I managed to solve. I don’t know it was the best way to fix, but it worked! Maybe It’s important I explain a little bit about my infrastructure.
We have a Rancher management three Kubernetes cluster: alpha and homolog and Rancher. Which ones have five machines ( three controllers with also workers role and two workers that have workers role). We’d like to update a Kubernetes version in both environments. In alpha and homolog trough RKE update running well. When we did on Rancher, we had problems: “FATA [controlPlane] Failed to bring up Control Plane: Failed to create [kube-controller-manager] container on host [x.x.x.x]: Failed to create [kube-controller-manager] container on host [x.x.x.x]: Error response from daemon: Conflict. The container name “/kube-controller-manager” is already in use by container 06ea1f6dc2879b30a1f8b72eb7d18b5abd193b7705564b4c36e6de0dd5e2e803. You have to remove (or rename) that container to be able to reuse that name”. When RKE tried to update a controller, It can’t due to already exist one kube-control-manager running. The suggestion by RKE was renamed a exist container or remove. I was afraid of remove or rename and make some crash in my environment
So I try on this alternative solution: First I’ve changed RKE file, removed the controller whom showing problem before I running RKE again. It worked all machines had a Kubernetes updated.
Second, I accessed the removed machine and clean up all containers existent there. I used this command to clean up all containers: docker rm $(docker ps -a) -f. Remember this machine didn’t be more in the cluster!!
And finally, I needed put the machine who I had removed in the cluster again. So once more I’ve changed RKE file, added the controller before I running RKE.
It worked for me, but I didn’t figure out why the RKE managed to do in alpha and homolog but it has problems in Rancher.