has anyone heard of the ucs blades crashing, like all of them, when/if you put esxi 5.5 into LACP?

[Network Manager]

our VM guy here was trying to use LACP to solve an issue with ESXi and vmotion. He was on the phone with vmware and when it came down to putting the one blade he was working on in LACP the vm ware tech said that he would not do it because it would crash all the blades? has anyone heard, read or seen this? we are using ucsm 2.1.3.

[Keny Perez]

I think he should have been confused or I am not understanding the scenario, but UCS uses LACP already... where were you trying to make that change?

-Kenny

[Network Manager]

from what i got from my vm guy, the change was going to happen on the blade in question, within the esxi program. 

[Walter Dey]

could it be related to

Note: LACP is only supported in vSphere 5.1 and 5.5, using vSphere Distributed Switches (VDS) or the Cisco Nexus 1000v.

Sample configuration of EtherChannel / Link Aggregation Control Protocol (LACP) with ESXi/ESX and Cisco/HP switches (1004048)

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1004048

[Network Manager]

yes, thanks for the article, we have 5.5. i read through it and found this one on activating LACP, 

http://kb.vmware.com/selfservice/search.do?cmd=displayKC&docType=kc&docTypeID=DT_KB_1_1&externalId=2034277

what im really wondering is if we do this will we crash the "entire" UCS blade array as the vmware guy told our vm guy?

[Network Manager]

ok, i talked to our vm guy some more and got some clarity. he stated that the vm ware techs told him that when they activate the LACP on vsphere 5.5 that it will drop all the hosts on the blade (much clearer than all the blades as i stated before). 

so what we are looking for is 1. proof of this statement. 2. if true, is there another option available that does not include dropping of the hosts? the documents from vmware say at the bottom minimal network disruption, so what is the delta for that? my idea of minimal disruption is no loss of hosts and a few lost packets. 

im good with a bit of spotty network connectivity while the system converges after a change but i wont complain if it does not drop anything.

[Walter Dey]

Are you using vswitch, or DVS, or N1k ?

I think even N1k, or VMware DVS wouldn’t help; they support LACP, but UCS FI don’t. Southbound !

To be precise:

You would not use LACP for NIC teaming because that would require the 6200s to be clustered together in a vPC domain, and that is not supported. For NIC teaming you could use standard active/standby teaming, the unique vNIC “fabric failover” capability in UCS, or in the case of a VMware host you can also use active/active NIC load balancing based on vPort-ID or Mac Pinning.

The UCS Fabric Interconnect (FI) supports LACP (Link Aggregation Control Protocol) for uplink ethernet port channels. Northbound !

[Network Manager]

thank you sir.

I kept coming back to that last sentence in the reply in my research on this, but your explanation helps very much on the southbound side. 

thanks again

[paul.mancuso]

Hello:

Has there been any change in this support for LACP? I do not believe so due to the same reasoning Walter mentioned in his post 10 months ago. But, i have a customer insisting that is not the case with the latest release of UCS and the UCS 13xx VICs? But, i have found nothing to corroborate their claim.

Paul