/
export HADOOP_COMMON_LIB_NATIVE_DIR=$HADOOP_HOME/lib/native
export HADOOP_OPTS="-Djava.library.path=${HADOOP_HOME}/lib/native"
export HADOOP_OPTS="-Djava.library.path=$HADOOP_HOME/lib:$HADOOP_COMMON_LIB_NATIVE_DIR"
export PATH=$PATH:$JAVA_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$JANUSGRAPH_HOME/bin:$SPARK_HOME/bin
export CLASSPATH=$HADOOP_CONF_DIR:$SPARK_CONF_DIR:$HBASE_CONF_DIR:$JANUSGRAPH_CONF_DIR
三、源码编译Janusgraph
GitHub地址：https://github.com/JanusGraph/janusgraph/tree/v0.3.1
2.1 修改janusgraph-core源码
打开并修改janusgraph-core/src/main/java/org/janusgraph/graphdb/database/idassigner/StandardIDPool.java
import com.google.common.base.inter.Stopwatch;
2.2 拷贝目录
cp -R janusgraph-hbase-parent/janusgraph-hbase-core/src/main/java/com/ janusgraph-core/src/main/java/
cd janusgraph-core/src/main/java/com/google/common/base
mkdir inter
mv Stopwatch.java inter/
# 打开并修改Stopwath.java
package com.google.common.base.inter;
import com.google.common.base.Ticker;
2.3 修改janusgraph-0.3.1/pom.xml文件
# 修改组件依赖 版本为cdh相同版本
<hadoop2.version>2.6.0-cdh5.11.1</hadoop2.version>
<hbase1.version>1.2.0-cdh5.11.1</hbase1.version>
<hbase.server.version>1.2.0-cdh5.11.1</hbase.server.version>
<zookeeper.version>3.4.5-cdh5.11.1</zookeeper.version>
#增加数据源：
<repository>
<id>cloudera</id>
<url>https://repository.cloudera.com/artifactory/cloudera-repos/</url>
</repository>
# 增加maven和java版本检查, 如果编译时使用了-Denforcer.skip=true 则不用添加以下内容
<execution>
<id>enforce-dependency-convergence</id>
<goals>
<goal>enforce</goal>
</goals>
<configuration>
<rules>
<requireMavenVersion>
<version>3.6.2</version>
</requireMavenVersion>
<requireJavaVersion>
<version>1.8.0</version>
</requireJavaVersion>
</rules>
</configuration>
</execution>
#修改janusgraph-cql/pom.xml
<directory>${basedir}/src/test/resources</directory>
2.3 编译命令
mvn clean install -Pjanusgraph-release -Dgpg.skip=true -DskipTests=true -Denforcer.skip=true --fail-at-end
编译成功后：janusgraph-dist/janusgraph-dist-hadoop-2/target/
可以看到对应的zip包： janusgraph-0.3.1-hadoop2.zip
四、替换依赖包
使用hdfs用户；解压janusgraph-0.3.1-hadoop2.zip到指定目录。
4.1 加入新依赖包
$CDH_HOME/jars/hadoop-yarn-server-web-proxy-2.6.0-cdh5.11.1.jar
$CDH_HOME/jars/zkclient-0.10.jar
$SPARK_HOME/jars/spark-core_2.11-2.2.0.cloudera2.jar
$SPARK_HOME/jars/spark-network-common_2.11-2.2.0.cloudera2.jar
$SPARK_HOME/jars/spark-yarn_2.11-2.2.0.cloudera2.jar
拷贝以上jar包到 $JANUSGRAPH_HOME/lib/目录下。
4.2 删除旧依赖包
$JANUSGRAPH_HOME/lib/spark-network-common_2.11-2.2.0.jar
$JANUSGRAPH_HOME/lib/spark-core_2.11-2.2.0.jar
五、修改属性文件
# 修改配置文件conf/janusgraph-hbase.properties
storage.backend=hbase
storage.hostname=Hbasetest1.com,Hbasetest2.com,Hbasetest3.com
#修改：conf/hadoop-graph/hadoop-load.properties
janusgraphmr.ioformat.conf.storage.backend=hbase
janusgraphmr.ioformat.conf.storage.hostname=Hbasetest1.com,Hbasetest2.com,Hbasetest3.com
spark.master=yarn
spark.executor.memory=1g
spark.executor.instances=2
spark.executor.cores=4
spark.serializer=org.apache.spark.serializer.KryoSerializer
spark.yarn.jars=/opt/cloudera/parcels/SPARK2/lib/spark2/jars/*
spark.yarn.am.extraJavaOptions=-Djava.library.path=/opt/cloudera/parcels/CDH/lib/hadoop/lib/native
spark.executor.extraJavaOptions=-Djava.library.path=/opt/cloudera/parcels/CDH/lib/hadoop/lib/native
spark.executor.extraClassPath=/home/we-op/janusgraph/lib/*:/etc/hadoop/conf:/etc/hbase/conf:/etc/spark2/conf
#修改：conf/hadoop-graph/read-hbase.properties
janusgraphmr.ioformat.conf.storage.backend=hbase
janusgraphmr.ioformat.conf.storage.hostname=Hbasetest1.com,Hbasetest2.com,Hbasetest3.com
janusgraphmr.ioformat.conf.storage.hbase.table=janusgraph
spark.master=yarn
spark.serializer=org.apache.spark.serializer.KryoSerializer
spark.yarn.am.extraJavaOptions=-Djava.library.path=/opt/cloudera/parcels/CDH/lib/hadoop/lib/native
spark.executor.extraJavaOptions=-Djava.library.path=/opt/cloudera/parcels/CDH/lib/hadoop/lib/native
spark.executor.extraClassPath=/home/we-op/janusgraph/lib/*:/etc/hadoop:/etc/hbase/conf:/etc/spark2/conf
六、测试
6.1 导入测试数据：
gremlin> :load data/grateful-dead-janusgraph-schema.groovy --执行一次
==>true
==>true
gremlin> graph=GraphFactory.open("conf/gremlin-server/offline_hbase_all_in_one_20190614.properties")
==>standardjanusgraph[hbase:[Hbasetest1.com, Hbasetest2.com, Hbasetest3.com]]
gremlin> g=graph.traversal()
==>graphtraversalsource[standardjanusgraph[hbase:[Hbasetest1.com, Hbasetest2.com, Hbasetest3.com]], standard]
gremlin> defineGratefulDeadSchema(graph)
==>null
gremlin> g.addV("song").property('name','123')
==>v[4272]
gremlin> g.tx().commit()
==>null
gremlin> g.V().count()
16:02:16 WARN org.janusgraph.graphdb.transaction.StandardJanusGraphTx - Query requires iterating over all vertices [()]. For better performance, use indexes
==>1
gremlin> graph.close()
==>null
6.2 执行读写测试：
image.png
