You could upgrade the tiller-deploy by reinitiate it.

# use following command to check whether the tiller-deploy pod is running or not
$kubectl get pod -n kube-system

# delete tiller-deploy deployment
$kubectl delete deployment -n kube-system tiller-deploy

# use the same command to confirm that the tiller-deploy is deleted
$kubectl get pod -n kube-system

# use the command below to deploy tiller-deploy again
$helm init

I've seen some tricky errors show up when people upgrade tiller on ICP (see this page: https://www-01.ibm.com/support/docview.wss?uid=ibm10730087). I don't think they intend for you to upgrade individual components like that.

The intended path is probably to install a newer version of IBM Cloud Private. Version 3.1.1 uses tiller 2.9.1, so it might be what you need. That will be the most "stable" procedure for you.

If you want to try something risky and don't mind if you break your cluster, you can try this:

1. Install a newer helm client (for example, 2.9.1).
2. Delete the tiller deployment on ICP using kubectl commands.
3. Create a new tiller deployment using helm init --tiller-tls.

I've heard this has worked before, but I've never actually tried it myself.