Configuring Argo CD Is Messy: Here's How We Can Fix It
summary
Argo CD's operator-facing settings are scattered across three ConfigMaps plus env vars — about 261 unique config keys — causing unpredictable hot-reload behavior, embedded-YAML typos, and opaque defaults. The proposed CRD-backed configuration provides native typed fields, apply-time validation, kubectl explain docs, and hot reload for most fields. Migration uses a CLI migration command to generate a blank CR, then iterative nullable-field migration, per-component grouping, log markers for legacy vs CR values, and eventual deprecation of ConfigMaps.
AI-written summary · about
chapters (10)
- 0:00Argo CD config issues
- 1:45Problems with config maps
- 5:38Maintainer frustrations
- 7:05Benefits of CRD config
- 8:55New CRD structure design
- 11:25Migration strategy
- 13:53User migration steps
- 15:20CRD config demo
- 18:51Resources and plans
- 19:56Q&A
related talks
25mExtending Argo Projects: Customizing Argo CD and Argo Rollouts...
08mArgo CD Update
42mLet's Learn About ArgoCD
27mExploring Argo
31mAttacking Argo CD with Argo CD (and then Defending)
39mSolving configuration drift using GitOps and ArgoCD
25mSailing Across the Ocean of Argo Ecosystem Like a Pro
23mArgo at Scale: Navigating Complex Multi-Dimensional Deployments Across H...