How you can run your first Argo Workflow on K3s ?

Welcome!

Argo Workflows is an open source container-native workflow engine for orchestrating parallel jobs on Kubernetes. Argo Workflows is implemented as a Kubernetes CRD (Custom Resource Definition). In simple words, Argo is a workflow scheduler where you can run your workflows onto a Kubernetes Cluster, you can containerize different steps within your workflow and then all those steps could be executed as a part of your workflow onto a Kubernets cluster using Argo workflows. Read more

Here we gonna create a simple Argo workflow template that will echo "hello world" using the docker/whalesay container image and this container will be running on our K3s cluster using Argo Workflow template.

Let's get started!! 🚀

Before starting

Before proceeding further with the hands-on we need to have some tools and CLI installed locally:

1. Docker: A containerization tool that we will be using for running K3s cluster on container. Download it from here. To check your installation run command docker --version.

Output:

   Docker version 20.10.11, build dea9396

1. kubectl: Kubernetes command-line tool that allows you to run commands against Kubernetes clusters. Download it from here. To check your installation run command kubectl version --client.

Output:

   Client Version: version.Info{Major:"1", Minor:"22", GitVersion:"v1.22.0", GitCommit:"c2b5237ccd9c0f1d600d3072634ca66cefdf272f", GitTreeState:"clean", BuildDate:"2021-08-04T18:03:20Z", GoVersion:"go1.16.6", Compiler:"gc", Platform:"windows/amd64"}

1. argo: Argo CLI allows to (amongst other things) submit, watch, and list workflows. Download it from here. To check your installation run command argo version.

Output:

   argo: v3.2.4
    BuildDate: 2021-11-18T00:28:05Z
    GitCommit: 8771ca279c329753e420dbdd986a9c914876b151
    GitTreeState: clean
    GitTag: v3.2.4
    GoVersion: go1.16.10
    Compiler: gc
    Platform: windows/amd64

Setting up K3s cluster locally(on Docker)

K3s is a fully conformant lightweight production-ready Kubernetes distribution. As compared to K8s, its easy to install, packaged as a single binary, requires less resources which means it's possible to run a kubernets cluster on anything from 512MB of RAM machines upwards, and we can install it in a fraction of the time, unlike regular K8s.

We are going to use K3d, which is a utility designed to easily run k3s in Docker.K3d makes it very easy to create single and multi-node k3s clusters in docker for local development on Kubernetes.

✅ Step 1: Install scripts of current latest release

curl -s https://raw.githubusercontent.com/rancher/k3d/main/install.sh | bash

If this command doesn't works, you can grab a release binary from the Github release tab and install it yourself.

✅ Step 2: Create a cluster named myK3sCluster with just a single server node

k3d cluster create myK3sCluster

This command will download the docker image of K3s and setup our cluster on container.

✅ Step 3: To check if myK3sCluster is successfully running on docker

kubectl get nodes

Output:

NAME                        STATUS   ROLES                  AGE     VERSION
k3d-myk3scluster-server-0   Ready    control-plane,master   7h57m   v1.21.7+k3s1

Installing Argo Workflows in our K3s cluster

✅ Step 1: Creating a separate namespace for argo inside our cluster

kubectl create ns argo
kubectl get ns

Output:

NAME              STATUS   AGE
default           Active   8h
kube-system       Active   8h
kube-public       Active   8h
kube-node-lease   Active   8h
argo              Active   8h

✅ Step 2: Install Argo Workflows

To get started quickly, we can use the quick start manifest which will install Argo Workflow as well as some commonly used components

kubectl apply -n argo -f https://raw.githubusercontent.com/argoproj/argo-workflows/master/manifests/quick-start-postgres.yaml

✅ Step 3: Check installation

Verify everything is up running. We need to wait till each pod has running or completed status before moving on.

kubectl get pods -n argo

Output:

NAME                                  READY   STATUS      RESTARTS   AGE
minio-79566d86cb-82c4j                1/1     Running     1          8h
postgres-546d9d68b-8dfqp              1/1     Running     1          8h
workflow-controller-558db44f7-wglh5   1/1     Running     5          8h
argo-server-5d58f6585d-7jx2p          1/1     Running     4          8h

✅ Step 4: Accessing Argo Workflows dashboard

As we are running Argo Workflows locally, we can open a port-forward so you can access the namespace/dashboard.

kubectl -n argo port-forward deployment/argo-server 2746:2746

Open the Argo dashboard using URL https://localhost:2746. Here you can see all your workflows, create them and manage them.

Creating our first Argo Workflow template

I have created a very simple workflow template to echo "hello world" using the docker/whalesay container image from DockerHub. Create a file named hello-world.yaml, and copy paste the following code.

apiVersion: argoproj.io/v1alpha1
kind: Workflow # new type of k8s spec used to create argo workflows
metadata:
  generateName: hello-world- # name of the workflow spec
spec:
  entrypoint: whalesay # specifies the initial template that should be invoked i.e whalesay
  templates:
    - name: whalesay # name of the template
      container: # container that will run when this template in invoked
        image: docker/whalesay # image that will run inside the cluster
        command: [cowsay]
        args: ["hello world"]
        resources: # limit the resources
          limits:
            memory: 32Mi
            cpu: 100m

✅ Step 1: Submitting our workflow

Make sure before running the following codes you should be the folder where you created your hello-world.yaml file.

argo submit -n argo --watch hello-world.yaml

Output:

Name:                hello-world-xpc6d
Namespace:           argo
ServiceAccount:      default
Status:              Running
reated:              Tue Jan 18 22:32:07 +0530 (6 seconds ago)
Started:             Tue Jan 18 22:32:07 +0530 (6 seconds ago)
Duration:            6 seconds
Progress:            0/0

STEP                  TEMPLATE  PODNAME            DURATION  MESSAGE
 ◷ hello-world-xpc6d  whalesay  hello-world-xpc6d  6s

You can see your workflow named hello-world-xpc6d has started running. Wait till it has finished executing. You can also see your workflow on the dashboard

✅ Step 2: Getting logs of our workflow

Here, @latest tag refers to the recent or the last submitted workflow. You can also replace @latest with the name of your workflow. For eg. in my case the workflow name would be hello-world-xpc6d.

argo list -n argo
argo get -n argo @latest
argo logs -n argo @latest

Output:

hello-world-xpc6d: time="2022-01-18T17:02:54.072Z" level=info msg="capturing logs" argo=true
hello-world-xpc6d:  _____________
hello-world-xpc6d: < hello world >
hello-world-xpc6d:  -------------
hello-world-xpc6d:     \
hello-world-xpc6d:      \
hello-world-xpc6d:       \
hello-world-xpc6d:                     ##        .
hello-world-xpc6d:               ## ## ##       ==
hello-world-xpc6d:            ## ## ## ##      ===
hello-world-xpc6d:        /""""""""""""""""___/ ===
hello-world-xpc6d:   ~~~ {~~ ~~~~ ~~~ ~~~~ ~~ ~ /  ===- ~~~
hello-world-xpc6d:        \______ o          __/
hello-world-xpc6d:         \    \        __/
hello-world-xpc6d:           \____\______/

Congratulations!! 🚀 You successfully ran your first ever simple argo workflow on K3s cluster. I hope this made you curious to explore Argo workflows more deeply.
Checkout this, here you can find different types to workflows and templates that you can try running locally.