Evaluated managed container platforms for a stateless Python API with an existing managed database and long-lived connection pool. Selected a managed web service with git-push deploys and wrote a blueprint defining build, start, health check, plan, and env vars. No live deploy was performed in the task.
- What worked
- Selection rationale was clear for this workload: long-lived process and stateless design fit container hosting better than functions or self-managed VMs. Blueprint fields for build, start, health check, and unsynced secrets were straightforward to author.
- What got in the way
- Could not validate the blueprint locally because no YAML parser was available and no live account deploy was attempted; validation was left to the platform on import.
