Why not? On my home server, k3s (single-node master), sitting idle (no workload, no changes), has amassed as much CPU time as a full virtual machine running Home Assistant: 17.75 hours in 106 hours of uptime.
All those abstractions and cluster features do not come for free.
Well hang on a second I have two nodes running k3s. 8%??
Did you accidentally also swap out the CPU governor?
Why not? On my home server, k3s (single-node master), sitting idle (no workload, no changes), has amassed as much CPU time as a full virtual machine running Home Assistant: 17.75 hours in 106 hours of uptime.
All those abstractions and cluster features do not come for free.
I think it's an 8% drop on the previous usage, not 8% of total compute drop.