Argo CD v0.12 Released

The Argo Team is proud to announce the general availability of Argo CD v0.12, a declarative GitOps deployment tool for Kubernetes. We recently presented this at a recent Kubernetes community meeting, and if you’re new to Argo CD, be sure to check out the demo!

At Intuit, Argo CD manages over 1500 applications across 100+ Kubernetes clusters. The number of our development teams and applications on-boarding to Kubernetes has exploded in just the past few months, and many improvements were implemented to make Argo CD scale gracefully. Here are the highlights for this release.

Improved UI

Many improvements to the UI were made to application search discovery.

• Table view when viewing applications
• Filters on applications
• Table view when viewing application resources
• YAML editor in UI
• Switch to text-based diff instead of JSON diff
• Ability to edit application specs

Custom Health Assessments (CRD Health)

Argo CD has long been able to perform health assessments on resources, however this would only assess the health for a handful of native Kubernetes types (deployments, statefulsets, daemonsets, etc…). Now, Argo CD can perform health assessments of any CRD, by writing Lua scripts. For example, Argo CD can now understands the cert-manager Certificate and Issuer CRD and will report a Degraded status when there are issues with the cert. Here is the Lua script to support this:

hs = {}
if obj.status ~= nil then
  if obj.status.conditions ~= nil then
    for i, condition in ipairs(obj.status.conditions) do
      if condition.type == "Ready" and condition.status == "False" then
        hs.status = "Degraded"
        hs.message = condition.message
        return hs
      end
      if condition.type == "Ready" and condition.status == "True" then
        hs.status = "Healthy"
        hs.message = condition.message
        return hs
      end
    end
  end
end
hs.status = "Progressing"
hs.message = "Waiting for certificate"
return hs

Please contribute a custom health check for your favorite CRD! For an example on how to contribute a custom health check to Argo CD, see the following pull request.

Configuration Management Plugins

Argo CD introduces Config Management Plugins to support additional configuration management tools (note that Argo CD already supports Helm, Kustomize, Ksonnet and Jsonnet out-of-the-box). Using config management plugins, Argo CD can be configured to run specified commands to render manifests. This makes it possible for Argo CD to support almost any config management tool (e.g. kubecfg, kapitan, shell scripts, etc…) without any Argo CD feature enhancements or code changes. A few examples of how to customize your Argo CD instance are provided in our example repo:

https://github.com/argoproj/argocd-example-apps/tree/master/plugins

High Availability

Argo CD is now fully highly available! A set of HA manifests are provided for users who wish to run Argo CD with HA. HA is particularly useful for preserving the contents of the Argo CD state cache in the face of cluster node upgrades or other cases where the Argo CD pods may be restarted.

Improved Application Sources

Improvements were made to Argo CD’s supported config management tools:

• Kustomize 2 support
• Directories of YAML/JSON/Jsonnet can now be recursed
• Support for Jsonnet external variables and top-level arguments

Additional Prometheus Metrics

In addition to the existing metrics about health, and sync status, Argo CD now exports the following additional prometheus metrics:

• Sync counter to track sync activity and results over time
• Application reconciliation (refresh) performance metrics to track Argo CD performance and controller activity
• Argo CD API Server metrics for monitoring HTTP/gRPC requests

Using these new metrics, sophisticated dashboards can be created to gain unique insights about applications and deployment activity. An example grafana dashboard is provided:

Fuzzy Diffing

Argo CD can now be configured to ignore known differences for resource types by specifying a json pointer to a list of field paths to ignore. This helps prevent OutOfSync conditions when a users have no control over the manifests, (e.g. upstream helm charts) and the differences are safe to ignore. Ignored differences can be configured at an application level, and/or at a system level, based on a group/kind.

Resource Exclusions

Argo CD can now be configured to completely ignore entire classes of resources group/kinds. Excluding high-volume resources improves performance and memory usage, and reduces load and bandwidth to the Kubernetes API server. It also allows users to fine-tune the k8s permissions that Argo CD needs by preventing Argo CD from attempting to watch resources of a specified group/kind.

gRPC-Web Support

The CLI is now able to communicate to the Argo CD API server using gRPC-Web (HTTP1.1) using a new CLI flag `—grpc-web`. This resolves some compatibility issues with ingresses and gRPC (HTTP2), and should enable the argocd CLI to work with virtually any load balancer, ingress controller, or API gateway.

CLI features

Finally, a few additional CLI commands were added:

• argocd app edit APPNAME — edit an application spec using preferred EDITOR
• argocd proj edit PROJNAME — edit an project spec using preferred EDITOR
• argocd app patch APPNAME — patch an application spec
• argocd app patch-resource APPNAME — patch a specific resource which is part of an application

Thank you for your support!

Many thanks to our users and contributors of the Argoproj community! Thank you for your contributions, testing, feedback, and support to make Argo CD what it is today. We still have a lot more ideas to make continuous delivery in Kubernetes fun and easy, so stay tuned and help us plan future directions for the project.