Open MikeTheCanuck opened 5 years ago
When we do this we will also need to think about what needs to be duplicated between staging and production environments, both to be mindful of costs as well as only testing pieces that are subject to change or impact production services in the staging environment.
For instances, I don't think we need a second VPC for the staging environment. As long as the staging services are using a different ALB and different ECS cluster, there's no way we could bring down production services.
On the cost front, duplicating the entire stack just to test an experimental ecs task configuration is much more expensive than the isolation is worth.
Summary
Creating multiple
master.yaml
templates to separate the DNS and other configuration values for the staging vs production environments.Impact
Enables us to scale the number of deployed environments, as well as to test continued CloudFormation evolution.
Impacts goals S5
Tasks
Image
URL to separate:staging
from:latest
or:production
or whatever the tagged docker images for each stage of deployment will beDefinition of Done
Containerized apps on staging URLs are reachable from a browser. All existing production services are still reachable on existing URLs.