Safe, resumable server-to-server file transfers via rsync over SSH — with a web UI, job queue, and dry-run safety gates.
🔒 Private source — this repository is a project showcase.
Copying large datasets between servers usually means routing all the data through the machine running the command — slow, wasteful, and brittle on flaky connections.
S2SC fixes this: the controller SSHes into the source host and runs rsync there, so data flows directly from source to destination without touching the controller machine.
The other problem is safety. Bulk copies are hard to undo. S2SC enforces a dry-run gate before every transfer, requires explicit approval to start, and never deletes or overwrites by default.
Controller (Linux)
│
├─── SSH ──► Source host
│ │
│ └─── rsync over SSH ──► Destination host
│
└─── Web UI ◄── Browser (any machine on the network)
- 📝 Create a job — pick source and destination endpoints, set paths
- 🔍 Dry-run prep — controller scans files, counts size, detects conflicts
- 👀 Review — inspect totals and any existing-file conflicts in the UI
- ✅ Approve & start — one click triggers the actual rsync
- 📊 Monitor — live progress, transfer rate, and ETA
- ✔️ Optional verify — post-copy checksum dry-run confirms integrity
- Dry-run gate — every job runs a prep scan before copy; copy never starts automatically
- No deletes, no overwrites —
--ignore-existingby default; destination files are never touched - Destination must exist — no accidental directory creation
- Allowlist-only paths — each endpoint declares its allowed base directories; paths outside are rejected
- Conflict policy — choose Ask / Skip / Stop for existing destination files
- Verify-only mode — checksum dry-run against an already-copied destination; reports mismatches without writing anything
- Sequential execution — one active transfer at a time
- Reorderable queue for waiting jobs
- Cancel, pause, and restart support
- Job states:
queued → preparing → ready → copying → finished / failed / canceled
- Queue overview with live status, progress bar, rate, and ETA
- Job detail view with per-job rsync log and file list
- New job form — endpoint picker, path entry, policy and flag configuration
- Admin UI for token management, endpoint management, and password rotation
| Option | Default |
|---|---|
| On existing files | Ask |
| Checksum verify | Off |
| Preserve permissions + timestamps | On |
| Write-in-place | Off |
| Auto-start after prep | Off |
| Verify after copy | Off |
| Bandwidth limit | None |
| Layer | Technology |
|---|---|
| Backend | Python 3.10+ / FastAPI |
| Web server | Uvicorn (ASGI) |
| Templating | Jinja2 |
| Database | SQLite (job state, logs) |
| Transfer engine | rsync over SSH via Paramiko |
| Auth | Token-based (queue UI) + session-based (admin UI) |
| Password storage | PBKDF2-SHA256 |
| Frontend | Vanilla JS + CSS |
app/
main.py FastAPI routes — queue UI, job API, admin UI
config.py Config loader + endpoint validation
transfer.py rsync argument builder + SSH execution (Paramiko)
queue.py Sequential job queue, state machine
state.py In-memory job state + SQLite persistence
storage.py SQLite schema, reads/writes
ssh.py SSH connection pool + remote command runner
security.py Token auth, password hashing, session management
validators.py Path allowlist enforcement
logging_setup.py Per-job + global log routing
templates/ Jinja2 HTML templates
static/ app.js + app.css (no build step required)
tests/ pytest suite
No agent on endpoints — the controller SSHes in and runs a single rsync command. Source and destination only need rsync + SSH.
Config-driven endpoints — servers are declared in config.json with host, user, port, OS type, and allowed base directories.
Queue — live status, filter, and reorder
New Job — endpoint picker with transfer options
Admin — default options and queue maintenance
Active development. 🔒 Private source — this repository is a project showcase.