Migrate a server to S3 without babysitting it

A server-to-cloud migration is usually a script somebody wrote at two in the morning: rsync or the AWS CLI in a loop, no record of what was copied, and a restart from zero when the connection dropped at 80 percent. It works once, for the person who wrote it.

> Measured from the product, September 2026: 240M+ objects indexed, 50 TB under management, across Amazon S3, Cloudflare R2, Google Cloud Storage and customers' own servers.

• •Connect the server over SSH — Linux, or Windows with OpenSSH — with a key BucketPilot generates. Nothing is installed on the machine.
• •Walk the tree, then copy. The scan uses the server's own tools, so a tree of millions of files is listed without loading it into memory; a re-run lists only what changed since the last one.
• •Resumable. Files over 64 MB go in pieces the server assembles; a dropped connection continues from the last piece, never from zero. Large files are the point of a migration, and they are the ones that failed before.
• •A log for every file, live progress, and per-file retries. Nothing silent.
• •Re-run any time. The same job, run again, copies only what is new — so the cut-over is a last short run, not a weekend.

1. 1.Add the server under Buckets & Servers → Add server: host, port, user, and the key BucketPilot shows you to authorise. The connection is tested before it is saved, and a Windows host is told exactly what to enable if OpenSSH is missing.
2. 2.Open Backups & Migrations → Create, choose the server as source and a bucket as destination, and a path on each side.
3. 3.Run it. Watch the log or come back later; run it again before you switch off the server.

Migrations are on the Pro plan and above; a server as a *destination* is Business.