Steps for Upgrading ArgoCD#

Here’s a list of relevant release pages you might find useful:

  • ArgoCD releases: link

  • KSOPS releases: link

  • Kustomize releases: link

  • SOPS releases: link

  • Helm-Secrets releases: link

First checkout the upgrade docs here to see if there have been any major/breaking changes, or additional steps that will be required when updating between versions.

Changes to operate-first/continuous-deployment#

Determine version Number#

The version number will be referred throughout this doc, it takes the form of X.X.X-Y. Where X.X.X refers to the ArgoCD version number, and the -Y suffix, refers to our custom version that iterates on the X.X.X ArgoCD version. The -Y value indicates the version number for our custom tooling. If we are simply updating the custom tooling, then we only bump -Y value. For example if we are updating our custom tooling for 1.7.8-1 but keeping ArgoCD at 1.7.8 then we would bump the version to 1.7.8-2.

We’ll use X.X.X-Y as the placeholder for this value throughout the doc.

Update Dockerfile#

Fork/Clone the operate-first/continuous-deployment repo.

Open the Dockerfile and bump the ARGO_CD_VERSION to the new ArgoCD Version. If upgrading KSOPS then also bump KSOPS_VERSION to the newer version. Similarly change SOPS_VERSION and HELM_SECRETS_VERSION if needed.

Create a pull request.

Create tag release#

Once it’s merged create a new tag in the operate-first/continuous-deployment in the form of vX.X.X-Y, where X.X.X-Y should resolve to the version number as described here. This tag will be used by downstream, and should never be changed once it’s created.

Once the tag is created, a bot will create the build and publish it to

Changes to operate-first/apps#

Fork/Clone the operate-first/apps repo.

Updating Redis image#

Upstream uses redis on docker for their image, we prefer to use quay so we don’t encounter rate-limit issues. Unfortunately this means we need to also push a new redis image to our quay org whenever newer redis is introduced in ArgoCD.

Check here, under the tag appropriate ArgoCD version tag. Search for the redis deployment and see if the tag version for redis under containers/0/image has been updated. If it has, then push this new image onto For example, if adding redis:5.0.10-alpine we would do the following:

$ podman pull redis:5.0.10-alpine
$ podman push redis:5.0.10-alpine

Next, under the images section here update the newTag field for the redis image accordingly. Folowing from our example above, we would have the following new image in our Kustomization:

  - name: redis
    newTag: 5.0.10-alpine

Update ArgoCD Manifests#

If updating ArgoCD’s version, perform the following:

  • Bump image ref to the new custom image in the root Kustomization here.

  • Ensure the CustomResourceDefinition, ClusterRoles, ClusterRoleBindings in the cluster-scope app are synced with upstream for the new ArgoCD version.

    To do this, run the following from repo root:

    # TAG is a tag from
    $ ./scripts/argocd/ ${TAG}
  • Do a search for other references to update accordingly where applicable.

Update Container Images#

Under the images section here, update the operate-first/argocd tag to vX.X.X-Y. For example, if updating to v1.8.1-2 we would have the following new image in our Kustomization:

  - name: argoproj/argocd
    newTag: v1.8.1-2

Run a manual kustomize build for the argocd overlay, and ensure that the images are updated in the deployment specs as intended.

Changes to operate-first/toolbox#

If updating any of the custom tooling (e.g. sops, ksops, helm, helm-secrets, etc.), these changes should also be brought to operate-first/toolbox, simply update the Dockerfile here, you will likely only need to update the version variables.

Update the version in the readme (to the version we’ll be tagging), and cut a new tag so that a new tag-release build is initiated.

Update Docs#

We make an attempt to list the current versions of our tooling here. So update these accordingly.