[IPython-User] IPython cluster: stopping

Ariel Rokem arokem@gmail....
Sat Jan 28 18:08:11 CST 2012


Hi again,

Just to follow up on that:

On Fri, Jan 27, 2012 at 3:44 PM, MinRK <benjaminrk@gmail.com> wrote:
>
>> Thanks for the report.
>>
>> It looks like your controller is being killed (hence exit status=-9),
>> and ipcluster knows that when you don't have a controller anymore,
>> your engines aren't useful so they get cleaned up.  I don't know what
>> would be killing your controller, but if, as you say, memory is
>> skyrocketing, something could be killing it to prevent gobbling up
>> resources.  I don't know how you would find that out.  If you are
>> using current master, you might use `--nodb` to disable to the Hub's
>> logging of tasks, depending on your usage.
>
>
Running ipcontroller with '--nodb' seems reduce the memory usage
substantially (thanks for making that option!) and my analysis has now
managed to run longer than in any of my other attempts. In other words,
that may have done the trick.

Thanks!

Ariel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/ipython-user/attachments/20120128/19320076/attachment.html 


More information about the IPython-User mailing list