[jira] [Created] (ZEPPELIN-2144) Python Interpreter although shows jobs are in running state , jobs are dependent on each other to finish

[vamsi (JIRA)]

vamsi created ZEPPELIN-2144:
-------------------------------

             Summary: Python Interpreter although shows jobs are in running 
state , jobs are dependent on each other to finish
                 Key: ZEPPELIN-2144
                 URL: https://issues.apache.org/jira/browse/ZEPPELIN-2144
             Project: Zeppelin
          Issue Type: Bug
          Components: python-interpreter
    Affects Versions: 0.6.0
         Environment: Linux , Anaconda Python 2.7
            Reporter: vamsi


Hi,

I have setup Python interpreter connected to Anaconda Python2, initially jobs 
when submitted where waiting in FIFO and got to know that Zeppelin provides two 
scheduler modes.
1 FIFO 
2.Parallel

So , I made the changes in PythonInterpreter.java to use 
SchedulerFactory.singleton().createOrGetParallelScheduler instead of FIFO.

The change though schedules jobs concurrently , I observe that when a long 
running jobs is scheduled it hold other smaller jobs which ideally should 
finish with no dependency                                                       
             ex. user 1 schedules %python
                                  import scipy
                                  scipy.test()

     user 2 schedules %python                         
                                  print ("Hello World") 


The job run by user 2 waits till job of user1 complete.

Please advise.

Thanks 
Vamsi



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)