Custom Prometheus Metrics from multi-process Rails app

Hi! I’m trying to add custom Prometheus metrics to my Rails app, but I can’t quite get my configuration right. I’m trying to use the prometheus_exporter gem (GitHub - discourse/prometheus_exporter: A framework for collecting and aggregating prometheus metrics). I’ve got the setup working in development, but I can’t quite get the pieces together in production.

Here’s my current fly.toml:

app = "my-app"
primary_region = "iad"
console_command = "/rails/bin/rails console"

app = "bin/rails server"
worker = "bundle exec good_job start"
metrics = "bundle exec prometheus_exporter -b"

port = 9394
path = "/metrics" # default for most prometheus exporters
processes = ["app", "worker"]


  internal_port = 3000
  force_https = true
  auto_start_machines = true
  min_machines_running = 1
  processes = ["app"]

  guest_path = "/rails/public"
  url_prefix = "/"

I tried to add a new process for the metrics exporter, and it seems to have properly started a new machine, but the problem is, my app process can’t find the metrics exporter. The logs have this:

app[...] ERROR -- : Prometheus Exporter, failed to send message Connection refused - connect(2) for "" port 9394 

So, I think it’s trying to find the right IP and port, but for some reason, it’s not finding anything. Can anyone suggest a way I might update my configuration to make this work?