Nexus 7010 and ESX Host

darren nixon · ‎06-08-2010

Hi,

We are currently seeing issues on a ESX Host using 10G Fibre dual connectivity to a pair of Nexus 7010's using vPC for the port channel to this ESX host which was working fine , up to this weekend. No changes had been made on the Nexus or ESX host.

We have changed the hardware path for the believed fault on a vmnic which when part of the virtual switch cause VM's on the Host to stop pinging, although we still see a CDP neighborship with the ESX from the Nexus, but changing fibre and Nexus ports has not worked.

As part of the testing , the vmware guy was removing this 'faulty' vmnic from the Virtual switch , which is part of an ether channel bundle his end. My first question is, how does the Nexus detect a link leaving the bundle for the ESX host , when the actual physical link is still up , and all they have done is software removed it from on the ESX, as the Nexus will still attempt to push traffic across both bundled links. I know there is the Cisco 1000V software , which can be used at an extra price, but is this the only option.

Any help will be gladly welcome.

Aaron Harrison · ‎06-11-2010

Hi

How the switch detects a link 'moving out' of the Etherchannel would depend on how you have configured it...

If you have used 'channel-group x mode on' under the physical ports in the channel, then it will not detect the change, and you will get problems. The solution is to ensure the config of the channel on both ends (server/switch) is consistent in this case. An inconsistent config will cause you connectivity issues.

If the switch automatically negotiates the Etherchannel (i.e. you are using LACP, and the server supports LACP) then it should detect the change.

Regards

Aaron

Please rate helpful posts...

Aaron Please remember to rate helpful posts to identify useful responses, and mark 'Answered' if appropriate!