<div dir="ltr">I do physical layer testing of ethernet switches. I want to virtualize the test servers for scalability, reliability, and a VM solution allows easy change of system configuration.<div><br></div><div>The bottom line is that testing often requires opening the network port in promiscuous mode to sniff traffic, many tests need to be able to use ethtool to view network properties (speed, duplex, link state, etc) and to set properties (auto-neg, forced speed, duplex, link state, etc).</div><div><br></div><div>We use TOE cards which also allow offloading data from the CPU to the card allowing higher throughput, some tests we use this to send ~1Gbps traffic and monitor the data rate.</div><div><br></div><div>So there are multiple scenarios, the main one being the ability to see and set physical layer properties, the other being able to offload traffic for maximum throughput. The latter can possibly be accomplished through OS layer drivers on the hypervisor but the former requires better driver access than what i currently see in the VM.</div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Feb 13, 2015 at 7:25 AM, Martin Pavlík <span dir="ltr"><<a href="mailto:mpavlik@redhat.com" target="_blank">mpavlik@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi David,<br>
<br>
maybe if you gave us a little bit more of the big picture we can find some way. What is the the reason for the actions you take? What are you trying to achieve?<br>
<span class="HOEnZb"><font color="#888888"><br>
Martin<br>
</font></span><div class="HOEnZb"><div class="h5"><br>
> On 13 Feb 2015, at 15:09, Dan Yasny <<a href="mailto:dyasny@gmail.com">dyasny@gmail.com</a>> wrote:<br>
><br>
> Still easy enough to do with vdsm-hooks<br>
><br>
> --<br>
> Dan<br>
><br>
> On Thu, Feb 12, 2015 at 6:28 PM, David Smith <<a href="mailto:dsmith@mypchelp.com">dsmith@mypchelp.com</a>> wrote:<br>
> Is there a way to directly connect a specific device (ie, NIC) to a particular VM?<br>
><br>
> I've figured out how to map a specific network interface to a VM, thats one step, but in the end I may need direct access to the PCI device itself.<br>
><br>
> Second to that, for network interfaces mapped to a particular VM, is there a way for the VM to be able to properly detect physical link state of that interface? Right now that's a 'configurable option' inside network interface settings on the VM in the manager (ie, plugged, not plugged, and link state up /down)<br>
><br>
> Thanks,<br>
> David<br>
><br>
> _______________________________________________<br>
> Users mailing list<br>
> <a href="mailto:Users@ovirt.org">Users@ovirt.org</a><br>
> <a href="http://lists.ovirt.org/mailman/listinfo/users" target="_blank">http://lists.ovirt.org/mailman/listinfo/users</a><br>
><br>
><br>
> _______________________________________________<br>
> Users mailing list<br>
> <a href="mailto:Users@ovirt.org">Users@ovirt.org</a><br>
> <a href="http://lists.ovirt.org/mailman/listinfo/users" target="_blank">http://lists.ovirt.org/mailman/listinfo/users</a><br>
<br>
</div></div></blockquote></div><br></div>