Is it normal behavior for all nodes the the Hyperscaler cluster to ran out of memory after few months without reboot. We had the issue with 11.32, no change with 11.36. Migrating from Hyperscale from 2.x to 3.x didn’t improve the situation. Only fix is to reboot each node.
They each have 512GB. After a reboot they have about each 300GB free. Most of is buff/cache.
When the have less then 10%, dashboard memory turns to red.

