[Users] PortXmitWait?

Florent Parent florent.parent at calculquebec.ca
Thu Mar 13 07:23:20 PDT 2014


On Thu, Mar 13, 2014 at 8:15 AM, Peter Kjellström <cap at nsc.liu.se> wrote:

> On Thursday, March 13, 2014 07:36:59 AM Hal Rosenstock wrote:
> > Some causes of congestion are: slow receiver,...
> ...
> > >> We recently migrated our opensm from 3.2.6 to 3.3.17. In this
> upgrade, we
> > >> moved to CentOS6.5 with the stock RDMA and infiniband-diags_1.5.12-5.,
>
> Did the CentOS-6.5 upgrade include all the (compute?) nodes in the fabric
> or
> just the node running OpenSM?
>
> The reason I ask is because the 6.5 kernel has problems correctly
> scheduling
> unpinned processes on nodes with NUMA (potentially slowing down receivers
> on
> your fabric).
>

Hi Peter,

The compute nodes are still at CentOS 5.10, but we're preparing a
transition. The NUMA issue is intriguing. Can you point us to more
information about this?

Thanks
Florent
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/users/attachments/20140313/b787be6b/attachment.html>


More information about the Users mailing list