A brief snapshot of PAWS usage shows worker cpu as less than 10% and memory as more than 60%. All of the worker systems are running a flavor that has 1 cpu to 2G ram, we can probably improve this by using a 1:4 flavor, and putting out half the number of cores (down to 22 worker cpu from 44). This should likely only be done as part of/after moving to magnum T308873.
rook@paws-k8s-control-1:~$ kubectl top nodes NAME CPU(cores) CPU% MEMORY(bytes) MEMORY% paws-k8s-control-1 332m 16% 2678Mi 69% paws-k8s-control-2 297m 14% 2707Mi 70% paws-k8s-control-3 304m 15% 2614Mi 67% paws-k8s-ingress-3 152m 7% 1251Mi 32% paws-k8s-ingress-4 152m 7% 1956Mi 50% paws-k8s-worker-2 358m 8% 4892Mi 62% paws-k8s-worker-3 312m 7% 3979Mi 50% paws-k8s-worker-4 366m 9% 3797Mi 48% paws-k8s-worker-5 513m 6% 12175Mi 76% paws-k8s-worker-6 403m 5% 12935Mi 81% paws-k8s-worker-7 336m 4% 11756Mi 73% paws-k8s-worker-8 455m 5% 9995Mi 62%