Unable to run compact on oplog.rs in local database in atlas cluster

I seem to be having an issue with the oplog.rs@local collection causing a huge size expansion resulting in significant costs. I can run db.stats() to confirm this. It will not release the data back and I cannot seem to find a way to do it myself via compact() (permissions issues), scaling down or testing resilience. I have tried to create roles via the atlas UI and the atlas CLI to no avail. I cannot access the local db which I presume is by design? Am I correct in assuming I need to engage premium support to be able to resolve this?