[IPython-User] Engines die during long parallel computation

Jascha Sohl-Dickstein jascha@stanford....
Mon Aug 12 00:05:15 CDT 2013


A note on this, mostly for future people searching for help on "got stale result" error messages.  I was getting almost identical errors using ipcluster on a starcluster on Amazon EC2, and the problem turned out to be that some of the computers running the engines were running out of memory.  So, stale results messages on an ipcluster can sometimes be caused by an engine running out of memory -- check that.

-jascha



More information about the IPython-User mailing list