In our Nexus environment:
Leaf: Nexus 93180YC-FX
Spine: Nexus 9332C
The leaf nodes often show above 90% RAM utilization, sometimes as high as 99% utilization even though the actual RAM utilization on the leaf is NOT in that range. I tried overriding the default threshold for the leafs for memory to 99% and 100%, for RAM. That seems to stop the false ALERTS, but the GROUP messages like this one: "Group xxx is in a Warning state due to member status. Nexus Switches is in a Warning state"... still fire due to this false memory state and I can't turn off default group messages at all. Ideally I'd like to stop this condition from being detected at all on these nodes but if I can stop the related events (event log, not alerts) from firing that's good enough.