Deploy on Fly.io

Fly.io runs your bot on small VMs in your choice of region. About $3/month for a minimal config, more if you want multi-region redundancy.

Why Fly.io

Cheap — small VMs start around $3/month.
Multi-region capability if you want geographic redundancy.
Persistent volumes available.
HTTPS termination handled by Fly's edge.
Good for users who want closer control of their infra than Railway/Render allow.

Prerequisites

A Fly.io account (signup at fly.io).
Credit card on file (no free tier as of 2026).
The Fly CLI installed (flyctl).
Your caput bundle.

Install flyctl

On macOS:

brew install flyctl

On Linux:

curl -L https://fly.io/install.sh | sh

On Windows:

iwr https://fly.io/install.ps1 -useb | iex

fly auth login

Steps

1. Initialize the Fly app

In the directory where you extracted your bundle:

fly launch --no-deploy

This creates a fly.toml configuration file. You'll be asked:

App name: caput-bot (or your choice; becomes part of your URL).
Region: pick the closest to you.
Database: skip (we use SQLite locally).
Postgres: skip.
Deploy now: no.

Answer "no" to deploy now — we need to configure first.

2. Configure fly.toml

Edit the generated fly.toml. Key settings for caput:

app = "caput-bot"
primary_region = "iad"  # or your region

[build]
  # Phase 2+ will have a Dockerfile; Phase 0 placebuyer doesn't deploy

[env]
  XRPL_NETWORK = "testnet"
  ADMIN_PORT = "7777"
  STATE_DB_PATH = "/data/caput.db"

[http_service]
  internal_port = 8080
  force_https = true
  auto_stop_machines = false   # IMPORTANT: must be false for caput
  auto_start_machines = true
  min_machines_running = 1     # always 1 instance running

[[mounts]]
  source = "caput_data"
  destination = "/data"

The critical setting is auto_stop_machines = false and min_machines_running = 1. If these are wrong, Fly will stop your bot during idle periods, breaking the polling loop and exposing you to operational risk.

3. Set secrets

Sensitive variables (license info, wallet address) should be set as Fly secrets, not in fly.toml:

fly secrets set LICENSE_TX_HASH=<your-tx-hash>
fly secrets set LICENSE_WALLET=<your-wallet-address>
fly secrets set LICENSE_TIER=<tier-number>
fly secrets set LICENSE_MAX_USES=<uses-count>
fly secrets set LICENSE_MEMO=<session-memo>
fly secrets set WALLET_ADAPTER=<your-adapter>
fly secrets set DEPLOY_HOST=fly

(Pull these values from your bundle's config.env.)

4. Create the persistent volume

fly volumes create caput_data --region iad --size 1

This creates a 1GB volume in your region, costing $0.15/month.

5. Deploy

fly deploy

Fly builds the container, pushes to its registry, and starts the VM.

6. Get your public URL

fly status

The hostname will look like caput-bot.fly.dev. Set this as your PUBLIC_HOST:

fly secrets set PUBLIC_HOST=https://caput-bot.fly.dev

7. First-run setup

fly logs

Find the admin setup token in the startup logs.

Visit https://caput-bot.fly.dev/admin?setup=XXXX-XXXX-XXXX to complete setup.

8. Set up uptime monitoring

Same recommendation as other hosts. External uptime monitor hitting /health.

9. Multi-region (optional)

If you want geographic redundancy:

fly regions add lhr  # London
fly regions add nrt  # Tokyo
fly scale count 3

This runs three instances across three regions. State is on the volume, which is region-pinned, so multi-region is more about HTTP availability than state replication. For caput, single-region with strong uptime monitoring is usually fine.

Costs

Fly shared-cpu-1x VM (256MB): ~$2/month
1GB volume: $0.15/month
HTTPS / certificates: included
Bandwidth: usage-based, typically pennies/month for caput's traffic
Total: ~$3-4/month

For multi-region (3 instances), multiply VM cost by region count.

Updating

fly deploy

Pushes the new code, restarts the machine. State persists on the volume.

Support

Fly docs at fly.io/docs.

Caput-specific issues: read the source.