Nexus1000 on ESX service console issue

[ravi.kumar]

Hi CSC members,

We have strange problem at our environment. Your help will be greatly appreciated.

Issue :  We have nexus 1000 switch with an old uplink portprofile named "uplink" now we have created a new uplink port profile named "newuplink"

            Now, when we are migrating the vmnics from old port profile to the new port profile there use to be the ping drop outs and it never stabalizes.

            While migrating vmnic0 and vmnic1, it goeas smooth without any ping drop out.

            but, while migrating vmnic2 and vmnic3 individual pring drop out to the service console happens and it never stablizes, though it does not effect,    

            no ping drops to the VMs.

Kindly assist me in resolving this issue.

Awaiting help.

Regards,

Ravi.

[simon.geary]

I'm having a similar problem at the moment (and also plan to make a post on it). The workaround for me seems to be to disable MAC Pinning on the uplinks prior to making any changes, so run 'no channel-group auto mode on mac-pinning'

That MAC pinning command is the recommended way to configure the uplinks when using Nexus 1000V with UCS but it seems to cause all sorts of trouble when moving a vmnic around, and also when creating a new VLAN. Disabling MAC pinning until the changes have been made seems to fix it but I don't consider this as a proper fix as it does interrupt traffic briefly.

[chlee3]

Ravi:  That's an interesting problem, however we'll need a bit more information to really understand what might be wrong.  Can you provide the answers to the following?

- "newuplink" port-profile configuration

- ESX vmnic to upstream switch network topology

- Upstream switch port configuration for those ports facing vmnic0,1,2,3

Alternatively if you open a TAC service request we can review track our troubleshooting efforts and review the issue over webex.

Simon:  Mac pinning is our recommended configuration in UCS and Nexus 1000v deployments.  It should not cause an outage when moving vmnics around or adding vlans.  Removing the mac pinning configuration during the change shouldn't be necessary and i think given the symptoms we may be looking at a bigger problem.  If you haven't already it might be worthwhile opening a TAC service request so we can review this issue in more depth over webex.

For your reference this is a white paper on Nexus 1000v and UCS deployments.

Best Practices in Deploying Cisco Nexus 1000V Series Switches on Cisco UCS B and C Series Cisco UCS Manager Servers

http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9902/white_paper_c11-558242.html