[Tips] – How to troubleshooting/debugging a Kubernetes installation?

Following some commands and checks for you troubleshoot a Kubernetes installation.

Checking the installation

• Check if you are able to connect in the cluster. Run:

$ kubectl version

Following an example of expected output.

Client Version: version.Info{Major:"1", Minor:"16", GitVersion:"v1.16.0", GitCommit:"2bd9643cee5b3b3a5ecbd3af49d09018f0773c77", GitTreeState:"clean", BuildDate:"2019-09-18T14:36:53Z", GoVersion:"go1.12.9", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"16", GitVersion:"v1.16.0", GitCommit:"2bd9643cee5b3b3a5ecbd3af49d09018f0773c77", GitTreeState:"clean", BuildDate:"2019-09-18T14:27:17Z", GoVersion:"go1.12.9", Compiler:"gc", Platform:"linux/amd64"}

NOTE: The certifications used to connect and be authorized is configured in the  ~/.kube/config

• Check the nodes

$ kubectl get nodes

All need to be with the Ready status, otherwise, check if the nodes are configured/joined correctly.  This is because there is no pod network for the control plane to communicate. We will install a Container Network.

Following an example off the expected result.

 Ensure that the following command, for example, was executed in the nodes after the kubeadm init. For further information see Creating a single control-plane cluster with kubeadm 

You can now join any number of machines by running the following on each node
as root:

  kubeadm join <control-plane-host>:<control-plane-port> --token <token> --discovery-token-ca-cert-hash sha256:<hash>

• Check the namespaces

$ kubectl get namespaces
NAME              STATUS   AGE
default           Active   2m47s
kube-node-lease   Active   2m50s
kube-public       Active   2m50s
kube-system       Active   2m50s

• Check the kube-system

$ kubectl get pods --namespace kube-system
NAME                               READY   STATUS    RESTARTS   AGE
coredns-5644d7b6d9-7qx89           1/1     Running   0          3m49s
coredns-5644d7b6d9-8mpgt           1/1     Running   0          3m49s
etcd-minikube                      1/1     Running   0          2m44s
kube-addon-manager-minikube        1/1     Running   0          2m40s
kube-apiserver-minikube            1/1     Running   0          2m59s
kube-controller-manager-minikube   1/1     Running   0          2m58s
kube-proxy-9z2z4                   1/1     Running   0          3m49s
kube-scheduler-minikube            1/1     Running   0          3m1s
storage-provisioner                1/1     Running   0          3m48s

Troubleshooting

• Checking the cluster logs

Unexpected outputs or misbehaving pods can be debugged in various ways. The simplest is to request the pod’s logs. Following an example.

kubectl logs --follow --namespace kube-system kube-apiserver-node-0

NOTE: Logs can be echoed to the screen in real-time with the –follow flag.

• Checking pod logs

Try to get the pod’s logs when its status is not Running

$ kubectl logs --namespace kube-system weave-net-68xm5

NOTE: When a pod has more than one container use the flag –container to inform the container that you would like to check the logs.

• Checking Cluster Events

kubectl get events --watch

 

Sponsored by: