[IPython-User] ipengines frequently being purged/stalling during registration (Win7)

MinRK benjaminrk@gmail....
Thu Sep 27 18:41:40 CDT 2012


I'm sorry for the trouble.  What IPython version, pyzmq version, etc. are
you using?  If it's not at least IPython-0.13 / pyzmq-2.2.0, please upgrade
first.

On Wed, Sep 26, 2012 at 2:17 PM, Matt Scruggs <mdscruggs@gmail.com> wrote:

> I'm trying out IPython's parallel capabilities for a new project at work.
> I enjoyed reading many of the tutorials and videos so far and look forward
> to using the parallel capabilities, but I've encountered a major hurdle on
> my local machine that makes me nervous about relying on IPython for this.
> I've used IPython for years as "the cool Python interpreter" but now it's
> time to get serious.
>
> Here's my issue (keep in mind, this is on my local Win7 machine with only
> background apps running):
>
> When I start an "IPcluster" on my local machine (either using
> >ipcluster.exe start -n 4, or >ipcontroller.exe and adding engines with
> >ipengine.exe manually), very frequently this happens on the engine side:
>
> [IPEngineApp] Registering with controller at tcp://127.0.0.1:60821
> [IPEngineApp] Completed registration with id 0
>
> So the engine thinks it registered! Great! But it ends up getting purged
> by the controller /very/ often (50% or more of the time!):
>
> [IPControllerApp] client::client 'e71bc993-ba22-49d0-bf70-a9871783556b'
> requested u'registration_request'
> [IPControllerApp] registration::register_engine(0,
> 'e71bc993-ba22-49d0-bf70-a9871783556b',
> ['e71bc993-ba22-49d0-bf70-a9871783556b'],
> 'e71bc993-ba22-49d0-bf70-a9871783556b')
> [IPControllerApp] registration::purging stalled registration: 0
> [IPControllerApp]
> heartbeat::handle_new_heart('e71bc993-ba22-49d0-bf70-a9871783556b')
> [IPControllerApp] heartbeat::ignoring new heart:
> 'e71bc993-ba22-49d0-bf70-a9871783556b'
>
> I've even tried setting the ping value to some absurd number and it
> doesn't seem to be a timing issue. This pretty much makes it impossible to
> even go through the most basic tutorials/tests/exploration! And I'm hoping
> to be able to load these clusters en masse automatically in the
> future...clearly I can't use IPython if this is going to happen. Although I
> suppose I could just try to launch engines, add a delay, check to see if
> they were created, and if not try again...but...ugh!
>
> I appreciate your help and insights. My next step is to see if I can
> reproduce this issue on another Windows machine. I suppose I could use a VM
> or PP (the /other/ parallel processing package :P ) but I want to use
> IPython!
>
> --
> Matthew D. Scruggs
>
>
> _______________________________________________
> IPython-User mailing list
> IPython-User@scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/ipython-user/attachments/20120927/2500aaaa/attachment.html 


More information about the IPython-User mailing list