Recently a customer has added a new ESXi server to his vSphere cluster. Since he has vCloud Standard licenses, I already created a distributed switch to allow an automatic, consistent and error-free configuration of the network settings of all servers. The network migration wizard had worked great and the new server was added and configured into the distributed switch.
Nonetheless, every vmotion attempt (both started by DRS as soon as the server entered the cluster or manually started) was failing. Customer, still used to run the classic vSphere Client, sent me a screenshot with the error. The same error was visible in the Web Client:
the “old” client was showing in fact a warning in the “VDS Status” column, but there was no further information to help debugging and solving it, as already happened in the past:
So, we started the Web Client, and because even here we had no more informations:
we decided to enable the vSphere Distributed Switch health check, a control tool that allow to check (as its name implies) the health status of a distributed switch and all its components. To enable it, you need to open the Web Client, go into the distributed switch section, open the “Manage” upper tab, then “settings”, and finally the “Health Check” section. With the “Edit” you open a popup where you can modify controls about VLAN, MTU, Teaming and Failover.
If you wait for some minutes until vCenter completes its controls, you would eventually see a summary like this one:
Funny indeed, Health Check activation made me discover other two warnings on another server!
About the “VDS Status” warning, please always check trunk configuration if you are using a teaming policy as Etherchannel/IP Hash: in my situation customer configured 5 ports on the trunk in the switch, but the ESXi server had 6 network cards in its IP Hash configuration, so we found why vmotion was failing (together with further random problems).
About the warnings on the other server, at the end MTU and VLAN were correctly configured. Health Check instead was compleining about missing CDP information, so ESXi was not able to retrieve information from the switch it was connected to, scecifically on the vmnic1:
One we modified the configuration on the switch, CDP information were finally available:
So, if you have vSphere licenses to use vSphere Distributed Switch, than also activate health check, it’s a powerful diagnostic tool to find and correct network misconfigurations.
And, unless you are forced for some reason, avoid Etherchannel and use PortID; but this is another story…