Day 0: Training Ground
Goal: Three VMs are ready for kubeadm: kernel, containerd, runc and the Kubernetes packages are installed. · Duration: approx. 60 min · Prerequisite: three VMs with Ubuntu 24.04
Briefing
CKA: Cluster Architecture, “Prepare underlying infrastructure” (part of the 25 %). In the exam the nodes are usually prepared – in operations you do this yourself. Anyone who has built it by hand once finds every fault faster later on.
cp1 .126 worker1 .228 worker2 .83
control-plane worker worker
4 vCPU / 12 GB 2 vCPU / 4 GB 2 vCPU / 4 GB
|________________________|________________________|
172.16.0.0/24 · Pods 192.168.0.0/16 (Calico)
Replace the IPs with your own. All drills of this day run on all three VMs unless stated otherwise.
Drill 1: Hostnames and name resolution
Set the matching name on each VM, here for cp1:
sudo hostnamectl set-hostname cp1On worker1 and worker2, use
worker1 and worker2 accordingly. Then on all
three:
printf '172.16.0.126 cp1\n172.16.0.228 worker1\n172.16.0.83 worker2\n' \
| sudo tee -a /etc/hosts
getent hosts cp1 worker1 worker2Expected:
172.16.0.126 cp1
172.16.0.228 worker1
172.16.0.83 worker2
Why? The node name in Kubernetes is the hostname. Unique names save
you guesswork in kubectl get nodes later.
Save your progress? Sign up for free
Drill 2: Swap off, load kernel modules
swapon --show
free -h | head -2Expected: swapon prints nothing, the
swap line is missing or shows 0B.
total used free shared buff/cache available
Mem: 3.8Gi 714Mi 433Mi 1.9Mi 3.0Gi 3.1Gi
If swap is active: sudo swapoff -a and comment out the
swap line in /etc/fstab. By default, the kubelet does not
start with swap active.
printf 'overlay\nbr_netfilter\n' | sudo tee /etc/modules-load.d/k8s.conf
sudo modprobe overlay
sudo modprobe br_netfilter
lsmod | grep -E '^(overlay|br_netfilter)'Expected: two lines, one with
br_netfilter and one with overlay.
Save your progress? Sign up for free
Drill 3: sysctl for the pod network
cat <<'EOF' | sudo tee /etc/sysctl.d/k8s.conf
net.ipv4.ip_forward = 1
net.bridge.bridge-nf-call-iptables = 1
net.bridge.bridge-nf-call-ip6tables = 1
EOF
sudo sysctl --system >/dev/null
sysctl net.ipv4.ip_forward net.bridge.bridge-nf-call-iptablesExpected:
net.ipv4.ip_forward = 1
net.bridge.bridge-nf-call-iptables = 1
Why? Without IP forwarding, packets between pods do not leave the
node; kubeadm init aborts with an error in the preflight
check.
Save your progress? Sign up for free
Drill 4: Install containerd
The container runtime straight from the project, with checksum:
cd /tmp
V=2.4.2
U=https://github.com/containerd/containerd/releases/download/v$V
curl -sSLO $U/containerd-$V-linux-amd64.tar.gz
curl -sSLO $U/containerd-$V-linux-amd64.tar.gz.sha256sum
sha256sum -c containerd-$V-linux-amd64.tar.gz.sha256sumExpected:
containerd-2.4.2-linux-amd64.tar.gz: OK
sudo tar Cxzf /usr/local containerd-$V-linux-amd64.tar.gz
sudo mkdir -p /usr/local/lib/systemd/system
sudo curl -sSL -o /usr/local/lib/systemd/system/containerd.service \
https://raw.githubusercontent.com/containerd/containerd/v$V/containerd.service
containerd --versionExpected:
containerd github.com/containerd/containerd/v2 v2.4.2 b7a25261e029...
Save your progress? Sign up for free
Drill 5: runc and systemd cgroups
R=1.5.3
U=https://github.com/opencontainers/runc/releases/download/v$R
curl -sSLO $U/runc.amd64
curl -sSLO $U/runc.sha256sum
grep ' runc.amd64$' runc.sha256sum | sha256sum -c
sudo install -m 755 runc.amd64 /usr/local/sbin/runc
runc --version | head -1Expected:
runc.amd64: OK
runc version 1.5.3
Now generate the default configuration and switch to the systemd cgroup driver – the kubelet has used it by default since Kubernetes 1.22, and the two must match:
sudo mkdir -p /etc/containerd
containerd config default | sudo tee /etc/containerd/config.toml >/dev/null
sudo sed -i 's/SystemdCgroup = false/SystemdCgroup = true/' /etc/containerd/config.toml
grep -n 'SystemdCgroup' /etc/containerd/config.toml
sudo systemctl daemon-reload
sudo systemctl enable --now containerd
systemctl is-active containerdExpected:
89: SystemdCgroup = true
active
Save your progress? Sign up for free
Drill 6: kubeadm, kubelet and kubectl
The packages come from the community repository
pkgs.k8s.io. The repository is split per minor version –
here v1.36, so that you practice a real upgrade to 1.37 on Day 13.
sudo apt-get update
sudo apt-get install -y apt-transport-https ca-certificates curl gpg
sudo mkdir -p -m 755 /etc/apt/keyrings
curl -fsSL https://pkgs.k8s.io/core:/stable:/v1.36/deb/Release.key \
| sudo gpg --dearmor -o /etc/apt/keyrings/kubernetes-apt-keyring.gpg
K=/etc/apt/keyrings/kubernetes-apt-keyring.gpg
echo "deb [signed-by=$K] https://pkgs.k8s.io/core:/stable:/v1.36/deb/ /" \
| sudo tee /etc/apt/sources.list.d/kubernetes.list
sudo apt-get update
apt-cache madison kubeadm | head -2Expected:
kubeadm | 1.36.6-1.1 | https://pkgs.k8s.io/core:/stable:/v1.36/deb Packages
kubeadm | 1.36.5-1.1 | https://pkgs.k8s.io/core:/stable:/v1.36/deb Packages
sudo apt-get install -y kubelet=1.36.6-1.1 kubeadm=1.36.6-1.1 kubectl=1.36.6-1.1 \
cri-tools
sudo apt-mark hold kubelet kubeadm kubectl
kubeadm version -o shortExpected:
kubelet set on hold.
kubeadm set on hold.
kubectl set on hold.
v1.36.6
Why hold? Otherwise an unintended
apt upgrade would update components outside the upgrade
order (Day 13).
Save your progress? Sign up for free
Drill 7: Point crictl at containerd
crictl talks to any CRI runtime – your tool when
kubectl no longer works.
cat <<'EOF' | sudo tee /etc/crictl.yaml
runtime-endpoint: unix:///run/containerd/containerd.sock
image-endpoint: unix:///run/containerd/containerd.sock
EOF
sudo crictl versionExpected:
Version: 0.1.0
RuntimeName: containerd
RuntimeVersion: v2.4.2
RuntimeApiVersion: v1
Save your progress? Sign up for free
Exercise: Speed Drills
- Show which packages are on
hold. Target time: 15 s - Check whether containerd uses the systemd cgroup driver – via the CRI, not via the file. Target time: 30 s
- Check in one command whether IP forwarding is active. Target time: 15 s
Solution
apt-mark showhold
sudo crictl info | grep -i systemdcgroup
sysctl net.ipv4.ip_forwardExpected:
kubeadm
kubectl
kubelet
"SystemdCgroup": true
net.ipv4.ip_forward = 1
Save your progress? Sign up for free
Checkpoint
On each of the three VMs:
systemctl is-active containerdExpected: active
kubeadm version -o shortExpected: v1.36.6
swapon --show | wc -lExpected: 0
systemctl is-active kubeletExpected: activating – the kubelet
restarts in a loop because it has no configuration yet. That changes
with kubeadm init and kubeadm join on Day 1
(depending on timing, it may also show active).
Challenge
Find out which pause image (sandbox image) containerd uses for pods.
Hint: It is in the containerd configuration.
Solution
grep -n "sandbox = " /etc/containerd/config.tomlExpected:
28: sandbox = 'registry.k8s.io/pause:3.10.2'
Save your progress? Sign up for free
Exam Tip
- In the exam, containerd and the packages are usually already installed. The question is more likely “why is this node NotReady” – then you check exactly today’s points.
pkgs.k8s.iohas one repository per minor version. For an upgrade you have to adjust the line inkubernetes.list(Day 13).
Enterprise Note
- Nodes are not built by hand but with images (Packer), Ansible or cloud-init.
- Package sources are mirrored (Artifactory, Nexus, apt-cacher) so that builds are reproducible.
- Instead of bare VMs, many teams use ready-made distributions (RKE2, Talos, OpenShift) – kubeadm remains the reference tool that the exam builds on.