deleted by creator

trust us! our logos are blue and our salespeople are bald and/or buff. lol ya

what is this study? why does the article not link to it and the data? what is the sample size, located where? waste of time post, downvoted.

Lenovo ThinkCentre M715q, Lenovo ThinkCentre M93p

separate cheap newer N100 cpu node for jellyfin, other encoding

Intel NUC NUC8i5BEHS for k3s control plane, little more expensive but reliable.

i usually replace Thinkcentre fans w noctua for power draw, performance, and noise. and remove wifi module, not needed, draws power, closed blob firmware, is a risk. pops out easy, no config changes needed in Debian.

amazing, well done! i run Debian on cheap used Thinkcentre PCs, run as k3s worker nodes just fine.

makertube.net, youtube is mask off at this point

https://github.com/fosrl/pangolin

use this instead of tailscale or other corporate options. tailscale is going for IPO, their service cannot survive investor greed.

just get a cheap n100 box, don’t overspend

it would be amazing if geerlingguy was the one to make youtube go mask off

ansible has a learning curve but will save you a lot of time in long run over bash.

write playbooks rhat target groups of similar hosts instead of a playbook for each distinct host, target specific hosts with -l flag of ansible-playbook.

look into molecule for testing sooner than later. helps you be more confident your plays will work as expected vs running trial and error on a host and getting it into a bad state. i run on bare metal so more important for my workflow not to wipe a folder with a typo, etc.