vPC issue using standard vSwitch

[samcatoe]

I am trying to set up a server to run an etherchannel to two different Nexus switches over vPC. Because of the license, I am using a standard vSwitch which to my knowledge does not support LACP. Therefore I am using static Port Channels and a standard vSwitch with Route Based on IP hash enabled.

I am running into a few issues. On one server, if both NICs are enabled the server is not accessible. If either is online it is accessible. On the Nexus, only one MAC address shows for the NICs.

I opened a ticket with Cisco and they said it is likely specific to the configurations on the server and not the configurations on the Nexus switches.

Any ideas on possible causes for this issue?

[samcatoe]

Some additional information:

It appears this is only happening on the Management VMK. It is not happening on the VMs that have 1 NIC. The management VMK has two physical NICs allocated, but is acting in an active standby mode, so when both are on it is not sending traffic out of both ports, only one until the other comes online. Both NICs show as active in vCenter.

[daphnissov]

Are you open to not using a LAG? It would be far simpler for you and wouldn't require any special upstream switch configuration while still allowing you to utilize multiple uplinks simultaneously.

[samcatoe]

I have not used it before, but my understanding is that it is only available if you are using distributed vSwitches. We are using standard vSwitches. If you know of another way to create a LAG on a standard switch please let me know.

[daphnissov]

What I mean to say is it is still possible to utilize multiple uplinks simultaneously without using a LAG of any sort through a variety of teaming algorithms present in ESXi.

[a_p_]

... but is acting in an active standby mode ...

The Active/Standby configuration might be causing of your issue.

from https://kb.vmware.com/s/article/1004048​

Do not configure standby or unused uplinks with IP HASH load balancing

André

[samcatoe]

Oh I see what you're saying. As in using another teaming method and and just connecting them as trunks? That it was I am doing for the management NICs right now because I cannot get it to work the other way. I was hoping I could get the management NICs to use both physical NICs at the same time instead of only being dedicated to one physical NIC at a time.

[daphnissov]

Yes, as in just use regular switch ports and no form of LAG.

I was hoping I could get the management NICs to use both physical NICs at the same time instead of only being dedicated to one physical NIC at a time.

But what does that buy you? Using regular ports, you can still fail over and fail back if you like, and there's not enough traffic on management to utilize multiple uplinks at one time (most likely).