@icedquinn @phnt @newt

> psychotic thing and freeze rolling distros in time

you bring up a good point. i think if one were to do this, having their own cr would be optimal. i used to do this when i ran kubernetes

> like do you actually have CI scripts rebuild all the kube images every week to update the base image?

when i was running kubernetes, i had every bit of software in crafted alpine images squeezing them down to nearly nothing. i got to a point that the cluster's controller (flux) would just pick up the changes automatically from git source and remedy state. very hands-off after the initial month of getting it all wired up.

these days though i only consult with kubernetes/openshift/eks/aks/etc. my personal system is much simpler on bsd and using simple mechanisms to keep things running. the most complex thing i have is monitoring tied to an alert pipeline that goes to my laptop if something starts to rock which it never does. (so far)