Even though the (horrible) official docs from Kubernetes web site say
you have the option of using kubeadm
, kops
, or kubespray
you
really don’t have those options when doing a secure, on-prem private
deployment.
The kops
tool appears to be only for AWS.
The kubespray
tool is definitely designed primarily for cloud provider
deployments and requires that “your firewall must be disabled” in order
for deployments to work when doing it on-prem, which is a non-starter,
obviously. And kubespray
is just using kubeadm
anyway:
“Kubespray has started using kubeadm internally for cluster creation since v2.3 in order to consume life cycle management domain knowledge from it and offload generic OS configuration things from it, which hopefully benefits both sides.
And, really, why the fuck is official Kubernetes documentation referring to a stupid, defunct article about the “new operator world” from 20-fucking-16!?
This leaves kubeadm
as the only true option for people who actually
care about their internal infrastructure of an on-prem deployment.
I must say, the more I get into Kubernetes the more disappointed I am in the base technical ability and IT acumen of those involved. This shit is just stupid dumb, and I’m not even talking about the partial sentences, lack of basic grammar, and punctuation in the documentation. I don’t know whether I should run screaming or embrace it and work to improve it with community contributions. Don’t get mad, Rob, get busy.