[IPython-dev] Parallel magic and exception

MinRK benjaminrk at gmail.com
Thu Dec 4 15:43:33 EST 2014


Do the engine processes actually die, or does the particular execution
stop? Can you provide code that reproduces the error?

CompositeError is a wrapper class on the Client for representing multiple
Exceptions that have been raised on the engines. It is raised when the
Client tries to get a result of multiple executions, and one or more of
them fail, e.g. in blocking %px, or AsyncResult.get.

-MinRK
​

On Thu, Dec 4, 2014 at 7:57 AM, thwiouz <maidos93 at laposte.net> wrote:

> Hi guys,
>
> I'm currently testing the parallel mode of IPython for a project. My
> problem
> is that I need to run a lot of
> simulations (for loop) for a Monte Carlo program with some code I wrote.
> Because I had no idea about the
> computation cost of my program, I totally handled it into a notebook. Thus
> everything is self contained
> (definition of new classes, functions and so on).
> In my last cell I do my loop and then I get the result I want.
>
> In order to parallelize my Notebook without modifying it a lot, I started
> to
> read the doc and use the %autopx
> magic which works well. In order to do so, I first create a client, then a
> view that I put into synchronous mode
> and then I execute my whole notebook.
>
> At the end, I use the view dict to get my variables back to get the results
> of the computations and everything
> should work well. Everything looks like okay, but I have got a f***ing
> error
> in my code, well this is a
> RuntimeWarning (that I filtered to an 'error') for some of my engines.
> Obviously, I do not handle the error
> right because the  engines that encountered the error stop without getting
> any stack of what happened.
>
> I saw something about CompositeError, but how do I manage it into a loop
> into the engine side?
>
> Thanks a lot,
>
>
>
> --
> View this message in context:
> http://python.6.x6.nabble.com/Parallel-magic-and-exception-tp5079639.html
> Sent from the IPython - Development mailing list archive at Nabble.com.
> _______________________________________________
> IPython-dev mailing list
> IPython-dev at scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20141204/0e08ef75/attachment.html>


More information about the IPython-dev mailing list