Failover

How to set up a backup deployment in case of server issues

Introduction

In the rare case of unexpected server failure/crash during your event, it is highly recommended for projects to also run a backup deployment/server. This guide will guide you through the process of setting up a 2nd deployment, and the steps required to reconnect all users from your error-state world to the new one.

Setting up your Failover World

After setting up your main world, spin up a 2nd world with all of the same configurations (you can name it FAILOVER [same-name-as-main-world] for clarity on the dashboard). Please check these settings before launching:

• Matching Mod and Content version

• World Scale and Duration (if running a large world, you must set the failover world to also be large in order to handle the same capacity of users)

• Access Groups and Settings (you may want to set this as non-public until you need to run a failover).

• Streaming Settings (launch mode should match the main world)

If you require any internal configurations to be done, please reach out to your support engineer who will configure this for you.

Setting LiveConfig on your Main World

For your current world, you need to retrieve these 3 values in order to trigger a failover into your backup world:

1. World ID

2. Launch Context ID

3. Deployment ID

   1. For this value, please reach out to your Support Engineer, as this value is only accessible through internal tooling

When you need to trigger failover to your backup world, go to your main world's live config settings:

Fill in this section with the values retrieved, and click "Save Changes".

Do not save these values until you need to failover into your backup world! Once these values are saved, anyone joining your main world will be redirected into the backup world automatically.

Once saved, increment the AllClients value , found under ForceDisconnect in live config. This will force all current clients connected to your main world to rejoin into your backup world:

If successful, you should see all players going back to the loading screen, and rejoining into your failover deployment