Start custom k3s image including Argo CD

[Sebast1aan]

Using k3d I'm trying to start a custom docker image that contains Argo CD.

Dockerfile:

ARG K3S_TAG="v1.21.2-k3s1"
FROM rancher/k3s:$K3S_TAG AS k3s

RUN chmod 1777 /tmp

VOLUME /var/lib/kubelet
VOLUME /var/lib/rancher/k3s
VOLUME /var/lib/cni
VOLUME /var/log

COPY entrypoint.sh /entrypoint.sh

RUN chmod +x /entrypoint.sh

ENTRYPOINT ["/entrypoint.sh"]

entypoint.sh:

#!/bin/sh

sleep 50

# Install Argo CD
echo "Installing Argo CD..."
kubectl create namespace argocd
sleep 100
kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml

# Wait for Argo CD components to be up and running
echo "Waiting for Argo CD to be ready..."
kubectl wait --for=condition=available --timeout=600s deployment -n argocd -l app.kubernetes.io/name=argocd-server

echo "Argo CD installed. Access the UI via NodePort or Ingress."

# Keep the container running
tail -f /dev/null

When then triggering k3d using the custom image:
sudo k3d cluster create test --image=k3s-argocd
I can see that the cluster gets created correctly. I don't find any trace however from the entrypoint.sh script being executed (and the argocd namespace is non-existent as well).
If I manually go inside the docker container that is started, the entrypoint.sh can be executed without errors. Is k3d somehow overruling the entrypoint of the container, or am I missing something else?

[iwilltry42]

Hey!
Yes, k3d uses a custom entrypoint for various pre-start operations.
Using a custom image just for installing ArgoCD seems overkill 🤔
Also that entrypoint script you shared would not start K3s at all, right? It's blocking at the end (and K3s would have to start before you install ArgoCD).

What you'd probably want to do here is using K3s auto-deploy manifests functionality - checkout auto-deploying manifests here: https://k3d.io/v5.7.2/usage/k3s/?h=manifests#general-k3s-documentation

If you really need to use entrypoint scripts, you may be able to add them to the entrypoint scripts dir which will then be picked up by the main entrypoint (I'd have to look up where that was again though)

[Sebast1aan]

The auto-deploy manifests functionality indeed does the trick 👍 .
For reference, this is the manifest file I added to deploy both Argo CD and Argo workflows into the cluster:

apiVersion: v1
kind: Namespace
metadata:
  name: argocd
---
apiVersion: helm.cattle.io/v1
kind: HelmChart
metadata:
  name: argocd
  namespace: kube-system
spec:
  repo: https://argoproj.github.io/argo-helm
  chart: argo-cd
  targetNamespace: argocd
  valuesContent: |-
    server:
      service:
        type: ClusterIP
      ingress:
        enabled: true
        hosts:
          - argocd.example.com
    redis:
      enabled: true
    dex:
      enabled: true
    controller:
      metrics:
        enabled: true
---
apiVersion: helm.cattle.io/v1
kind: HelmChart
metadata:
  name: argoworkflows
  namespace: kube-system
spec:
  repo: https://argoproj.github.io/argo-helm
  chart: argo-workflows
  targetNamespace: argocd

Using k3s v1.24.17-k3s1 in my custom docker image, I was able to automatically have both Argo CD and Argo workflows in my cluster that was created by k3d.