Normal Traffic on vPC Peer Link Nexus 5000 with UCS

[attavit_reungratanakul]

Normally, I use vPC with vlan 99 [Allow vlan 99 one vPC peer link] but if I have new two UCS-6100 and only one uplink per 6100

so, Can I connect 6100 directly to Nexus 5000 with non-vPC [Interface connect as trunk and allow vlan 99 to UCS 6100] ?

Because  In Best Practice in Cisco document recommend to seperate non-vPC  Traffic to non-vPC peer link but in this case I cannot to that.

P.S. because when I try to use this setting it cause loop between two 6100 uplink interfaces.

Thank you.

[cpaggen]

This topology is indeed not recommended because a) you're sending potentially large amounts of traffic over the peer-link and that isn't its intended primary function   b) you could end up isolating your orphan ports (the ports going to UCS) if the peer-link fails.

However that topology isn't creating a loop. You simply have a topology with 2 orphan ports (see http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9670/design_guide_c07-625857.pdf for more information about orphan ports). If you issue a "show vpc orphan-ports" you should see the two ports going to UCS.

What is leading you to believe this is creating a loop?

[attavit_reungratanakul]

What is leading you to believe this is creating a loop?

- First of all.CNA card is Q-Logic M71KR-Q

Because when I start to implement Nexus 1000v on ESX host (in UCS Blade) . vSwitch0 I choose to go out on vmnic0 [6100-A] only but when I install Nexus1000v [VSM] on ESX host then add host to VSM using vmnic1 [6100-B] in vCenter.    VSM and vCenter freeze and I cannot do anything on these two so I look at the Nexus 5000 Interface and I see huge broadcast packet come from interface Eth1/1 that connected to 6100-A and 6100-B

Later my Core Switch [6500 Gateway] CPU Peak to 99% so I unplug Interface Eth1/1 that connect to 6100-B, process CPU on 6500 come back to normal

That's why I think it can crate loop.