Revue Kubernetes Deployment

In this repository, you can find all the configuration files needed to deploy the Revue within a Kubernetes cluster.

Remember that to deploy Revue, you have to use deploy script present in the main Revue repository.

In the following, you can find a step-by-step guide to create a K3s (a lightweight Kubernetes distribution) cluster on Raspberry PIs 5.

Creating a K3s cluster running on Raspberries PI 5

Prerequisites

• At least two Raspberry PIs that can communicate with each other.
• Preinstalled OS and available SSH connection.
• A basic understanding of Kubernetes, K3s and kubectl.

First of all, you need to set up the network configuration of the Raspberry PIs to access them. To do this, you have to edit the /etc/NetworkManager/system-connections/ file and add the following configuration:

[ipv4]
method=manual
address1=RASPBERRY_STATIC_IP(`E.g.192.168.1.30/24,192.168.1.1`)
dns=192.168.1.1;8.8.8.8;

And then

sudo systemctl restart NetworkManager

Then, to set the configuration of K3s in the Raspberry PIs, you have to go to the config.txt and cmdline.txt files in the /boot/firmware/ folder and:

In config.txt: add a line enabling arm_64bit=1 And in cmdline.txt add this: (N.B. no new line allowed) cgroup_enable=cpuset cgroup_memory=1 cgroup_enable=memory

Then activate iptables:

sudo apt-get update
sudo apt-get install iptables
sudo iptables -F

iptables has been activated because K3s will use it to manage the network internally.

Now everything is ready to install K3s in the primary node, to install it simply run:

curl -sfL https://get.k3s.io | K3s_KUBECONFIG_MODE="644" sh -

Everything can work only with one node, but to have a real cluster you need to add at least a child node. To do this, you have to get the token of the primary node and use it to add the child node.

Primary node token retrieval:

sudo cat /var/lib/rancher/k3s/server/node-token

The output will be something like this:

K106b7da6871bbae660bcf20134027051018d8cdab619fd12c184ba388c1::server:0d0ead5bedbcacc01a74878f1dcc597a

Now, repeat the installation of K3s on the child node, but this time you have to specify the token of the primary node:

curl -sfL https://get.k3s.io | K3S_TOKEN="YOUR_TOKEN" K3S_URL="https://YOUR_PRIMARY_NODE_STATIC_IP:6443" K3S_NODE_NAME="YOUR_CHILD_NODE_NAME" sh -

The cluster, if everything went well, is now ready, and you can check the status of the nodes by running:

kubectl get nodes

Now one more part is needed to complete the cluster, the Load Balancer. In this guide, MetalLB is used, a load balancer for bare metal Kubernetes clusters.

Command to add a Load Balancer:

kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.14.8/config/manifests/metallb-native.yaml

Now that MetalLB is installed, you have to configure it giving it an IP address pool to use. This pool must be in the same subnet as the Raspberry PIs, and it must be a range of IP addresses that are not used by other devices in the network. The function is to assign an IP address to the Ingress Controller, which will be used to expose the services outside the cluster. Moreover, every service that needs to be exposed outside the cluster will use an IP address from this pool.

Example of address pool configuration:

apiVersion: metallb.io/v1beta1
kind: IPAddressPool
metadata:
    name: ip-address-pool
    namespace: metallb-system # namespace of the LB MetalLB
spec:
    addresses:
        - 192.168.1.50-192.168.1.200

To apply the configuration, you have to save it in a file and then run:

kubectl apply -f your_file.yaml

Other useful information

ARP problem

Running the cluster on Raspberries can bring to a problematic about ARP tables. This happens because Raspberries by default do not respond to ARP request when using Wi-Fi. The possible workaround is to set the raspberries in promiscuous mode running the command:

sudo ifconfig <<DEVICE>> promisc

E.g. sudo ifconfig wlan0 promisc

What is suggested to do is to add the following line at the end of the /etc/rc.local file, before the exit(0) so that the command will be executed at every boot.

To check if the ARP problem is present or to check if the command has been executed during the boot, you can run the following command:

ip link show <DEVICE>

In the output the PROMISC flag should be present.

More info in the MetalLB troubleshooting section.

Helm

specifications To properly run Revue, you need to install Helm, a package manager for Kubernetes.

Helm is used to install Traefik, Prometheus and Grafana, which are needed to run Revue.

Note that you have to set the KUBECONFIG environment variable to the path of the K3s configuration file:

export KUBECONFIG=/etc/rancher/k3s/k3s.yaml

SSH

To access the Raspberry PIs, you can use the ssh command, but to make it easier, you can add the Raspberry PIs to the ~/.ssh/config file.

To copy some configuration files in the Raspberry PIs, you can use the scp command:

scp -r your_folder/ pi@your_pi_ip:/path/to/destination

Revue deployment

Take a look to the deployment section of the Revue repository for a detailed guide to run Revue on the cluster