I am using HA in docker on a NUC, but experience the same issue; it seems there is a slight memory leak. memory usage after a fresh start is apprx 45% and after a couple of hours is is ramping up to 55-60%. By rebooting the NUC in time, I can keep it stable, but i has never been like that.
I stripped down a lot of HACS integrations, but without the desired result memory-wise.
Looking for a solid way to diagnose this