Setting up Raspberry PI Cluster with Raspberry Pi OS Lite (64 bits) 2022-04-04 Release

[jpconstantineau]

I am still having issues getting the ansible playbook to play nice and get everything installed and running.

Hardware is a set of 7 Raspberry Pi 4 B 8Gb (bought them a while ago... wouldn't be able to buy that today...) with external 120Gb USB3 SSDs

Each SSD is imaged with the Raspberry Pi Imager with the latest Raspberry Pi OS Lite (64 bits) 2022-04-04 Release.
Passwordless SSH was setup in the imager. Hostname and Wifi are not setup, Locale settings are set (timezone and keyboard layout)

All have a static reservation in the DHCP server so that they always get the same IP address.

Host.ini file is pretty simple:

[master]
192.168.50.51
192.168.50.52
192.168.50.53

[node]
192.168.50.54
192.168.50.55
192.168.50.56
192.168.50.57

[k3s_cluster:children]
master
node

The variables to run the playbook are as follows:

---
k3s_version: v1.23.4+k3s1
# this is the user that has ssh access to these machines
ansible_user: pierre
become: true
systemd_dir: /etc/systemd/system

# Set your timezone
system_timezone: "America/Regina"

# interface which will be used for flannel
flannel_iface: "eth0"

# apiserver_endpoint is virtual ip-address which will be configured on each master
apiserver_endpoint: "192.168.50.50"

# k3s_token is required  masters can talk together securely
# this token should be alpha numeric only
k3s_token: "mynotsosecrettokentthatiuse"

# change these to your liking, the only required one is--no-deploy servicelb
extra_server_args: "--no-deploy servicelb --no-deploy traefik --write-kubeconfig-mode 644"
extra_agent_args: ""

# image tag for kube-vip
kube_vip_tag_version: "v0.4.3"

# image tag for metal lb
metal_lb_speaker_tag_version: "v0.12.1"
metal_lb_controller_tag_version: "v0.12.1"

# metallb ip range for load balancer
metal_lb_ip_range: "192.168.50.60-192.168.50.69"

The following step fails:

TASK [k3s/master : Verify that all nodes actually joined (check k3s-init.service if this fails)]

With some investigation, I was able to tell that the first node started up but the second and third master haven't joined at all.
On the 2 other master nodes, doing a k3s kubectl get nodes results in the dreaded The connection to the server 127.0.0.1:6443 was refused - did you specify the right host or port?

Some more investigation of the k3s-init.service log details with `` returns a bunch of things with the following being the most alarming:

Apr 22 22:43:49 raspberrypi k3s[1203]: time="2022-04-22T22:43:49-06:00" level=error msg="CA cert validation failed: Get \"https://127.0.0.1:6443/cacerts\": x509: certificate signed by unknown authority"

I'll do one more pull from the repo and set my IPs and give it one more try as is.

From the above, is there anything that seems off in my setup?

[jpconstantineau]

Hostname is not setup

You need to set your hostnames so that they are not the same. The default hostname is "raspberrypi" and if they all have the same name, you will have problems when joining. Remember that when you do kubectl get nodes the node hostname is what's displayed in the first column.

You can refer to this post on how to setup the hostname based on what the DHCP provides. This will make your process of a simple re-image consistent across all your nodes before you apply the playbook.