High Availability Requirement for Greenlight Deployment

[rautniraj]

Hello everyone,

We have a new requirement to run Greenlight in a high-availability (HA) setup.

Current Setup - We currently run the following containers on a single machine:

• Greenlight
• Redis
• PostgreSQL

This setup introduces a critical risk - if the machine goes down, all running containers become unavailable, directly affecting our uptime.

---

New Requirement

We need a solution that ensures:

1. 2–3 machines are always available as part of the Greenlight deployment.
2. A load balancer sits in front of the application.
3. If one VM becomes unavailable, the system should automatically -
   • Route traffic to another healthy VM.
   • Start/recover the required containers on that VM.

This configuration should allow us to maintain service uptime even during machine failures.

---

Assumptions

Database data is stored on an NFS-backed storage, which is assumed to have continuous availability. Because of this, DB data replication is not required.

---

Proposed Approach

We are considering implementing Docker Swarm mode to orchestrate the containers across multiple machines and provide failover capabilities.

---

Request for Input

I would appreciate everyone’s thoughts, suggestions, or alternative approaches for setting up a high-availability architecture for Greenlight.

Thank you, Niraj