kubectl version issue

[hongbo-miao]

--replicas was added in version 1.19 based on https://stackoverflow.com/a/66794176/2000548

Hmm, this repo kubectl server is using 1.18.2

cloud-provider-cloudlycke/vagrant/Vagrantfile

Line 11 in e5dd154

K8S_VERSION = "1.18.2"

This is when I run it in host machine (macOS):

kubectl version --short

Client Version: v1.22.3
Server Version: v1.18.2

In README, one step it runs

kubectl run --image nginx --replicas 3 nginx-demo

When I run it in host machine, it will show error

Error: unknown flag: --replicas

[hongbo-miao]

As workaround way, I found if I ssh into one master node, then

vagrant@master-c2-1:~$ kubectl version --short
Client Version: v1.18.2
Server Version: v1.18.2
vagrant@master-c2-1:~$ export KUBECONFIG=/etc/kubernetes/admin.conf
vagrant@master-c2-1:~$ kubectl run --image nginx --replicas 3 nginx-demo
Flag --replicas has been deprecated, has no effect and will be removed in the future.
error: error loading config file "/etc/kubernetes/admin.conf": open /etc/kubernetes/admin.conf: permission denied
vagrant@master-c2-1:~$ sudo kubectl run --image nginx --replicas 3 nginx-demo
Flag --replicas has been deprecated, has no effect and will be removed in the future.
pod/nginx-demo created
vagrant@master-c2-1:~$ kubectl get pods -A
error: error loading config file "/etc/kubernetes/admin.conf": open /etc/kubernetes/admin.conf: permission denied
vagrant@master-c2-1:~$ sudo kubectl get pods -A
NAMESPACE     NAME                                  READY   STATUS    RESTARTS   AGE
default       nginx-demo                            0/1     Pending   0          21s
kube-system   coredns-66bff467f8-cffjq              0/1     Pending   0          13m
kube-system   coredns-66bff467f8-gsfmp              0/1     Pending   0          13m
kube-system   etcd-master-c2-1                      1/1     Running   0          13m
kube-system   kube-apiserver-master-c2-1            1/1     Running   0          13m
kube-system   kube-controller-manager-master-c2-1   1/1     Running   0          13m
kube-system   kube-flannel-ds-amd64-nzjp6           1/1     Running   0          11m
kube-system   kube-flannel-ds-amd64-p6l2h           1/1     Running   0          13m
kube-system   kube-proxy-rjzfs                      1/1     Running   0          13m
kube-system   kube-proxy-tpj86                      1/1     Running   0          11m
kube-system   kube-scheduler-master-c2-1            1/1     Running   0          13m
vagrant@master-c2-1:/vagrant$ exit
logout
Connection to 127.0.0.1 closed.

Howeve, as above shows,

Flag --replicas has been deprecated, has no effect and will be removed in the future.

So in this case, it only created one nginx-demo instead of 3.

If I am back to host machine at this moment, and run

kubectl apply -f manifests/cloudlycke-ccm.yaml

After a while, nginx-demo status will change from Pending to Running which is expected as README (however only 1 nginx-demo instead of 3).

kubectl get pods -A
NAMESPACE     NAME                                        READY   STATUS    RESTARTS   AGE
default       nginx-demo                                  1/1     Running   0          11m
kube-system   cloudlycke-cloud-controller-manager-zmdsn   1/1     Running   0          4m43s
kube-system   coredns-66bff467f8-cffjq                    1/1     Running   0          24m
kube-system   coredns-66bff467f8-gsfmp                    1/1     Running   0          24m
kube-system   etcd-master-c2-1                            1/1     Running   0          25m
kube-system   kube-apiserver-master-c2-1                  1/1     Running   0          24m
kube-system   kube-controller-manager-master-c2-1         1/1     Running   0          24m
kube-system   kube-flannel-ds-amd64-nzjp6                 1/1     Running   0          22m
kube-system   kube-flannel-ds-amd64-p6l2h                 1/1     Running   0          24m
kube-system   kube-proxy-rjzfs                            1/1     Running   0          24m
kube-system   kube-proxy-tpj86                            1/1     Running   0          22m
kube-system   kube-scheduler-master-c2-1                  1/1     Running   0          24m

[mikejoh]

When I run it in host machine, it will show error

Error: unknown flag: --replicas

Ah, right, that flag is now gone. I think it makes sense to bump the whole setup to v1.22+ or v1.23 of k8s basically. The same goes for the module versions of the dependencies to client and cloud provider code that is used in the CCM.

It would be fun to implement more of the cloud provider interfaces and in particular the InstancesV2, just to be on top of things.