Cattle-cluster-agent can't bind to port 6443

I’ve been using Rancher 2 in our development environment for a few months, hosted on DigitalOcean. We’re now working on a bare-metal installation in a test environment, with all sorts of restrictions and requirements, and have had a few problems, that we’ve managed to overcome.
But we don’t really know what to do with the following.

We’re running Rancher 2 in HA mode with a single node for the Rancher cluster.
Everything set up nicely, and we have pods running fine on the User cluster, communicating, everything is great.
But I suddenly noticed under the cattle-system namespace of the local (system) cluster, that the cattle-cluster-agent is constantly updating, and has thousands of restarts since we installed the environment earlier this week.
When viewing the log, it terminates on
time=“2020-06-30T13:22:03Z” level=info msg=“Refreshing all schemas”
time=“2020-06-30T13:22:03Z” level=fatal msg=“steve exited: listen tcp bind: address already in use”

The cattle-cluster-agent pod is running on the rancher cluster node, and running netstat on the node shows that port 6443 is taken by kube-apiserve.

As I mentioned before, currently we’re not seeing any problem with functionality in our cluster. Is this a problem that should be dealt with? Or when we add another node to the Rancher cluster this will go away?
Alternatively, is there anyway to change the port of the kube-apiserve through rancher, or to allow the cattle-cluster-agent to use a different port?
Thanks for making an amazing product!

We’ll take a look at this.

What version of rancher are you running?

I’m running Rancher v2.4.5.

Are you able to ssh onto the host where that cluster agent container is running, do a docker inspect on it and post the results here?

Hello @bicycledan.
I was not able to reproduce this issue on a 1 node HA setup running Rancher v2.4.5.

  • Ubuntu 20.04
  • Docker 19.03
  • k8s: v1.18.3

I also tried a 1 node imported RKE cluster.

  • RKE v1.1.3
  • Ubuntu 20.04
  • k8s: v1.18.3

The cattle-cluster-agent logs did not show this error in both setups.
If there are any more details that would be helpful in investigating, please post them.
Thank you.