Abstracting primary and secondary region deployments in Kargo

[scott-hpe]

So my focus today was to find a pattern in Kargo that abastracts away primary and secondary region deployments. What we do today:

• two clusters mira-us-west-2, mira-us-east-2
• at any point in time, one might be the primary and the other the secondary
• we keep track of this in a state file with simple key - value mapping to let us know which cluster is primary and which is secondary at that point in time
• the pipeline should always deploy to primary first, and then secondary after some soak time on primary.

My first thought in Kargo was to have a control stage mira and two promotion stages, mira-primary and mira-secondary. I would use a configmap to decide which region is primary and which is secondary and set the argocd-update config.apps.name to myapp-mira-us-west-2 or myapp-mira-us-east-2 from either the mira-primary or mira-secondary Kargo Stage as indicated in the configmap.

That plan came to a full stop when I realized that I cannot have multiple Kargo Stages assocated with a single ArgoCD application.

So I'm on the hunt for how others might solve the above use case.