Automating upgrades with Ansible, fighting with APT auto-starting Cassandra service

I'm preparing to upgrade nodes from 3.11.17 to 4.1.7. I've done the upgrade successfully by hand with some dev nodes, but I am wanting to make this more repeatable with Ansible. I am running on Ubuntu servers, so I have to fight with apt auto starting the service. Wondering what would be the most kosher process of doing these upgrades without apt tossing default files at me and possibly messing stuff up. My thoughts are: 1. drain and stop one node 2. Template in my necessary cassandra.yaml and rack properties files 3. Do the install of cassandra 4.1.7 4. Cope with apt auto starting cassandra Does this spell disaster? My data isn't in the default path of the config that comes from the package maintainer (/var/lib/cassandra), and some of my config variables are no longer present in 4.1 so I would expect cassandra to crash on startup, seeing the unrecognized variables (which I came across with my dev tests). What I would REALLY like is to install cassandra and give it a good look over before starting, but I am not afforded that luxury. In numerous tests, stopping cassandra immediately after a successful install did not work, cassandra would persist.