fly.toml

# Fly.io deployment for the emotion API.
#   fly launch --no-deploy      # create the app from this file
#   fly deploy                  # build remotely and roll out
#
# Runs in offline (stub) mode by default so a deploy works with zero external
# dependencies. For the real model, set OFFLINE=0 and build from
# requirements-ml.txt (expect a larger image and more memory).

app = "distilbert-emotion-api"
primary_region = "iad"

[build]
  dockerfile = "Dockerfile"

[env]
  OFFLINE = "1"
  PORT = "8000"
  LOG_LEVEL = "INFO"
  MAX_BATCH_SIZE = "64"
  BATCH_MAX_DELAY_MS = "5"

[http_service]
  internal_port = 8000
  force_https = true
  auto_stop_machines = "stop"
  auto_start_machines = true
  min_machines_running = 0
  processes = ["app"]

  [http_service.concurrency]
    type = "requests"
    soft_limit = 40
    hard_limit = 60

  [[http_service.checks]]
    method = "GET"
    path = "/healthz"
    interval = "15s"
    timeout = "3s"
    grace_period = "10s"

[[vm]]
  size = "shared-cpu-1x"
  memory = "512mb"   # bump to >=2gb when OFFLINE=0 (torch + weights)