Would another (non-OpenSM) SM running on another switch mess things up? We have an IBM blade server with its own IB switch attached to our main switch. Detaching that connection made the problem go away. Reattaching it makes the problem come back. Do I have to configure something differently in opensmd.conf?<br>
<br><br><div class="gmail_quote">On Fri, Feb 15, 2008 at 3:33 PM, Hal Rosenstock <<a href="mailto:hrosenstock@xsigo.com">hrosenstock@xsigo.com</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d">On Fri, 2008-02-15 at 15:26 -0500, Chuck Hartley wrote:<br>
> I direct connected the two machines instead of running them to the<br>
> switch and I get good BW numbers. What does that indicate?<br>
<br>
</div>Maybe you have a degraded link in your subnet.<br>
<div><div></div><div class="Wj3C7c"><br>
><br>
><br>
> On Fri, Feb 15, 2008 at 3:05 PM, Chuck Hartley <<a href="mailto:hartlch14@gmail.com">hartlch14@gmail.com</a>><br>
> wrote:<br>
> Nope - none of those. The closest thing is osm_pr_rcv_init<br>
> right when the SM starts up.<br>
><br>
> BTW, we have OpenSM Rev:openib-3.0.14<br>
><br>
><br>
> On Fri, Feb 15, 2008 at 2:56 PM, Hal Rosenstock<br>
> <<a href="mailto:hrosenstock@xsigo.com">hrosenstock@xsigo.com</a>> wrote:<br>
> On Fri, 2008-02-15 at 14:47 -0500, Chuck Hartley<br>
> wrote:<br>
> > Yes, but I don't see anything in the log file tagged<br>
> with anything<br>
> > like that. Is there some other string I can search<br>
> on to locate the<br>
> > request/response messages?<br>
><br>
><br>
> Maybe search for osm_pr_rcv_respond ?<br>
><br>
><br>
> > What osm_xxx function would be sending the<br>
> request/responses? Or what<br>
> > is the message or signal called - I see various ones<br>
> starting with<br>
> > OSM_<br>
> ><br>
> ><br>
> ><br>
> > On Fri, Feb 15, 2008 at 1:52 PM, Hal Rosenstock<br>
> > <<a href="mailto:hrosenstock@xsigo.com">hrosenstock@xsigo.com</a>> wrote:<br>
> > On Fri, 2008-02-15 at 13:46 -0500, Chuck<br>
> Hartley wrote:<br>
> > > Yes, we are using OpenSM and I found where<br>
> to set the -V<br>
> > switch<br>
> > > in /etc/opensm.conf. I started opensmd<br>
> service on the<br>
> > current node<br>
> > > and looked at the log file it created. I<br>
> have no idea what<br>
> > I am<br>
> > > looking at / for... there are two items<br>
> that stand out -<br>
> > PortInfo<br>
> > > dumps and SMP dumps. What is it I am<br>
> looking for or what<br>
> > can I post<br>
> > > here?<br>
> ><br>
> ><br>
> > SA PathRecord requests/responses to your end<br>
> node.<br>
> ><br>
> ><br>
> > > On Fri, Feb 15, 2008 at 12:47 PM, Hal<br>
> Rosenstock<br>
> > > <<a href="mailto:hrosenstock@xsigo.com">hrosenstock@xsigo.com</a>> wrote:<br>
> > > On Fri, 2008-02-15 at 09:30 -0800,<br>
> Hal Rosenstock<br>
> > wrote:<br>
> > > > On Fri, 2008-02-15 at 09:23<br>
> -0800, Arlin Davis<br>
> > wrote:<br>
> > > > > I would concentrate on the<br>
> path record<br>
> > information<br>
> > > returned<br>
> > > > > from the SA and compare<br>
> against the straight<br>
> > verbs test<br>
> > > > > configuration.<br>
> > > > ><br>
> > > > > Hal/Sean, is there an easy way<br>
> to see path<br>
> > record<br>
> > > information<br>
> > > > > from the query?<br>
> > > ><br>
> > > > Most SMs have a way to display<br>
> the PathRecord<br>
> > responses<br>
> > > being returned<br>
> > > > by the SA. Would that work ? If<br>
> OpenSM is being<br>
> > used, use -V<br>
> > > on the<br>
> > > > command line for this. Contact<br>
> your vendor if a<br>
> > vendor<br>
> > > specific SM is<br>
> > > > being used and this technique<br>
> would work for your<br>
> > purposes.<br>
> > ><br>
> > ><br>
> > > Another alternative would be to<br>
> use madeye on the<br>
> > end node but<br>
> > > I don't<br>
> > > think there's much decode there so<br>
> that would need<br>
> > to be done<br>
> > > by "hand".<br>
> > ><br>
> > > -- Hal<br>
> > ><br>
> > > ><br>
> > > > -- Hal<br>
> > > ><br>
> > > > > -arlin<br>
> > > > ><br>
> > > > ><br>
> > > ><br>
> _______________________________________________<br>
> > > > general mailing list<br>
> > > > <a href="mailto:general@lists.openfabrics.org">general@lists.openfabrics.org</a><br>
> > > ><br>
> <a href="http://lists.openfabrics.org/cgi-" target="_blank">http://lists.openfabrics.org/cgi-</a><br>
> > > bin/mailman/listinfo/general<br>
> > > ><br>
> > > > To unsubscribe, please visit<br>
> > ><br>
> <a href="http://openib.org/mailman/listinfo/openib-general" target="_blank">http://openib.org/mailman/listinfo/openib-general</a><br>
> > ><br>
> _______________________________________________<br>
> > > general mailing list<br>
> > > <a href="mailto:general@lists.openfabrics.org">general@lists.openfabrics.org</a><br>
> > > <a href="http://lists.openfabrics.org/cgi-" target="_blank">http://lists.openfabrics.org/cgi-</a><br>
> > bin/mailman/listinfo/general<br>
> > ><br>
> > > To unsubscribe, please visit<br>
> > ><br>
> <a href="http://openib.org/mailman/listinfo/openib-general" target="_blank">http://openib.org/mailman/listinfo/openib-general</a><br>
> > ><br>
> ><br>
> ><br>
><br>
><br>
><br>
><br>
</div></div></blockquote></div><br>