fly.machines.toml

app = "udns"

kill_signal = "SIGINT"
kill_timeout = 10

[build]
  dockerfile = "node.Dockerfile"

[env]
NODE_ENV = "production"
CLOUD_PLATFORM = "fly"
# (set for machines only) node proc shuts itself after this threshold:
# on Fly, the in-memory DNS cache is evicted on process exit, hence
# more efficient to let the Machine linger for as much time affordable.
# also: fly can't start vms that shutdown quick: community.fly.io/t/6091/12
MACHINES_TIMEOUT_SEC = "180"
LOG_LEVEL = "info"

[experimental]
auto_rollback = true

# DNS over HTTPS
[[services]]
internal_port = 8080
protocol = "tcp"

  [services.concurrency]
  hard_limit = 575
  soft_limit = 500
  type = "connections"

  [[services.ports]]
  port = 443

  [[services.ports]]
  port = 8080

  [[services.tcp_checks]]
  grace_period = "10s"
  interval = "15s"
  restart_limit = 3
  timeout = "3s"

# DNS over TCP/TLS
[[services]]
internal_port = 10000
protocol = "tcp"

  [services.concurrency]
  hard_limit = 575
  soft_limit = 500
  type = "connections"

  [[services.ports]]
  port = 853

  [[services.ports]]
  port = 10000

  [[services.tcp_checks]]
  grace_period = "10s"
  interval = "15s"
  restart_limit = 3
  timeout = "3s"

# community.fly.io/t/5490/3
[checks]
  [checks.up]
    grace_period = "30s"
    interval = "15s"
    method = "get"
    path = "/check"
    port = 8888
    timeout = "10s"
    type = "http"