mysql - 使用 Oozie 永远运行的 Sqoop 作业

Question

我们有两个节点的 hadoop yarn 集群，它是 hadoop 2.2，在其上我们使用 oozie 在单个工作流中安排了两个操作，第一个操作是 python map-reduce 流操作，第二个是 sqoop export 作业，它实际上是将 map-reduce 流操作的输出传输到 mysql 数据库。

流式操作成功执行，导致 sqoop 作业启动，并一直运行。

stdout 结果如下。

Sqoop command arguments :
         export
         --connect
         jdbc:mysql://localhost/database
         --username
         root
         --password
         root
         --table
         tableName
         --direct
         --export-dir
        /user/hduser/oozieProject/workflow/output

 =================================================================
Invoking Sqoop command line now >>>

2137 [main] WARN  org.apache.sqoop.tool.SqoopTool  - $SQOOP_CONF_DIR has not been set in the environment. Cannot check for additional configuration.
2158 [main] INFO  org.apache.sqoop.Sqoop  - Running Sqoop version: 1.4.4.2.0.6.1-102
2170 [main] WARN  org.apache.sqoop.tool.BaseSqoopTool  - Setting your password on the command-line is insecure. Consider using -P instead.
2178 [main] WARN  org.apache.sqoop.ConnFactory  - $SQOOP_CONF_DIR has not been set in the environment. Cannot check for additional configuration.
2197 [main] INFO  org.apache.sqoop.manager.MySQLManager  - Preparing to use a MySQL streaming resultset.
2197 [main] INFO  org.apache.sqoop.tool.CodeGenTool  - Beginning code generation
2464 [main] INFO  org.apache.sqoop.manager.SqlManager  - Executing SQL statement:     SELECT t.* FROM `missedCalls` AS t LIMIT 1
2483 [main] INFO  org.apache.sqoop.manager.SqlManager  - Executing SQL statement: SELECT t.* FROM `missedCalls` AS t LIMIT 1
2485 [main] INFO  org.apache.sqoop.orm.CompilationManager  - HADOOP_MAPRED_HOME is /usr/local/hadoop
3838 [main] INFO  org.apache.sqoop.orm.CompilationManager  - Writing jar file: /tmp/sqoop-hduser/compile/21bd1d5fe13adeed4f46a09f8b3d38fe/missedCalls.jar
3847 [main] INFO  org.apache.sqoop.mapreduce.ExportJobBase  - Beginning export of missedCalls
Heart beat
Heart beat
Heart beat
Heart beat
Heart beat
Heart beat
Heart beat
Heart beat

作业属性如下

nameNode=hdfs://master:54310
jobTracker=master:8035
queueName=default

oozie.libpath=${nameNode}/user/hduser/share/lib
oozie.use.system.libpath=true
oozie.wf.rerun.failnodes=true

oozieProjectRoot=${nameNode}/user/hduser/oozieProject
appPath=${oozieProjectRoot}/workflow
oozie.wf.application.path=${appPath}
oozieLibPath=${oozie.libpath}

mapred.tasktracker.map.tasks.maximum=4
mapred.tasktracker.reduce.tasks.maximum=4

inputDir=${oozieProjectRoot}/data/*
outputDir=${appPath}/output

工作流xml如下

<!--Oozie workflow file: workflow.xml --> 
<workflow-app name="WorkflowStreamingMRAction-Python" xmlns="uri:oozie:workflow:0.1">
<start to="streamingaAction"/>
<action name="streamingaAction">
    <map-reduce>
        <job-tracker>${jobTracker}</job-tracker>
        <name-node>${nameNode}</name-node>
        <prepare>
            <delete path="${outputDir}"/>
        </prepare>
        <streaming>
            <mapper>python mapper.py</mapper>
            <reducer>python reducer.py</reducer>
        </streaming>
        <configuration>
            <property>
                <name>oozie.libpath</name>
                <value>${oozieLibPath}/mapreduce-streaming</value>
            </property>
            <property>
                <name>mapred.input.dir</name>
                <value>${inputDir}</value>
            </property>
            <property>
                <name>mapred.output.dir</name>
                <value>${outputDir}</value>
            </property>
            <property>
                <name>mapred.reduce.tasks</name>
                <value>4</value>
            </property>
        </configuration>
        <file>${appPath}/mapper.py#mapper.py</file>
        <file>${appPath}/reducer.py#reducer.py</file>
    </map-reduce>
    <ok to="sqoopAction"/>
    <error to="killJobAction"/>
    </action>

    <action name="sqoopAction">
        <sqoop xmlns="uri:oozie:sqoop-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <command>export --connect jdbc:mysql://localhost/database --username root --password myPwd --table tableName --direct --export-dir /user/hduser/oozieProject/workflow/output</command>
        </sqoop>        
    <ok to="end"/>
    <error to="killJobAction"/>
    </action>
<kill name="killJobAction">
<message>"Killed job due to error: ${wf:errorMessage(wf:lastErrorNode())}"</message>
</kill>
<end name="end" />

请指教哪里出了问题？

谢谢

Answer 1

它不会永远运行。你只需要等待。

首先，您在上面看到的 Sqoop 导出作业只是一个 Oozie 计划作业。和 Heart beat意味着它现在正在运行。你只需要等待。其实你可以去 YARN 资源管理器页面(通常是 http://$namenode:8088/cluster)，然后你可以找到“真正的”Sqoop 导出作业。 (我猜映射器的默认数量是 4。)

其次，Sqoop 使用 INSERT 进行“导出”语句，所以比较慢。当表很大时，我不建议使用 Sqoop 导出，例如，当它有超过 100 万个条目时。

第三，由于我注意到您尝试导出到 MySQL，您可以尝试批处理模式，它运行 INSERT这样查询:INSERT INTO <TABLE> VALUES (<ROW1>), (<ROW2>), etc.

因此您可以将命令更改为: sqoop export -D sqoop.export.records.per.statement=1000 --connect jdbc:mysql://localhost/database --username root --password myPwd --table tableName --direct --export-dir /user/hduser/oozieProject/workflow/output --batch