Lab 05 - First look at the cluster

Goal: see what the cluster is made of, and the difference between imperative and declarative. Cheat sheet: kubectl

# 1 - where am I connected, and as whom?
kubectl cluster-info
kubectl config get-contexts
kubectl config view --minify | grep server

Get into the habit of config get-contexts before anything destructive. Most production accidents are a command typed against the right object in the wrong cluster.

# 2 - the machines
kubectl get nodes -o wide
kubectl describe node control01 | head -40
kubectl describe node worker01 | grep -A6 'Allocated resources'

In describe node, three blocks matter: Conditions (is it healthy?), Taints (what refuses to run here?) and Allocated resources (how full is it?). Note that allocation is counted from Pod requests, not actual usage.

# 3 - what does the control plane run for itself?
kubectl get pods -n kube-system -o wide

You should recognise most of it from the architecture diagram: the API server, etcd, the scheduler, the controller manager, CoreDNS, and one CNI Pod per node.

Imperative

# 4 - an action, executed once
kubectl run web01 --image nginx:1.27
kubectl get pods -o wide
kubectl run web01 --image nginx:1.27      # error: already exists - not idempotent
kubectl delete pod web01

Declarative

# 5 - generate the manifest instead of writing it
cd ~
kubectl run web01 --image nginx:1.27 --dry-run=client -o yaml > web01.yaml
cat web01.yaml

--dry-run=client builds the object locally and prints it instead of sending it. This is the fastest route to a correct manifest - generate, then edit.

# 6 - apply it twice
kubectl apply -f web01.yaml
kubectl apply -f web01.yaml          # "unchanged" - idempotent
kubectl get pod web01

The second apply is a no-op because the desired state already matches. This is what makes apply safe in a pipeline and run unsuitable for one.

# 7 - change the file, apply again
sed -i 's/nginx:1.27/nginx:1.28/' web01.yaml
kubectl apply -f web01.yaml
kubectl get pod web01 -o jsonpath='{.spec.containers[0].image}{"\n"}'

Clean up

# 8 - delete with the same file you created it from
kubectl delete -f web01.yaml