Once you have deployed your app and are ready to launch you will probably want to disable auto-scaling to 0 and beed up your machine to handle more traffic.

You will likely see a warning on your fly dashboard stating - “Your app is currently running on a single machine To ensure high availability, especially for production apps, we’d strongly recommend running at least 2 machines”

As we are using sqlite there is no built in data replication so we have to use 1 machine!

We can however improve the single machine by bumping up the RAM or CPU count - note this will have affect on price.

​
Disable auto-stop-machines

By default your fly.toml is set to shut down the machine after a period of inactvity. This is useful during development but once you launch it much better to avoid the cold start time of starting up the machine when someone hits your url.

​
1. Navigate to your fly.toml

Update http_service to:

[http_service]
  internal_port = 3000
  force_https = true
  auto_stop_machines = false
  auto_start_machines = true
  min_machines_running = 0
  processes = ['app']

Fly will now keep your machine running until stopped. But will automatically restart the machine if it crashes for any reason.

​
Scale up RAM or CPU count

You can safely change the ram or cpu count to a higher number to handle more traffic. (e.g during a launch week)

​
1. Navigate to your fly.toml

Update vm to :

[[vm]]
  memory = '512mb'
  cpu_kind = 'shared'
  cpus = 2

​
Future improvements

LiteFs is a solution for Sqlite replication. I’m currently working on implementing this into the boring database repository so keep an eye out!