# Your first restore

> A 20-minute practice run that never touches your real data.

Source: https://rowsafe.sh/docs/first-restore

Practise a restore now, not during an outage. You make a practice database, delete rows on purpose, and get them back with **Rewind**. About 20 minutes.

## Before you start

- The database's **first backup has finished** ([Quickstart](https://rowsafe.sh/docs/quickstart)).
- You are an **owner or admin**. Members can't restore.
- Free disk for a copy: about **1.3 × the database size + 1 GB**. Rowsafe tells you if there isn't enough.

### Make a practice database

Open the server, then **Databases & users** → **Create database**. Name it `practice`. You don't need the password it shows. [Create databases and users](https://rowsafe.sh/docs/guides/databases-and-users).

### Add a table with 100 rows

The one SQL step, because it's your data. On the server:

```sh
sudo -u postgres psql -d practice
```

```sql
CREATE TABLE customers (id int PRIMARY KEY, name text);
INSERT INTO customers SELECT g, 'Customer ' || g FROM generate_series(1, 100) AS g;
```

In Docker: `docker compose exec postgres psql -U postgres -d practice` (your own service name and user).

Rowsafe needs a primary key to compare and bring back rows.

### Wait two minutes, then delete 40 rows

Waiting makes the "before" moment easy to spot.

```sql
DELETE FROM customers WHERE id <= 40;
SELECT count(*) FROM customers;   -- 60
```

### Find the moment

Open the database, then **Rewind** → **Find the moment**. Pick **deleted rows** and the last hour. You see your delete: 40 rows from `customers`, to the second. Click **Rewind to just before this**. [Find the moment](https://rowsafe.sh/docs/guides/find-the-moment).

### Restore a copy

Click **Restore a copy**. Production keeps running. It takes about as long as a [Proof](https://rowsafe.sh/docs/concepts/restore-drills) run.

The copy has no network port, so no app connects to it by mistake. It **deletes itself** after 24 hours by default. [Restore a copy](https://rowsafe.sh/docs/guides/restore#restore-a-copy).

### Compare with production

On the copy's card, click **Compare with production** and pick `customers`. You should see **Missing in production: 40**. [Compare with production](https://rowsafe.sh/docs/guides/restore#compare-with-production).

### Bring the rows back

Select the table, click **Bring back 40 missing rows** and type the database name. Rowsafe sets a Mark first, then puts back only the missing rows, in one transaction. [Bring back rows](https://rowsafe.sh/docs/guides/restore#bring-back-rows).

```sql
SELECT count(*) FROM customers;   -- 100 again
```

### Clean up

- **The copy:** **Delete** on its card, or let it expire.
- **The practice database:** in **Databases & users**, its menu → **Remove database**, then type its name.

Note how long the copy took: a real one on this database takes about as long.

## Other ways to practise

- **With a Mark.** Set one before you delete (on the database's page, or `rowsafe mark app before-practice`). Then on **Marks**, click **Restore a copy at this Mark**. [Marks](https://rowsafe.sh/docs/concepts/restore-points).
- **On another server.** [Fork](https://rowsafe.sh/docs/guides/fork) the database to another of your servers. That's close to what [Restore to another server](https://rowsafe.sh/docs/guides/restore#server-lost-restore-to-another-server) does when a server is lost.

> **Rewinding the whole database is not a practice step:** [Rewind the whole database](https://rowsafe.sh/docs/guides/restore#rewind-the-whole-database) replaces your live database and stops PostgreSQL while it works. Never practise it on production.
