Argo CD 3.0 and the Future of Promotions

Argo CD 3.0 and the Future of Promotions

Katie Lamkin-Fulsher: Product Manager of Platform and Open Source @ Intuit

Michael Crenshaw: Staff Software Developer @ Intuit and Lead Argo Project CD Maintainer

The core promise of GitOps is powerful: a declarative, auditable, and automated system for managing infrastructure and applications. Yet, as organizations scale, the promotion pipeline—moving changes from development to production—often becomes a complex bottleneck. What starts as a clean workflow can devolve into a tangle of CI scripts, manual gates, and fragmented tooling.

But what if the solution wasn't to build more complex scripts, but to return to Git-native principles? In a recent discussion, leading experts from Intuit—the creators of the Argo project—shared their vision for the next evolution of GitOps. This article summarizes their insights, covering the foundational updates in Argo CD 3.0 and a new tool, GitOps Promoter, that is set to redefine deployment workflows.

Argo CD 3.0: A Foundation for the Future

Argo CD 3.0 is more than just a feature release; it represents a significant leap in the project's maturity. As lead maintainer Michael Crenshaw explained, one key goal was to "flex the muscle" of cutting major versions. This prepares the community for future API evolutions without the fear and disruption often associated with major version bumps.

The release is built on three pillars:

1. Hardened Security & Performance: The new version ships with more secure defaults and delivers major performance optimizations. It specifically targets lock contention issues in the cluster cache, making Argo CD far more capable of handling large-scale, high-churn environments with thousands of concurrent resource updates.

2. The Rendered Manifest Pattern: A standout feature is first-class support for the rendered manifest pattern via the sourceHydrator field. Instead of only tracking "dry" templates (like Helm charts or Kustomize bases), Argo CD can now render the final, complete YAML and commit it to a separate Git branch. This creates a crystal-clear and immutable audit trail of the exact state being deployed, forming a critical building block for the new promotion model.

The Universal Challenge of Scaling Promotions

Many organizations today use CI tools like Jenkins to orchestrate promotions. This often involves disabling Argo CD's autosync and using CLI commands within a CI pipeline to trigger deployments. While this works, it introduces friction.

Katie Lamkin-Fulsher, a Product Manager at Intuit, detailed how this pattern created challenges even for their highly advanced platform, which handles over 2,000 pull requests a week:

• Fragmented Visibility: Developers were forced to hunt for information across Jenkins logs, the Argo CD UI, and monitoring tools like Splunk just to diagnose a single failure.

• Brittle, Imperative Scripts: Managing promotion logic in thousands of imperative Jenkins scripts made updating best practices a complex and risky endeavor.

• Difficult Quality Gates: Implementing and managing consistent quality gates across every environment was difficult to scale declaratively.

• Risky Rollbacks: Using the Argo CD UI to perform a rollback would instantly create a drift between the cluster's state and Git's state, risking the change being overwritten by the next automated CI job.

This disjointed experience, as Katie noted, "adds time which ultimately takes away from developer efficiency and developer velocity."

The Declarative Solution: GitOps Promoter

To solve these problems, the team at Intuit developed GitOps Promoter, a new controller in the Argo Labs ecosystem. It moves orchestration logic out of CI scripts and into a declarative, Kubernetes-native controller that operates entirely on pull requests.

The workflow is elegant and powerful:

1. A developer merges a feature into the main branch.

2. Argo CD's Source Hydrator renders the final manifests and commits them to a "next" branch (e.g., staging-next).

3. GitOps Promoter detects this change and automatically opens a pull request to promote the change from staging-next to the active staging branch.

4. The PR itself becomes the quality gate. The Promoter adds automated commit status checks, such as verifying that the previous environment is healthy or that integration tests have passed. The PR remains un-mergeable until all gates are green.

5. Once all checks pass, GitOps Promoter automatically merges the PR.

6. With autosync now safely enabled, Argo CD detects the change on the staging branch and deploys it.

A key point emphasized in the discussion is that these environment-specific branches are for automation, not humans. Your developers continue to work from a single main branch, following established best practices. The Promoter handles the complexity of branch management behind the scenes.

Conclusion and Further Resources

This new paradigm, built on the mature foundation of Argo CD 3.0, offers a compelling path forward for platform teams. It makes promotions declarative, auditable, and centered around the pull request—a workflow developers already master. It solves the visibility and brittleness problems of CI-driven promotions and finally allows teams to embrace autosync with confidence.

For those who want to see a live demonstration of this entire workflow, the full discussion with Katie and Michael provides an invaluable deep-dive. For those ready to get started, the official documentation is the best place to begin.

• Explore the project: GitOps Promoter Official Documentation