[Ipython-tickets] [IPython] #160: IPython1 remote controller problems triggered by mlabwrap

[IPython]

#160: IPython1 remote controller problems triggered by mlabwrap
--------------------+-------------------------------------------------------
 Reporter:  jitter  |       Owner:  bgranger
     Type:  defect  |      Status:  new     
 Priority:  normal  |   Milestone:          
Component:  saw     |     Version:          
 Severity:  normal  |    Keywords:          
--------------------+-------------------------------------------------------
 Hi!

 I'm running the current SVN version of ipython and the "saw" branch of
 ipyton1. In "normal operation" (e.g. using scipy) the cluster and the task
 farming interface work nicely. (Excellent work, btw. Thanks a lot!) The
 problem is that we have lots of legacy code written in Matlab and I'd like
 to use mlabwrap on the engines to perform some image processing tasks.

 The attached script, when executed either with "python testcase.py" or
 with "%run testcase.py" within an ipython shell aborts with a NameError
 exception:


 {{{
 Starting controller:
 Starting engines:
 Sarting 2 engines on graceland2
 Sarting 2 engines on graceland1

 Your cluster is up and running.

 For interactive use, you can make a Remote Controller with:

 import ipython1.kernel.api as kernel
 ipc = kernel.RemoteController(('graceland1',10105))

 You can then cleanly stop the cluster from IPython using:

 ipc.killAll(controller=True)

 Traceback (most recent call last):
   File "testcase.py", line 31, in ?
     testmlabwrap()
   File "testcase.py", line 24, in testmlabwrap
     remoteController.executeAll("mlabwrap.mlab.help('svd')")
   File "/usr/lib64/python2.4/site-
 packages/ipython1-0.9alpha2-py2.4.egg/ipython1
 /kernel/multienginexmlrpc.py", line 533, in executeAll
     return self.execute('all', lines, block)
   File "/usr/lib64/python2.4/site-
 packages/ipython1-0.9alpha2-py2.4.egg/ipython1
 /kernel/multienginexmlrpc.py", line 520, in execute
     result = self._executeRemoteMethod(self._server.execute,
 self._clientID, loc
 alBlock, targets, lines)
   File "/usr/lib64/python2.4/site-
 packages/ipython1-0.9alpha2-py2.4.egg/ipython1
 /kernel/multienginexmlrpc.py", line 390, in _executeRemoteMethod
     result = self._unpackageResult(rawResult)
   File "/usr/lib64/python2.4/site-
 packages/ipython1-0.9alpha2-py2.4.egg/ipython1
 /kernel/multienginexmlrpc.py", line 399, in _unpackageResult
     return self._returnOrRaise(result)
   File "/usr/lib64/python2.4/site-
 packages/ipython1-0.9alpha2-py2.4.egg/ipython1
 /kernel/multienginexmlrpc.py", line 403, in _returnOrRaise
     result.raiseException()
   File "/usr/lib64/python2.4/site-packages/twisted/python/failure.py",
 line 259,
  in raiseException
     raise self.type, self.value, self.tb
 NameError
 }}}


 What's really surprising is that the same sequence of instructions, when
 invoked line by line on an ipython shell work nicely (i.e. I don't get an
 error but the output of the help('svn') command).

 I guess that this either is caused by some race condition or by a timeout
 which is exceeded due to a blocking call to the Matlab engine.

 Note that I didn't add "clusterfile.py", but this only says to start two
 engines on graceland1 and graceland2, respectively.

 - jitter

-- 
Ticket URL: <http://projects.scipy.org/ipython/ipython/ticket/160>
IPython <http://ipython.scipy.org>
The IPython interactive Python system