[Beowulf] tcp error: Need ideas!

Bruno Coutinho coutinho at dcc.ufmg.br
Tue Jan 27 03:25:19 PST 2009


2009/1/24 Gerry Creager <gerry.creager at tamu.edu>

> Couple of follow-up notes.
>
> MTU=4500:  Had one node fall over with the same overflow errors.
> MTU=3000:  A WRF model is running, but single timesteps are executing 2.5x
> slower than MTU=1500
>
>
Are you using openmpi? Some openmpi versions always enable the naggle
algorithm:
http://www.open-mpi.org/community/lists/users/2007/07/3752.php

Perhaps tunning openmpi buffer parameters could help with performance.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.scyld.com/pipermail/beowulf/attachments/20090127/b07cf021/attachment.html


More information about the Beowulf mailing list