[IPython-dev] ipcluster(LSF): some engines do not start

Florian M. Wagner wagnerfl@student.ethz...
Fri Aug 23 10:24:39 CDT 2013


Dear all,

I am starting an IPython Cluster using LSF. Sometimes all engines 
startup as expected, other times a few engines are missing. So it might 
depend on the clusters load? The ipengine error logs yield:

    2013-08-23 16:58:27.855 [IPEngineApp] Using existing profile dir:
    u'/cluster/home02/erdw/wagnerfl/.ipython/profile_cluster'
    2013-08-23 16:58:27.881 [IPEngineApp] ERROR | Couldn't start the Engine
    Traceback (most recent call last):
       File
    "/cluster/home02/erdw/wagnerfl/.local/lib64/python2.7/site-packages/IPython/parallel/apps/ipengineapp.py",
    line 342, in init_engine
         connection_info=self.connection_info,
    AttributeError: 'IPEngineApp' object has no attribute 'connection_info'

Other engines are interrupted after missing four heartbeats (3010 ms).

My changes from the default settings are:

c.IPClusterStart.delay = 5.0
c.IPClusterStart.early_shutdown = 90
c.IPEngineApp.wait_for_url_file = 60
c.EngineFactory.timeout = 20

Anything else you would recommend to tweak?

Thank you!
Florian








-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/ipython-dev/attachments/20130823/97aec39d/attachment.html 


More information about the IPython-dev mailing list