Haha yeah could have worded better. I was running a 3 node k8s cluster on Talos Linux as my homelab. Downsized/simplified to a single NAS (aka NixNAS) + NUC on nixos and split the services between them. Apps that heavily use the NAS live on it for direct file access, the rest hum along on the nuc. I sleep easier with this, rather than fighting nose breakage during upgrades or wondering when my next ceph failure will occur.
I blocked ML instance pretty quickly on my account. Heavy handed but probably for the best