It looks like `openshift_sanitize_inventory` is called from `basic_facts`, which should be called during an upgrade. I will investigate.
I've confirmed the install correctly fails when the `openshift_master_console_port` is set and the upgrade_control_plane.yml playbook is run. This is expected since we have removed that option from openshift-ansible. You will need to set up a service node port manually to run the console on a different port. We are removing some infrequently used options to simplify the install. If this is not what you're seeing, please let us know.
To be clear, both 3.9 install and upgrade should fail early with a message if `openshift_master_console_port` is set, even if you were previously running with a different port. You might need the following fix to see the message: https://github.com/openshift/openshift-ansible/pull/7471
Steps to Verified: 1. Install a 3.7 ocp cluster with console port is set to 7443, and api port is set to 443. Then upgrade to 3.9 with openshift-ansible v3.9.7, could see the upgrade failed with the error: "The OpenShift web console port must now match the API server port. If you were previoiusly running the web console on a different port than the API server, please open a support ticket." 2. Install a 3.7 cluster with console port and api port are defaulted 8443, upgrade to 3.9 with openshift-ansible v3.9.7 with parameter "openshift_master_console_port=7443" set, then the upgrade failed also with the error: "The OpenShift web console port must now match the API server port. If you were previoiusly running the web console on a different port than the API server, please open a support ticket." The bug is fixed, so move it to Verified.