Skip to content

README.md

Latest commit

 

History

History
145 lines (103 loc) · 3.35 KB

README.md

File metadata and controls

145 lines (103 loc) · 3.35 KB

1100 — Volumes & Persistent Storage

The Problem with Ephemeral Machines

By default, Fly Machines are stateless — when a Machine is replaced (on deploy), all data written to the container's filesystem is lost. For persistent data, you need a Fly Volume.

Fly Volumes

A Fly Volume is a persistent block storage device attached to a single Machine in a specific region. Think of it as a hard disk for your Machine.

  • Built on NVMe SSDs (fast, low-latency)
  • Tied to one region
  • One Volume per Machine (1:1 relationship)
  • Survives Machine restarts and deploys
  • Does NOT survive fly volumes destroy

Creating Volumes

fly volumes create myapp_data \
  --region ams \
  --size 10          # 10 GB

With replication (across 2 zones)

fly volumes create myapp_data \
  --region ams \
  --size 10 \
  --count 2          # Creates 2 volumes in different availability zones

Mounting in fly.toml

[[mounts]]
  source = "myapp_data"     # Volume name (not ID)
  destination = "/data"     # Mount path inside container

After mounting, your app can read/write /data like a normal filesystem.

Volume Management

fly volumes list                        # List all volumes
fly volumes show <volume-id>            # Show volume details
fly volumes extend <volume-id> --size 20  # Grow to 20GB (can't shrink)
fly volumes destroy <volume-id>         # Delete permanently

Snapshots (Backups)

fly volumes snapshots list <volume-id>      # List snapshots
fly volumes snapshots create <volume-id>    # Manual snapshot

Fly takes automatic daily snapshots (retained for 5 days).

Restore from snapshot

fly volumes create restored_data \
  --from-snapshot <snapshot-id> \
  --region ams \
  --size 10

Multi-Region Volume Strategy

A Volume exists in one region only. For a 3-region app:

fly volumes create myapp_data --region ams --size 10
fly volumes create myapp_data --region lhr --size 10
fly volumes create myapp_data --region syd --size 10

The Machine in each region gets its own Volume. Data is NOT replicated between volumes automatically — use LiteFS or Postgres for replicated state.

Common Use Cases

Use case Solution
SQLite database Volume at /data/app.db
User uploaded files Volume at /data/uploads
Redis data (self-hosted) Volume at /data/redis
Logs Volume at /data/logs
App config files Volume at /data/config

Example: SQLite App

fly.toml:

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

[env]
  DATABASE_PATH = "/data/app.db"

App code (Node.js):

const Database = require('better-sqlite3');
const db = new Database(process.env.DATABASE_PATH);

Tasks

Task 1 — Create and use a Volume

  1. fly volumes create test_data --region ams --size 1 -a my-app
  2. Add [[mounts]] to fly.toml.
  3. Deploy.
  4. SSH in: fly ssh console
  5. Run: echo "hello persistent world" > /data/test.txt
  6. Exit, redeploy, SSH back in, run: cat /data/test.txt — data should persist.

Task 2 — Snapshot a Volume

  1. Run fly volumes snapshots create <vol-id>.
  2. Run fly volumes snapshots list <vol-id> to see it.

Next Steps

→ Continue to 1200 — Private Networking (WireGuard / 6PN)