Storage Size exceeding ILM rollover max_size

I have read through the ILM documentation and it's my understanding that if I have a rollover policy with a max_age and a max_size set, that the index should rollover when one of these conditions is met, but that's not what I'm experiencing.

I have attached a screenshot that shows my policy and a Storage Size exceeding that policy. What am I missing or doing wrong?

Thank you.

The size is based on the primary size. What you are seeing is likely the total primary+replica size.

Thanks @warkolm. Ultimately my goal is to keep the disk on my Instance from filling up. I'm trying to do that by tuning the logs to rotate and delete using ILM in order to stay under the limit, but my limits don't seem to be working.

I'm attaching a screenshot of my instance as well as my Index Management. Maybe a better question is - How does my Instance disk allocation relate to the Elasticsearch storage size?