Keep a directory in step with a master copy.

Any number of clients can share a repo this way, and a separate machine can hold a replica of the master. Every run is a short-lived CLI process suitable for cron or the Windows Task Scheduler: there is no daemon, no database and no web interface.

   laptop  ──┐
   desktop ──┼──  push / pull  ──▶  master on the server
   tablet  ──┘                          │
                                        └──▶  replica (optional, pull-only)

Why it is built this way

• No daemon to supervise

  Every run is a short-lived process. Cron or the Task Scheduler starts it, it finishes, and nothing is left listening.

• Only changes move

  A size-and-mtime cache decides what to send. Data frames are deflated when that helps, and files of at least 64 KiB go as rsync-style block deltas when the other side holds an older copy.

• Conflicts are kept

  The newer mtime wins; the loser is preserved beside it as report.conflict-laptop-1-20261022T101500Z.txt, never silently discarded.

• Deletions are recoverable

  Removed files go to a per-repo trash, kept for 30 days by default. filesync restore puts a path back, or lists every version.

• Snapshots for replicas

  A replica can take a daily snapshot of its tree, by hard link or by copy, and roll back to a date. Fourteen are kept by default.

• Boring to operate

  JSON configuration and run logs you can read with cat, people-readable messages from a catalogue, and distinct exit codes for cron.

A repo in six commands

Two machines are involved: the server, which holds the master, and each client, which holds a copy. Install App::FileSync on both. The example keeps the master in the server account's home directory, so nothing runs as root.

# on the server, once
filesync-admin --root ~/filesync repo create documents

# on the server, once: filesync-server needs the same --root
cat > ~/bin/filesync-server-home <<'EOF'
#!/bin/sh
exec /usr/local/bin/filesync-server --root "$HOME/filesync" "$@"
EOF
chmod +x ~/bin/filesync-server-home

# on each client, once
filesync init --repo documents --root /home/foo/Documents \
        --server foo@srv --server-command /home/foo/bin/filesync-server-home

# on each client, whenever you like
filesync sync --repo documents --dry-run
filesync sync --repo documents

The client reports its host name as its id; there is nothing else to name. An id is not a login, and it does not divide the repo: every client of a repo syncs the whole tree.

Where to go next

• Quick start

  Server and client setup, the first sync, cron, Windows, and a backup-and-restore drill.

• Commands

  Every option of filesync, filesync-admin and filesync-server, and what each exit code means.

• Settings

  Process-wide settings, the client's repo list, the server's per-repo policy, and where each value came from.

• How it works

  The session, the line-delimited JSON protocol, deltas, conflicts, locking and the layout on disk.