Creating Fly S3

When to use this page

• You need object storage for uploads, media assets, exports, or backups.
• You deploy on Fly and want an S3-compatible storage backend.

Prerequisites

• Fly app and organization are set up.
• You know whether objects should be public or private.
• You have a naming pattern for buckets and object keys.

Documentation: https://fly.io/docs/tigris/

Overview

Fly.io uses Tigris for S3-compatible globally distributed object storage.

Usage considerations

• Use for file uploads, media assets, backups, and exports
• S3-compatible API - works with standard S3 SDKs and tools
• Globally distributed for low-latency access
• Prefer signed URLs for private access

Setup checklist

Follow the official Fly.io Tigris documentation for the latest CLI commands and options.

1. Choose a bucket name and region
2. Decide on public vs private access
3. Plan your folder structure and lifecycle rules
4. Create access keys for your app or CI/CD pipeline

After creation

• Store credentials in your app's environment variables (example):

  fly secrets set AWS_ACCESS_KEY_ID=<your-access-key-id>
  fly secrets set AWS_SECRET_ACCESS_KEY=<your-secret-access-key>
  fly secrets set BUCKET_NAME=<your-bucket-name>
  fly secrets set S3_ENDPOINT=<your-endpoint>
  fly secrets set S3_REGION=<your-region>

• In app you configure as regular S3

Verify

• Upload and download a test file from the running application.
• Signed URL flow works for private objects.
• Bucket/object naming and access policies match your security model.

Troubleshooting

• Upload permission errors: Check key permissions and bucket policy configuration.
• Broken object URLs: Verify endpoint/region and path-style vs virtual-host style settings.
• Slow delivery of public assets: Add CDN caching rules and optimize object size.