Quickstart

You need a ZFS pool running OpenZFS and a ZFSRecv.com API token. No other tools, libraries, or applications required, the send is plain zfs send piped over HTTPS.

Create an account

A pool on ZFSRecv is the receive target for your dataset trees. Create one from your account page — sign up first if you don’t have one yet — and pick a region: tank in the US Southwest region becomes us-sw-01.zfsrecv.com/v1/recv/tank. Your API token is shown once, right there when the pool is created — copy it immediately, and save it somewhere safe, as it isn’t displayed again.

Take a snapshot

Snapshot recursively so each child datasets gets the same snapshot as the parent. You can use whatever snapshot scheme or tools you are already using (sanoid, a cron script, or your own tooling). Nothing is pruned automatically on our end. Snapshots stay until you destroy them yourself or through whatever tools you already use to manage snapshots.

# zfs snapshot -r tank@2026-09-14T03:00

Send it

The first send is a full stream. Pipe it straight into the endpoint; curl -T - reads the stream from stdin and uploads it as it arrives, so nothing is staged on disk.

# zfs send -Rw tank@2026-09-14T03:00 | \
    curl -T - "https://us-sw-01.zfsrecv.com/v1/recv/tank?token=YOUR-TOKEN-HERE"

Send an increment

Every send after the first should be incremental: -i names the snapshot ZFSRecv.com already has, and only the changed blocks go over the wire.

# zfs send -Rw -i @2026-09-14T03:00 \
      tank@2026-09-15T03:00 | \
    curl -T - "https://us-sw-01.zfsrecv.com/v1/recv/tank?token=YOUR-TOKEN-HERE"

Next steps

• Setup automatic replication using cron, a systemd unit, or syncoid --recursive.
• Confirm the pool holds what you expect with GET /v1/list.