This is the second part of the note. Previous one you are able to find here

In this note I’m describing which packages and how they had been installed and configured.

Setting up Docker CE and configuring containerd

Step 1

First, add the Docker CE repository by installing the necessary plugins:

dnf install -y dnf-plugins-core
yum config-manager --add-repo=https://download.docker.com/linux/rhel/docker-ce.repo

Step 2

Next, install containerd.io. Containerd is a runtime used for managing containers.

dnf install -y containerd.io

Step 3:

Enable and start the containerd service:

systemctl enable --now containerd

Step 4:

My environment is running behind the proxy, I added proxy environment variables to the containerd service configuration:

SERVICE_FILE="/etc/systemd/system/multi-user.target.wants/containerd.service"
PROXY_SETTINGS=$(cat <<EOF
# {mark} ANSIBLE MANAGED BLOCK
Environment="NO_PROXY=127.0.0.1,localhost,192.168.11.0/8,10.96.0.1/16,192.168.0.0/16
Environment="HTTP_PROXY=http://192.168.11.51:3128"
Environment="HTTPS_PROXY=http://192.168.11.51:3128"
EOF
)

if grep -q "^\[Service\]" $SERVICE_FILE; then
  sed -i "/^\[Service\]/a $PROXY_SETTINGS" $SERVICE_FILE
else
  echo "[Service]" >> $SERVICE_FILE
  echo "$PROXY_SETTINGS" >> $SERVICE_FILE
fi

Reload the systemd daemon and restart containerd to apply the proxy settings:

systemctl daemon-reload
systemctl restart containerd

Step 5

Copy the custom config.toml file to the containerd configuration directory:

cp files/config.toml /etc/containerd/config.toml

The content of the configuration file which I use is given below.

version = 2

[grpc]
  address = "/run/containerd/containerd.sock"
  gid = 0
  uid = 0
  max_recv_message_size = 16777216
  max_send_message_size = 16777216
  tcp_address = ""
  tcp_tls_ca = ""
  tcp_tls_cert = ""
  tcp_tls_key = ""
  


[plugins]
  [plugins."io.containerd.grpc.v1.cri"]
   sandbox_image = "registry.k8s.io/pause:3.10"
   [plugins."io.containerd.grpc.v1.cri".containerd]
      [plugins."io.containerd.grpc.v1.cri".containerd.runtimes]
        [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc]
          runtime_type = "io.containerd.runc.v2"
          [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc.options]
            SystemdCgroup = true

version = 2: Specifies the version of the containerd configuration format being used. [grpc]: Starts the configuration section for gRPC, which is the protocol containerd uses for communication. address = “/run/containerd/containerd.sock”: Defines the Unix socket that containerd will use for gRPC communication. gid = 0: Sets the group ID that will have access to the containerd socket (0 usually represents the root group). uid = 0: Sets the user ID that will own the containerd socket (0 typically represents the root user). max_recv_message_size = 16777216: Sets the maximum size, in bytes, for incoming gRPC messages (16 MB in this case). max_send_message_size = 16777216: Sets the maximum size for outgoing gRPC messages. tcp_address = "": Indicates that there is no TCP address configured for containerd to listen on, implying it will use the Unix socket only. tcp_tls_ca, tcp_tls_cert, tcp_tls_key: These lines set the TLS parameters for gRPC over TCP. In this case, they are empty, meaning no TLS is configured for TCP connections.

[plugins]: Begins the configuration section for containerd plugins.

[plugins."io.containerd.grpc.v1.cri"]: This section is for configuring the CRI (Container Runtime Interface) plugin, which allows containerd to work with Kubernetes. sandbox_image = "registry.k8s.io/pause:3.10": Specifies the container image to use for Kubernetes pods’ “pause” container, which manages the pod’s network namespace. [plugins."io.containerd.grpc.v1.cri".containerd]: Configures containerd-specific settings within the CRI plugin. [plugins."io.containerd.grpc.v1.cri".containerd.runtimes]: Begins the configuration for different container runtimes managed by containerd. [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc]: Configures the “runc” runtime, which is the default runtime for containerd. runtime_type = "io.containerd.runc.v2": Specifies the runtime type for runc. Version 2 (runc.v2) provides enhanced support for container lifecycle management. [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc.options]: Begins the options section for the “runc” runtime. SystemdCgroup = true: Configures containerd to use systemd for cgroup management, which is preferred in Kubernetes environments as it provides better resource management and compatibility.

Setting Up Kubernetes

Step 1:

To install Kubernetes components, first create the repository file and set correct permissions

cat <<EOF > /etc/yum.repos.d/kubernetes.repo
[kubernetes]
name=Kubernetes
baseurl=https://pkgs.k8s.io/core:/stable:/v1.31/rpm/
enabled=1
gpgcheck=1
gpgkey=https://pkgs.k8s.io/core:/stable:/v1.31/rpm/repodata/repomd.xml.key
exclude=kubelet kubeadm kubectl cri-tools kubernetes-cni
EOF
chmod 0644 /etc/yum.repos.d/kubernetes.repo
chown root:root /etc/yum.repos.d/kubernetes.repo

Step 2:

Install the Kubernetes tools (kubelet, kubeadm, kubectl) along with socat:

yum install -y kubelet kubeadm kubectl socat --disableexcludes=kubernetes
  • kubelet: Manages containerized applications on a local node.
  • kubeadm: Bootstraps and manages the Kubernetes cluster.
  • kubectl: Provides command-line tools to interact with the Kubernetes cluster.
  • socat: A utility for data relay, often used for debugging and network communication.
  • --disableexcludes=kubernetes: Temporarily disables any excludes set by the Kubernetes repository, ensuring the required packages can be installed.

Step 3:

Enable the kubelet service to ensure it starts automatically on system boot:

systemctl enable kubelet

The kubelet service is enabled but not started immediately.

I chose not to start Kubelet immediately because it didn’t start correctly on initial attempts. However, the kubeadm init command starts Kubelet automatically when initializing the Kubernetes cluster.

In many Kubernetes setups, it is better to delay starting kubelet until the system is fully configured:

  1. Manual Control: You may need to control when kubelet starts, especially if additional configuration steps are pending.
  2. Dependencies: The kubelet service requires other components, such as the control plane, to be ready, and starting it too early may cause issues.
  3. Avoid Errors: Starting kubelet before the environment is fully prepared can lead to failures or unexpected behavior.

You may also enjoy