I have 2 Nexus 5548 switches set up with a vPC peer-link between them and am trying to migrate to a pair of Nexus 9000 switches in my datacenter. Presently I have copied my config from the 5k switches over to the new 9k switches and have the 9k pair up and running with the vPC peer link active between them. One one of the 5k switches I have shut down the various ethernet and port-channel interfaces for the links that I need to move, but have left the vPC peer-link active. I then moved my cabling from that switch to one of the new 9ks.
For our first attempt at a move we shut down the same port channels and ethernet links on the remaining 5k and brought up the links on the 9k. This seemed to work ok but we were having problems pinging a few machines. Then we started getting notifications of certain services being out. We reversed the changes and nearly everything came back, but we had to go to a separate switch in the network and shut/no shut a port channel on that switch as it had been error-disabled due to a vlan misconfiguration error. Our assumption on that is when we switched over that downstream switch saw the change in MAC addresses and error disabled the port-channel. And we also thought that maybe the vPC peer-link on the 5k caused us some error here as well.
For the second attempt I tried disabling the vPC peer-link on the switch that had everything shut down already, it immediately disabled the ports on the active 5k. As this wasn't expected I reenabled the peer-link and everything came back up.
What would be the suggested method for moving my traffic to the new 9k switch pair with the minimal amount of traffic disruption. Our thought was to try things the way we did on the first attempt and be logged into that downstream switch and if the port-channel were to error-disable again we do the shut no shut and that would restore traffic going through the 9k, but I am looking for some advice on this. The traffic that gets killed is all of our voice traffic which is critical traffic.
Thanks in advance.
