How do I import a local python module when running a python script in Oozie?

Question

I have two python files - my_python_A.py and my_python_B.py. The first file references the second ( from my_python_B import * ).

I'm executing the first python file from a shell action in Oozie (ie the script is simply python my_python_A.py ), and am receiving the following error:

Traceback (most recent call last):
  File "my_python_A.py", line 2, in <module>
    from my_python_B import *
ImportError: No module named my_python_B
Failing Oozie Launcher, Main class [org.apache.oozie.action.hadoop.ShellMain], exit code [1]

Both python files are located under the same directory in HDFS. How can I get this import statement to work?

Answer 1

I faced the same issue and the way I worked around this problem was by setting the environment variable PYTHONPATH to the current working directory inside the shell script before I execute my python code

export PYTHONPATH=`pwd`
python m_python_A.py

Make sure that in your shell action you have included all the required python modules inside the <file></file> tags. Assuming that you have a shell script called sample_script.sh (inside which you have the aforementioned commands) your workflow.xml file should look something like this

<workflow-app name="shellTest" xmlns="uri:oozie:workflow:0.4">
    <start to="shell-action"/>
    <action name="shell-action">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>                
                <property>
                    <name>oozie.launcher.mapred.job.queue.name</name>
                    <value>${launcherqueue}</value>
                </property>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${mapredqueue}</value>
                </property>
            </configuration>
            <exec>sample_script.sh</exec>
            <file>${appPath}/sample_script.sh#sample_script.sh</file>
            <file>${appPath}/m_python_A.py#m_python_A.py</file>
            <file>${appPath}/m_python_B.py#m_python_B.py</file>
            <capture-output/>
        </shell>
        <ok to="end"/>
        <error to="shell-action-failed"/>
    </action>

    <kill name="shell-action-failed">
        <message>Shell action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
    </kill>

    <end name="end" />

</workflow-app>

Answer 2

What about to add

sys.path.append(os.path.join(os.path.dirname(__file__), "lib"))

in your m_python_A.py to access anything which is stored into (IE) lib/ ?