You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many hosted operators use AWS for production relays
Reduces guesswork around Postgres, Redis, WebSockets, HTTPS, and env configuration
Proposed architecture (primary path)
Document deployment using:
Component
AWS service
Nostream (Docker)
Amazon ECS on Fargate
Container image
Amazon ECR (or existing registry)
PostgreSQL 14+
Amazon RDS for PostgreSQL
Redis
Amazon ElastiCache for Redis
HTTPS + WebSockets
Application Load Balancer (ALB)
TLS
AWS Certificate Manager (ACM)
Secrets (SECRET, credentials)
Secrets Manager or SSM Parameter Store
Logs
Amazon CloudWatch Logs
Networking
VPC (public subnets for ALB, private for ECS/RDS/ElastiCache)
ECS is proposed as the primary path because Nostream is already container-first (Dockerfile, Railway template), and managed RDS + ElastiCache fit a production relay better than a single VM running everything.
Optional appendix
EC2 + Docker Compose as a small-relay option
Non-goals (v1)
Terraform / CloudFormation / IaC in repo (can be a follow-up issue)
EKS, multi-region HA
Azure / GCP (separate issues)
Happy to implement the guide after scope is confirmed, like :-
Is ECS Fargate + RDS + ElastiCache + ALB acceptable as the primary documented path?
Docs-only v1 OK, or should IaC be in scope from the start?
Activity