No dedicated repository, and gitops-duel does not implement multi-region delivery (its EVIDENCE.md lists multi-cluster / multi-region as Not yet). This page documents the pattern; single-cluster Argo CD vs Flux sync, self-heal and recovery timings are measured in the GitOps duel.
Case 12 · Pattern
Multi-Region GitOps
One Git intent, many regions, honest lag: fan-out sync with visibility and local rollback so multi-region delivery stays auditable.
00
Evidence home
Pattern case — proof lives next door.
01
How it works
One intent — staggered truth per region.
Multi-Region GitOps by Andrey Lesnikov — declare once, sync many places, and treat lag as a first-class signal.
-
01
Declare once
Desired state and overlays live in Git — regions do not invent their own source of truth.
-
02
Fan out deliberately
Controllers sync each region with promotion rules you can explain in a review.
-
03
See the lag
Arrival time differences are visible — silent drift is the failure mode you refuse.
-
04
Rollback locally
Reversibility stays regional when blast radius demands it — without rewriting global history by panic.
02
Evidence
From problem to proof.
How the system is framed, operated, and proven — not a slide deck summary.
01
Promotion across regions is manual: overlays, gates, and rollback contracts are tribal knowledge.
02
You need a single-cluster GitOps ownership model first — start with GitOps: Argo CD & Flux or Argo CD App of Apps.
03
GitOps: Argo CD & Flux and Architecture Rehearsal prove change before multi-sky promote. EKS Platform Foundation keeps each region rebuildable.
Constraints
Regional overlays, failover state, secret distribution, traffic switching, drift, and environment-specific policy.
03
Related
Nearby systems.
GitOps: Argo CD & Flux
Teams need a clear delivery model before GitOps becomes another layer of operational confusion
Delivery as explicit platform contracts
Argo CD App of Apps
As platforms grow, application onboarding, add-ons, and environment drift become hard to reason about
Platform bootstrapable from Git
EKS Platform Foundation
Kubernetes clusters become inconsistent when networking, identity, ingress, storage, and observability are assembled per project
Clusters as a repeatable product
Available for meaningful infrastructure conversations