hwestman / k3-infrastructure

install nodes

• Fedora media writer : https://getfedora.org/en/workstation/download/ (You can get the image there as well)
• ssh-copy-id root@192.168.1.x
• dnf check-update
• sudo dnf upgrade
• Set up NTP - it was set to manual by default in the webui for fedora
• Set NIC name to the same for all nodes like so:

sudo vim /etc/systemd/network/25-eth0.link

variant: fcos
version: 1.4.0
storage:
  files:
    - path: /etc/systemd/network/25-eth0.link
      mode: 0644
      contents:
        inline: |
          [Match]
          MACAddress=4c:cc:6a:65:1b:28
          [Link]
          Name=eth0

• During installation of fedora - make sure partition full disk if we need to, defaults to 15 gb if set to auto

  • We are attempting to stick to 15 gb til we need a new partion, if we need it we create a new one and leave OS to 15

• Add disk 2

  • mkdir /mnt/disk2
  • find UUID - lsblk -f
  • vim etc/fstab
    • Append UUID=xxxx-xxxx /mnt/disk2 ext3 defaults,errors=remount-ro 0 1
  • reboot - check disk is mounted

Install cluster

• Checkout repo (this one) - based on onedr0p/cluster-template#311
• Change interface in templates:

# ./provision/ansible/playbooks/templates/calico-installation.yaml.j2
---
apiVersion: operator.tigera.io/v1
kind: Installation
metadata:
  name: default
spec:
  calicoNetwork:
    nodeAddressAutodetectionV4:
      interface: "eth0"

and

# ./provision/ansible/playbooks/templates/kube-vip-daemonset.yaml.j2

          env:
            - name: vip_interface
              value: "eth0"

and

# ./tmpl/cluster/kube-vip-daemonset.yaml

          env:
            - name: vip_interface
              value: "eth0"

install disks on nodes

https://www.techotopia.com/index.php/Adding_a_New_Disk_Drive_to_a_Fedora_Linux_System lsblk -f

fdisk /dev/sd*

vim /etc/fstab UUID= /mnt/disk2 ext3 defaults,errors=remount-ro 0 1

Expand volume

lvextend -l +100%FREE --resizefs fedora/root

Updates from template

Changes from the template was updated with git git merge v3.14.0 --allow-unrelated-histories

Taints and labels

kubectl label nodes k3s-node-4 accelerator=coral kubectl taint nodes k3s-node-4 coral=true:NoSchedule kubectl taint nodes k3s-node-4 coral=true:NoSchedule-

kubectl label nodes k3s-node-2 ups=true

Bootstrap secrets

• See home-assistant crd for password ref

Migrate volumes from Longhorn

• Set backup target in longhorn to nfs://192.168.1.49:/volume1/longhorn-backup
• Restore backup volumes
• Set backup-routines
• Set

Adding nodes

Replacing nodes

1. kubectl delete node
2. SSH into node and run

• /usr/local/bin/k3s-killall.sh
• /usr/local/bin/k3s-uninstall.sh

3. Reboot node
4. Update ansible inventory ansible/inventory/hosts.yaml
5. Check that version of k3s_release_version still matches the cluster in ansible/inventory/group_vars/kubernets/k3s.yaml
6. Run task ansible:ping
7. Run task ansible:prepare
8. Run task ansible:install

• Based on
  • https://discord.com/channels/673534664354430999/1066771341212127242/1172721839324270622
  • https://discord.com/channels/673534664354430999/1066771341212127242/1078088800577785876
  • onedr0p/cluster-template#589

Postgres & TeslaMate

create database x; create user x with encrypted password ''; grant all privileges on database x to x;