umm... what's that?
you don't have an on-prem multi-node kubernetes cluster with HA control plane, off-site karmada workload rebalancing, automated multi-site backups with point-in-time db recovery, replicated NVMe-oF storage over thunderbolt, full IaC gitops end2end, SSO, fine-grained RBAC, OTel-based distributed observability, airgapped KubeVirt agent VMs, multi-GPU local inference and a full MLOps stack in a rack next to your desk?
little homelab action today
- added 3x Noctua NF-A12x15 on the radiator as pull fans
- lowered the node by 1U in the rack for better airflow
results:
cpu idle temp from 42°C to 36°C
hdd idle temp from 46°C to 39°C
cpu peak temps are also much better only at 80°C vs throttled 95°C before
noise levels are also much lower under load!
homelab cluster is now finally ready for clankers to do their best work uninterrupted
scheduled some stuff while I was sleeping and at work













