pythonipythonjupyteripython-parallel

Tasks that throw exceptions block IPython-Parallel queue

I am using IPython-Parallel's load_balanced_view and map to perform some simulations. I use the asynchronous version of the view so it doesn't block.

Whenever the function applied by mapthrows an exception, the error property of the asynchronous map is set correctly. If this happens however the errorous tasks stays in the queue forever. Is there any possibility to cancel/abort it?

As far as I understand it cancel and abort only work on jobs that are not yet running (waiting in queue).

Any Ideas how to cancel the job so the engine becomes free again?

Thanks, Martin

Solution

  • This was a bug in ipyparallel and is fixed in master.

    The related GitHub issue is here.