# ArgoCD Installation on the Kubernetes
💡
At the time of writing this blog, the latest version available on github was Kubernetes available is 1.28.3, ArgoCD -2.8.6 and Calico Open Source v3.26.3, Ubuntu 23.10, Helm - 3.13.1
💡
What I tested is Kubernetes - 1.26.10, ArgoCD - 2.8.4 and Calico - 3.26.3, Ubuntu - 22.4.03, Helm - 3.13.1
💡
Observed many issues at the time of installing the ArgoCD with Kubernetes, ArgoCD, Redis, Calico Networking issues, coredns issues, etc. So, took a lot of time to find to strike the right balance and try and test why. This Blog is for those who can just follow along (fingers crossed!!, as I have installed this way thrice to see, to check the validity of the blog)
💡
In this blog, we are going to install ArgoCD in Kubernetes as a pod which is generally the standard practice.
### Pre-requisites Make sure that Git, K8s, Calic, Helm are already installed 1. Test the Networking is it working fine or not in the current Kubernetes, because ArgoCD will heavily rely on it. Make sure that the KUBECONFIG environment variable is already configured on your system ```bash echo $KUBECONFIG ``` 1. Ports --- Keep an eye on these ports (Make sure that they are not used by other software or if required check if the firewall is blocking them, some of them are Inbound and some of them are outbound ports, Make sure, you firewalled the rules properly) -- * *Kubernetes - 2379, 2380, 6443, 10250, 10255, 10256, 10257, 10259, 30000-32767(Node Ports)* * *Calico and ArgoCD- 5556,5557,5558,6379,6443,7000,8080,8081,8082,8084,9001, 5473, 4789(UDP), 179* 1. ```bash kubectl get pods -o wide --all-namespaces ``` * When you run the above command, you should see all the pods from all the namespaces along the nodes where they are deployed. This command should work as expected on all the nodes connected in the Kubernetes 1. In the master - create a yaml file named dnsutils ```yaml apiVersion: v1 kind: Pod metadata: name: dnsutils namespace: default spec: containers: - name: dnsutils image: k8s.gcr.io/e2e-test-images/jessie-dnsutils:1.3 command: - sleep - "3600" imagePullPolicy: IfNotPresent restartPolicy: Always ``` Save it and run *kubectl apply -f dnsutils.yaml* Run the following command ```bash kubectl exec -i -t dnsutils -- nslookup kubernetes.default ``` if you see the error as *\*\*\* Can't find kubernetes.svc.cluster.local: No answer* then restart the pods of Kube-Proxy (once it's done), CoreDNS pods (once it's done) then Calico Pods. (You can observe that, some pods, shifted from the master to the worker nodes which were the pods that were getting blocked in the communication) How to restart - kubectl delete -n <namespace of the pod> pods <pod name> they will be automatically restarted because they come under replica sets or deployments or Statefulsets Run the following command ```bash kubectl exec -i -t dnsutils -- nslookup kubernetes.default ``` this time, you should be seeing *Name: kubernetes.default.svc.cluster.local* if you do, then you are in luck to go the next steps, If not and still have some issues here in this step, The Problem can persist in Networking and configuration, please look at the Calico or CoreDNS areas and also check the permissions, read the logs!! to check the issues in the coredns pods ```bash kubectl logs --namespace=kube-system -l k8s-app=kube-dns ``` If you are still finding issues - then Don't know why, by some users did have luck, in seeing the output when deleting the following network policies ```bash kubectl delete networkpolicy argocd-redis-network-policy kubectl delete networkpolicy argocd-server-network-policy ``` once that is done, restart all the pods then try the above command kubectl exec command provided above. Lastly, try this!!, Written by K8s Guys !! , Excellent Debugging Process - [Link](https://kubernetes.io/docs/tasks/administer-cluster/dns-debugging-resolution/) Once you can see the *kubernetes.default.svc.cluster.local as the output for kubectl exec -it dnsutils --nslookup kubernetes.default command, then only perform the next steps. Remember !! this command should work as expected even when run on the other nodes.* ### ArgoCD Installation ```bash kubectl create namespace argocd ``` ```bash kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml ``` I set it to NodePort process (you can set it to [port forward](https://argo-cd.readthedocs.io/en/stable/getting_started/) process if you want or if you have a load balancer already, you can set it to it) ```bash kubectl patch svc argocd-server -n argocd -p '{"spec": {"type": "NodePort"}}' ``` ### Installing ARGO CD CLI ```bash curl -LO https://github.com/argoproj/argo-cd/releases/latest/download/argocd-linux-amd64 ``` ```bash chmod +x argocd-linux-amd64 sudo mv argocd-linux-amd64 /usr/local/bin/argocd chown : /usr/local/bin/argocd ``` Now let's run the command to check the default password, provided by the ArgoCD ```bash argocd admin initial-password -n argocd ``` Node down the passphrase provided by it. As we set it to NodePort, let's see which port it is set to ```bash kubectl describe svc argocd-server -n argocd ``` In the output of the above command, we can see the HTTP and HTTPS NodePorts. Now Type ```bash argocd login : --insecure ``` observe that it will ask you for a username and password, once you provide them, it will show as login updated successfully. Now change the password as shown below ```bash argocd account update-password ``` Once, you change the password from the above step, it checks the context of k8s. ```bash kubectl config get-contexts -o name ``` ```bash kubectl config set-context --current --namespace=argocd ``` That's it !! head to your browser!! provide the <server public ip>:<nodeport> , once asked , provide username and password!! you are good to go!!