This repository is dedicated to my personal k3s cluster deployed with k3sup backed by Flux, SOPS and maintained by Renovate.
- flannel - default CNI provided by k3s
- local-path-provisioner - default storage class provided by k3s
- flux - GitOps tool for deploying manifests from the
cluster
directory - metallb - bare metal load balancer
- cert-manager - SSL certificates - with Cloudflare DNS challenge
- traefik - ingress controller
- system-upgrade-controller - upgrade k3s
- nfs-subdir-external-provisioner - external cluster storage
- Prometheus - cluster monitoring
- Loki - log management
- Grafana - visualization for cluster monitoring and log management
- Enable packet forwarding on the hosts and increase
max_user_watches
cat <<EOF | sudo tee /etc/sysctl.d/k8s.conf
net.bridge.bridge-nf-call-iptables = 1
net.ipv4.ip_forward=1
fs.inotify.max_user_watches=65536
EOF
sysctl --system
-
Configure DNS on nodes to use an upstream provider (e.g.
1.1.1.1
,9.9.9.9
), or router's IP (Ad-blockers should only be used on devices with a web browser) and set a static IP with/etc/dhcpcd.conf
file. -
Disable swap with
sudo dphys-swapfile swapoff && sudo dphys-swapfile uninstall && sudo update-rc.d dphys-swapfile remove && sudo rm -f /etc/init.d/dphys-swapfile && sudo service dphys-swapfile stop && sudo systemctl disable dphys-swapfile.service
-
Disable syslog to reduce pressure on SD card
sudo service rsyslog stop
sudo systemctl disable rsyslog
- Configure logrotate for all logs with
/etc/logrotate.conf
# use daily rotation
daily
# keep 3 days worth of backlogs
rotate 3
copytruncate
- Disable leds all on Raspberry Pi with
/boot/config.txt
[pi4]
# Disable the PWR LED
dtparam=pwr_led_trigger=none
dtparam=pwr_led_activelow=off
# Disable the Activity LED
dtparam=act_led_trigger=none
dtparam=act_led_activelow=off
# Disable ethernet port LEDs
dtparam=eth_led0=4
dtparam=eth_led1=4
# Create local directory to mount
sudo mkdir -p /nfs/rpi-tftpboot
# Mount NFS directory
sudo mount -t nfs -O proto=tcp,port=2049,rw,all_squash,anonuid=1001,anongid=1001 192.168.1.10:/volume1/rpi-tftpboot /nfs/rpi-tftpboot -vvv
# Get node UUID
vcgencmd otp_dump | grep 28: | sed s/.*://g
# Create directory for each node (replace b676264b with node UUID)
sudo mkdir -p /nfs/rpi-tftpboot/b676264b
# Copy /boot to NFS directory
sudo cp -r /boot/* /nfs/rpi-tftpboot/b676264b/
# Change /boot entry in /etc/fstab :
# 192.168.1.10:/volume1/rpi-tftpboot/b676264b /boot nfs defaults,vers=3,proto=tcp 0 0
sudo nano /etc/fstab
π CLI tools required on workstation.
Tool | Purpose |
---|---|
k3sup | Tool to install k3s on your nodes |
kubectl | Allows you to run commands against Kubernetes clusters |
flux | Operator that manages your k8s cluster based on your Git repository |
SOPS | Encrypts k8s secrets with GnuPG |
GnuPG | Encrypts and signs your data |
pinentry | Allows GnuPG to read passphrases and PIN numbers |
direnv | Exports env vars based on present working directory |
pre-commit | Runs checks pre git commit |
kustomize | Template-free way to customize application configuration |
helm | Manage Kubernetes applications |
go-task | A task runner / simpler Make alternative written in Go |
prettier | Prettier is an opinionated code formatter. |
sops-pre-commit will check to make sure we are not by accident commiting secrets un-encrypted.
After pre-commit is installed :
pre-commit install-hooks
The Git repository contains the following directories under cluster
and are ordered below by how Flux will apply them.
- base directory is the entrypoint to Flux
- crds directory contains custom resource definitions (CRDs) that need to exist globally in your cluster before anything else exists
- core directory (depends on crds) are important infrastructure applications (grouped by namespace) that should never be pruned by Flux
- apps directory (depends on core) is where applications (grouped by namespace) are placed. Flux will prune resources here if they are not tracked by Git anymore.
cluster
βββ apps
β βββ default
β βββ home-assistant
β βββ mosquitto
β βββ networking
β βββ system-upgrade
β βββ unifi
β βββ vaultwarden
βββ base
β βββ flux-system
βββ core
β βββ cert-manager
β βββ log-management
β βββ metallb-system
β βββ monitoring
β βββ monitoring-config
β βββ namespaces
β βββ nfs-provisioner
β βββ system-upgrade
βββ crds
βββ cert-manager
π SOPS with GnuPG permits to encrypt and decrypt secrets.
- Creation of a Personal GPG Key, password protected, and export of the fingerprint.
export GPG_TTY=$(tty)
export PERSONAL_KEY_NAME="First name Last name (location) <email>"
gpg --batch --full-generate-key <<EOF
Key-Type: 1
Key-Length: 4096
Subkey-Type: 1
Subkey-Length: 4096
Expire-Date: 0
Name-Real: ${PERSONAL_KEY_NAME}
EOF
gpg --list-secret-keys "${PERSONAL_KEY_NAME}"
# pub rsa4096 2021-03-11 [SC]
# 772154FFF783DE317KLCA0EC77149AC618D75581
# uid [ultimate] k8s@home (Macbook) <[email protected]>
# sub rsa4096 2021-03-11 [E]
export PERSONAL_KEY_FP=772154FFF783DE317KLCA0EC77149AC618D75581
- Creation of a Flux GPG Key and export of the fingerprint
export GPG_TTY=$(tty)
export FLUX_KEY_NAME="Cluster name (Flux) <email>"
gpg --batch --full-generate-key <<EOF
%no-protection
Key-Type: 1
Key-Length: 4096
Subkey-Type: 1
Subkey-Length: 4096
Expire-Date: 0
Name-Real: ${FLUX_KEY_NAME}
EOF
gpg --list-secret-keys "${FLUX_KEY_NAME}"
# pub rsa4096 2021-03-11 [SC]
# AB675CE4CC64251G3S9AE1DAA88ARRTY2C009E2D
# uid [ultimate] Home cluster (Flux) <[email protected]>
# sub rsa4096 2021-03-11 [E]
export FLUX_KEY_FP=AB675CE4CC64251G3S9AE1DAA88ARRTY2C009E2D
- Installation of the master node
Servicelb is replaced by metallb, traefik and metrics-server will be installed with Flux.
curl -sfL https://get.k3s.io | \
INSTALL_K3S_VERSION=v1.21.4+k3s1 sh -s - server \
--node-taint CriticalAddonsOnly=true:NoExecute \
--disable servicelb \
--disable traefik \
--disable metrics-server \
--tls-san <master-ip> \
--datastore-endpoint="postgres://<user>:<password>@<ip>:<port>/<db>?sslmode=disable"
- Join worker nodes
curl -sfL https://get.k3s.io | \
INSTALL_K3S_VERSION=v1.21.4+k3s1 \
K3S_TOKEN=<token> \
K3S_URL=https://<master-ip>:6443 sh -
- Check nodes are online
kubectl get nodes
# NAME STATUS ROLES AGE VERSION
# k3s-master Ready control-plane,master 46d v1.21.4+k3s1
# k3s-node-3 Ready <none> 46d v1.21.4+k3s1
# k3s-node-2 Ready <none> 46d v1.21.4+k3s1
# k3s-node-1 Ready <none> 46d v1.21.4+k3s1
- Verify Flux can be installed
flux --kubeconfig=./.kube/config check --pre
# βΊ checking prerequisites
# β kubectl 1.21.4 >=1.18.0-0
# β Kubernetes 1.21.4+k3s1 >=1.16.0-0
# β prerequisites checks passed
- Pre-create the
flux-system
namespace
kubectl create namespace flux-system --dry-run=client -o yaml | kubectl apply -f -
- Add the Flux GPG key in-order for Flux to decrypt SOPS secrets
gpg --export-secret-keys --armor "${FLUX_KEY_FP}" |
kubectl create secret generic sops-gpg \
--namespace=flux-system \
--from-file=sops.asc=/dev/stdin
- Export more environment variables for application configuration
source .env
- Create required files based on ALL exported environment variables.
envsubst < ./tmpl/.sops.yaml > ./.sops.yaml
envsubst < ./tmpl/cluster-secrets.sops.yaml > ./cluster/base/cluster-secrets.sops.yaml
envsubst < ./tmpl/cluster-settings.yaml > ./cluster/base/cluster-settings.yaml
envsubst < ./tmpl/gotk-sync.yaml > ./cluster/base/flux-system/gotk-sync.yaml
envsubst < ./tmpl/secret.sops.yaml > ./cluster/core/cert-manager/secret.sops.yaml
-
Verify all the above files have the correct information present
-
Encrypt
cluster/cluster-secrets.sops.yaml
andcert-manager/secret.sops.yaml
with SOPS
export GPG_TTY=$(tty)
sops --encrypt --in-place ./cluster/base/cluster-secrets.sops.yaml
sops --encrypt --in-place ./cluster/core/cert-manager/secret.sops.yaml
π Variables defined in cluster-secrets.sops.yaml
and cluster-settings.sops.yaml
will be usable anywhere in your YAML manifests under ./cluster
-
Verify all the above files are encrypted with SOPS
-
Push you changes to git
git add -A
git commit -m "initial commit"
git push
- Install Flux
π Due to race conditions with the Flux CRDs this command needs to be run twice. There should be no errors on this second run.
kubectl apply --kustomize=./cluster/base/flux-system
# namespace/flux-system configured
# customresourcedefinition.apiextensions.k8s.io/alerts.notification.toolkit.fluxcd.io created
# ...
# unable to recognize "./cluster/base/flux-system": no matches for kind "Kustomization" in version "kustomize.toolkit.fluxcd.io/v1beta1"
# unable to recognize "./cluster/base/flux-system": no matches for kind "GitRepository" in version "source.toolkit.fluxcd.io/v1beta1"
# unable to recognize "./cluster/base/flux-system": no matches for kind "HelmRepository" in version "source.toolkit.fluxcd.io/v1beta1"
# unable to recognize "./cluster/base/flux-system": no matches for kind "HelmRepository" in version "source.toolkit.fluxcd.io/v1beta1"
# unable to recognize "./cluster/base/flux-system": no matches for kind "HelmRepository" in version "source.toolkit.fluxcd.io/v1beta1"
# unable to recognize "./cluster/base/flux-system": no matches for kind "HelmRepository" in version "source.toolkit.fluxcd.io/v1beta1"
kubectl get pods -n flux-system
# NAME READY STATUS RESTARTS AGE
# helm-controller-5bbd94c75-89sb4 1/1 Running 0 1h
# kustomize-controller-7b67b6b77d-nqc67 1/1 Running 0 1h
# notification-controller-7c46575844-k4bvr 1/1 Running 0 1h
# source-controller-7d6875bcb4-zqw9f 1/1 Running 0 1h
VSCode SOPS is a neat little plugin for those using VSCode. It will automatically decrypt SOPS secrets when the file is open in the editor and encrypt them when save and exit the file.
Manually sync Flux with your Git repository
flux reconcile source git flux-system
# βΊ annotating GitRepository flux-system in flux-system namespace
# β GitRepository annotated
# β waiting for GitRepository reconciliation
# β GitRepository reconciliation completed
# β fetched revision main/943e4126e74b273ff603aedab89beb7e36be4998
Show the health of you kustomizations
kubectl get kustomization -A
# NAMESPACE NAME READY STATUS AGE
# flux-system apps True Applied revision: main/943e4126e74b273ff603aedab89beb7e36be4998 3d19h
# flux-system core True Applied revision: main/943e4126e74b273ff603aedab89beb7e36be4998 4d6h
# flux-system crds True Applied revision: main/943e4126e74b273ff603aedab89beb7e36be4998 4d6h
# flux-system flux-system True Applied revision: main/943e4126e74b273ff603aedab89beb7e36be4998 4d6h
Show the health of your main Flux GitRepository
flux get sources git
# NAME READY MESSAGE REVISION SUSPENDED
# flux-system True Fetched revision: main/943e4126e74b273ff603aedab89beb7e36be4998 main/943e4126e74b273ff603aedab89beb7e36be4998 False
Show the health of your HelmRelease
s
flux get helmrelease -A
# NAMESPACE NAME READY MESSAGE REVISION SUSPENDED
# cert-manager cert-manager True Release reconciliation succeeded v1.5.2 False
# default hajimari True Release reconciliation succeeded 1.1.1 False
# networking ingress-nginx True Release reconciliation succeeded 3.30.0 False
Show the health of your HelmRepository
s
flux get sources helm -A
# NAMESPACE NAME READY MESSAGE REVISION SUSPENDED
# flux-system bitnami-charts True Fetched revision: 0ec3a3335ff991c45735866feb1c0830c4ed85cf 0ec3a3335ff991c45735866feb1c0830c4ed85cf False
# flux-system hajimari-charts True Fetched revision: 1b24af9c5a1e3da91618d597f58f46a57c70dc13 1b24af9c5a1e3da91618d597f58f46a57c70dc13 False
# flux-system ingress-nginx-charts True Fetched revision: 45669a3117fc93acc09a00e9fb9b4445e8990722 45669a3117fc93acc09a00e9fb9b4445e8990722 False
# flux-system jetstack-charts True Fetched revision: 7bad937cc82a012c9ee7d7a472d7bd66b48dc471 7bad937cc82a012c9ee7d7a472d7bd66b48dc471 False
# flux-system k8s-at-home-charts True Fetched revision: 1b24af9c5a1e3da91618d597f58f46a57c70dc13 1b24af9c5a1e3da91618d597f58f46a57c70dc13 False
-
Renovate is a very useful tool that create PRs in this Github repository when Docker images, Helm charts or anything else that can be tracked has a newer version. The configuration for renovate is located here.
-
system-upgrade-controller watch for new k3s releases and upgrade your nodes when new releases are found.
A couple of Github workflows included in this repository help automate some processes.
- Flux upgrade schedule - workflow to upgrade Flux.
- Renovate schedule - workflow to annotate
HelmRelease
's which allows Renovate to track Helm chart versions.
kubectl taint nodes k3s-node-hdw-1 hardware=true:NoSchedule
kubectl label nodes k3s-node-hdw-1 nodetype=hardware-gateway