argo-cd/docs/developer-guide/running-locally.md

Running Argo CD locally

Prerequisites

1. Development Environment
2. Toolchain Guide
3. Development Cycle

Preface

During development, it is recommended to start with Argo CD running locally (outside of a K8s cluster). This will greatly speed up development, as you don't have to constantly build, push and install new Argo CD Docker images with your latest changes.

After you have tested locally, you can move to the second phase of building a docker image, running Argo CD in your cluster and testing further.

For both cases, you will need a working K8s cluster, where Argo CD will store all of its resources and configuration.

In order to have all the required resources in your cluster, you will deploy Argo CD from your development branch and then scale down all it's instances. This will ensure you have all the relevant configuration (such as Argo CD Config Maps and CRDs) in the cluster while the instances themselves are stopped.

Deploy Argo CD resources to your cluster

First push the installation manifest into argocd namespace:

kubectl create namespace argocd
kubectl apply -n argocd --server-side --force-conflicts -f manifests/install.yaml

The services you will start later assume you are running in the namespace where Argo CD is installed. You can set the current context default namespace as follows:

kubectl config set-context --current --namespace=argocd

Scale down any Argo CD instance in your cluster

Make sure that Argo CD is not running in your development cluster by scaling down the deployments:

kubectl -n argocd scale statefulset/argocd-application-controller --replicas 0
kubectl -n argocd scale deployment/argocd-dex-server --replicas 0
kubectl -n argocd scale deployment/argocd-repo-server --replicas 0
kubectl -n argocd scale deployment/argocd-server --replicas 0
kubectl -n argocd scale deployment/argocd-redis --replicas 0
kubectl -n argocd scale deployment/argocd-applicationset-controller --replicas 0
kubectl -n argocd scale deployment/argocd-notifications-controller --replicas 0

Running Argo CD locally, outside of K8s cluster

Prerequisites

1. Deploy Argo CD resources to your cluster
2. Scale down any Argo CD instance in your cluster

Start local services (virtualized toolchain)

When you use the virtualized toolchain, starting local services is as simple as running

cd argo-cd
make start

By default, Argo CD uses Docker. To use Podman instead, set the DOCKER environment variable to podman before running the make command:

cd argo-cd
DOCKER=podman make start

This will start all Argo CD services and the UI in a Docker container and expose the following ports to your host:

• The Argo CD API server on port 8080
• The Argo CD UI server on port 4000
• The Helm registry server on port 5000

You can now use either the web UI by pointing your browser to http://localhost:4000 or use the CLI against the API at http://localhost:8080. Be sure to use the --insecure and --plaintext options to the CLI. Webpack will take a while to bundle resources initially, so the first page load can take several seconds or minutes.

As an alternative to using the above command line parameters each time you call argocd CLI, you can set the following environment variables:

export ARGOCD_SERVER=127.0.0.1:8080
export ARGOCD_OPTS="--plaintext --insecure"

Start local services (local toolchain)

When you use the local toolchain, starting local services can be performed in 3 ways:

With "make start-local"

cd argo-cd
make start-local ARGOCD_GPG_ENABLED=false

With "make run"

cd argo-cd
make run ARGOCD_GPG_ENABLED=false

With "goreman start"

cd argo-cd
ARGOCD_GPG_ENABLED=false && goreman start

Any of those options will start all Argo CD services and the UI:

• The Argo CD API server on port 8080
• The Argo CD UI server on port 4000
• The Helm registry server on port 5000

Check that all programs have started:

$ goreman run status
*controller
*api-server
[...]

If some of the processes fail to start (not marked with *), check logs to see why they are not running. The logs are on DEBUG level by default. If the logs are too noisy to find the problem, try editing log levels for the commands in the Procfile in the root of the Argo CD repo.

You can now use either use the web UI by pointing your browser to http://localhost:4000 or use the CLI against the API at http://localhost:8080. Be sure to use the --insecure and --plaintext options to the CLI. Webpack will take a while to bundle resources initially, so the first page load can take several seconds or minutes.

As an alternative to using the above command line parameters each time you call argocd CLI, you can set the following environment variables:

export ARGOCD_SERVER=127.0.0.1:8080
export ARGOCD_OPTS="--plaintext --insecure"

Making code changes while Argo CD is running on your machine

Docs Changes

The doc-site can be run in a container with make serve-docs or locally using make serve-docs-local. When the doc-site process is run locally, modifying the docs will auto-reload changes in a preview of the documentation website. Once running, you can view your locally built documentation on port 8000.

The developer guide has detailed information about the doc-site process.

UI Changes

Modifying the User-Interface (by editing .tsx or .scss files) auto-reloads the changes on port 4000.

Backend Changes

Modifying the API server, repo server, or a controller requires restarting the current make start for virtualized toolchain. For make start-local with the local toolchain, it is enough to rebuild and restart only the respective service:

# Example for working on the repo server Go code, see other service names in `Procfile`
goreman run restart repo-server

CLI Changes

Modifying the CLI requires restarting the current make start or make start-local session to reflect the changes. Those targets also rebuild the CLI.

To test most CLI commands, you will need to log in.

First, get the auto-generated secret:

kubectl -n argocd get secret argocd-initial-admin-secret -o jsonpath="{.data.password}" | base64 -d; echo

Then log in using that password and username admin:

dist/argocd login localhost:8080

Running Argo CD inside of K8s cluster

Scale up Argo CD in your cluster

Once you have finished testing your changes locally and want to bring back Argo CD in your development cluster, simply scale the deployments up again:

kubectl -n argocd scale statefulset/argocd-application-controller --replicas 1
kubectl -n argocd scale deployment/argocd-applicationset-controller --replicas 1
kubectl -n argocd scale deployment/argocd-dex-server --replicas 1
kubectl -n argocd scale deployment/argocd-repo-server --replicas 1
kubectl -n argocd scale deployment/argocd-server --replicas 1
kubectl -n argocd scale deployment/argocd-redis --replicas 1
kubectl -n argocd scale deployment/argocd-notifications-controller --replicas 1

Run your own Argo CD images on your cluster

For your final tests, it might be necessary to build your own images and run them in your development cluster.

Create Docker account and login

You might need to create an account on Docker Hub if you don't have one already. Once you created your account, login from your development environment:

docker login

Create and push Docker images

You will need to push the built images to your own Docker namespace:

export IMAGE_REGISTRY=docker.io
export IMAGE_NAMESPACE=youraccount

If you don't set IMAGE_TAG in your environment, the default of :latest will be used. To change the tag, export the variable in the environment:

export IMAGE_TAG=1.5.0-myrc

Note

The image will be built for linux/amd64 platform by default. If you are running on Mac with Apple chip (ARM), you need to specify the correct buld platform by running:

export TARGET_ARCH=linux/arm64 

Then you can build & push the image in one step:

DOCKER_PUSH=true make image

To speed up building of images you may use the DEV_IMAGE option that builds the argocd binaries in the users desktop environment (instead of building everything in Docker) and copies them into the result image:

DEV_IMAGE=true DOCKER_PUSH=true make image

Note

The first run of this build task may take a long time because it needs first to build the base image first; however, once it's done, the build process should be much faster than a regular full image build in Docker.

Configure manifests for your image

With IMAGE_REGISTRY, IMAGE_NAMESPACE and IMAGE_TAG still set, run:

make manifests

or

make manifests-local

(depending on your toolchain) to build a new set of installation manifests which include your specific image reference.

Note

Do not commit these manifests to your repository. If you want to revert the changes, the easiest way is to unset IMAGE_REGISTRY, IMAGE_NAMESPACE and IMAGE_TAG from your environment and run make manifests again. This will re-create the default manifests.

Configure your cluster with custom manifests

The final step is to push the manifests to your cluster, so it will pull and run your image:

kubectl apply -n argocd --server-side --force-conflicts -f manifests/install.yaml