jedijeff
Enthusiast
Enthusiast

VDS teaming and Etherchannel

We are moving to Cisco switches in our Datacenter and am doing my first ESXi host config going to Cisco Etherchannel. I am seeing some bizarre behavior in  testing failover.

First off running ESXi 6 and VDS. We did have vMotion split out, but found it was not supported to have Unused Uplinks when using Etherchannel and IP hash. So now my VDS has all 4 physical nics(10gb and yes I know overkill but thats what the box has) as all Active Uplinks. We have no Standby or Unused uplinks. Using IP-hash load balancing.

Shutting down ports from Cisco or pulling actualy fiber the failover works well, couple of dropped packets. However if I try to down an interface from ESXi using esxcl network nic down, sometimes VMs will stop pinging until I bring it backup. On the Cisco side the link still shows up. I was told that esxcli command really doesnt work well with etherchanell and teaming. Which if that is the case, then if I have a marginal nic in a Host, how could I take it offline outside of having someone pull the cable? Removing the physical adatper from the Uplink on the hosts seems to exhibit same behavior. And I cant really removed the Uplink from the Active list in the VDS, because for one that would remove it even for good hosts. And Also then I am left with an Unused uplink which is unsupported with Etherchannel and Ip-hash apparently. Very confused.

0 Kudos
1 Reply
verto
Contributor
Contributor

?

0 Kudos