I'm consider using kubeadm
or Conjure-Up
for a bare metal deployment. I'm not sure how to set a cluster since it requires a Master and Worker Node. If I just have a single Linux machine, could the worker and master node be the same? If so, what's the best way to do that.
For kubeadm
I know that I can run kubeadm init
to set the Master Node and then use kudeadm join <token>
on subsequent machines to set as the Worker Node but I'm not sure if this is advisable.
For Conjure-Up, I just created the deployment using the kubernetes core
spell and it created a kubernetes worker and master but I'm not sure if this is the best way to go. Another option I heard is using MAAS but I don't have much experience with it.
Follow the instructions in the user manual including kubeadm init
and you will end up with a 1 node running as master. You do not have to add nodes with kubeadm join
so just skip (4/4) Joining your nodes. Do not execute kubeadm join
on your master.
To enable normal scheduling on the master you will have to remove the dedicated=master:NoSchedule
taint from it with this command:
kubectl taint nodes --all dedicated-