flecks/website/docs/database.mdx

---
title: Database
description: Define models and connect to a database.
---

import Create from '@site/helpers/create';
import InstallPackage from '@site/helpers/install-package';

# Database

flecks provides database connection through [Sequelize](https://sequelize.org/) and database
server instances through either flat SQLite databases or [Docker](https://www.docker.com/)-ized
database servers.

## Install and configure

We'll start from scratch as an example. Create a new flecks application:

<Create pkg="db_test" type="app" />

Now in your new application directory, add `@flecks/db`:

```bash
npx flecks add @flecks/db
```

Finally, `npm start` your application and you will see lines like the following in the logs:

```
  @flecks/db/server/connection config: { dialect: 'sqlite', storage: ':memory:' } +0ms
  @flecks/db/server/connection synchronizing... +107ms
  @flecks/db/server/connection synchronized +2ms
```

By default, flecks will connect to an in-memory SQLite database to get you started instantly.

## Your models

Astute observers may have noticed a line preceding the ones earlier:

```
  @flecks/core/flecks gathered '@flecks/db/server.models': [] +0ms
```

Let's create a fleck that makes a model so we can get a feel for how it works.

First, create a fleck in your application:

<Create pkg="tags" type="fleck" />

Now, let's hop into `packages/tags/src/index.js` and add a hook implementation:

```javascript title="packages/tags/src/index.js"
export const hooks = {
  '@flecks/db/server.models': (flecks) => {
    const {Model, Types} = flecks.fleck('@flecks/db/server');
    class Tags extends Model {
      static get attributes() {
        return {
          key: {
            type: Types.STRING,
            allowNull: false,
          },
          value: {
            type: Types.STRING,
            allowNull: false,
          },
        };
      }
    };
    return {
      Tags,
    };
  },
}
```

:::tip

`@flecks/db` uses [Sequelize](https://sequelize.org/) under the hood. You can dive into
[their documentation](https://sequelize.org/docs/v6/getting-started/) to learn even more.

The static `attributes` method above is sugar on top of
[`sequelize.define`](https://sequelize.org/docs/v6/core-concepts/model-basics/#using-sequelizedefine)
and you should consult that documentation for more details on how to define and validate your
models.

To implement associations between models, there is a static `associate` method. Suppose you also
had a `Post` model along with your `Tags` model. You might do something like this in your `Tags`
model:

```
static associate({Post}) {
  Post.hasMany(this);
  this.hasMany(Post);
}
```

:::

Now, `npm start` your application and you will see that line looks different:

```
  @flecks/core/flecks gathered '@flecks/db/server.models': [ 'Tags' ] +0ms
```

Our model is recognized!

## Gathering models

When building Real:tm: applications we are usually going to need a bunch of models. If we add all
of them into that one single file, things are going to start getting unwieldy. Let's create a
`src/models` directory in our `packages/tags` fleck and add a `tags.js` source file with the
following code:

```javascript title="packages/tags/src/models/tags.js"
export default (flecks) => {
  const {Model, Types} = flecks.fleck('@flecks/db/server');
  return class Tags extends Model {
    static get attributes() {
      return {
        key: {
          type: Types.STRING,
          allowNull: false,
        },
        value: {
          type: Types.STRING,
          allowNull: false,
        },
      };
    }
  };
}
```

Notice that this looks very similar to how we defined the model above, but this time we're only
returning the class.

Now, hop over to `packages/tags/src/index.js` and let's rewrite the hook implementation:

```javascript title="packages/tags/src/index.js"
export const hooks = {
  '@flecks/db/server.models': Flecks.provide(require.context('./models')),
}
```

We're passing the path to our models directory to `require.context` which is then passed to
`Flecks.provide`. This is completely equivalent to our original code, but now we can add more
models by adding individual files in `packages/tags/src/models` and keep things tidy.

:::info

For a more detailed treatment of gathering in flecks, see [the gathering guide](#todo).

:::

## Working with models

Let's do something with it. Edit `packages/tags/src/index.js` again like
so:

```javascript title="packages/tags/src/index.js"
export const hooks = {
  // highlight-start
  '@flecks/server.up': async (flecks) => {
    const {Tags} = flecks.get('$flecks/db.models');
    console.log('There were', await Tags.count(), 'tags.');
  },
  // highlight-end
  '@flecks/db/server.models': Flecks.provide(require.context('./models')),
}
```

Now, another `npm start` will greet us with this line in the output:

```
There were 0 tags.
```

Not very interesting. Let's add some, but only if there aren't any tags yet:

```javascript title="packages/tags/src/index.js"
export const hooks = {
  '@flecks/server.up': async (flecks) => {
    const {Tags} = flecks.get('$flecks/db.models');
    console.log('There were', await Tags.count(), 'tags.');
    // highlight-start
    if (0 === await Tags.count()) {
      await Tags.create({key: 'foo', value: 'bar'});
      await Tags.create({key: 'another', value: 'thing'});
    }
    console.log('There are', await Tags.count(), 'tags.');
    // highlight-end
  },
  '@flecks/db/server.models': Flecks.provide(require.context('./models')),
}
```

Another `npm start` and we see the tags created!

## Persistence

You'll notice that if you run it again, it will always say

```
There were 0 tags.
There are 2 tags.
```

What's up with that? Remember in the beginning:

> By default, flecks will connect to an in-memory SQLite database to get you started instantly.

This means that the database will only persist as long as the life of your application. When you
restart it, you'll get a fresh new database every time. Obviously, this isn't very helpful for
any real purpose. Let's make a change to our `build/flecks.yml`:

```yml title="build/flecks.yml"
'@db_test/tags:./packages/tags': {}
'@flecks/core': {}
'@flecks/db': {}
// highlight-start
'@flecks/db/server':
  database: './persistent.sql'
// highlight-end
'@flecks/server': {}
```

Now `npm start` again. You'll see our old familiar message:

```
There were 0 tags.
There are 2 tags.
```

This time though, our application wrote the SQLite database to disk at `./persistent.sql`. If we
give it one more go, we'll finally see what we expect:

```
There were 2 tags.
There are 2 tags.
```

A persistent database!

## Containerization

Sure, our database is persistent... kinda. That `persistent.sql` file is a bit of a kludge and
isn't much of a long-term (or production) solution. Let's remove it:

```
rm persistent.sql
```

Our small-core philosophy means that you don't pay for spinning up a database by default. However,
it's trivial to accomplish a *"real"* database connection if you have Docker installed on your
machine.

<details>
  <summary>How do I know if I have Docker running on my machine?</summary>

  A decent way to test if your machine is ready to continue with this guide is to run the following
  command:

  `docker run -e POSTGRES_PASSWORD=password postgres`

  if the command appears to spin up a database, you're in good shape!

  If not, follow the [Docker installation documentation](https://docs.docker.com/engine/install/)
  before proceeding.
</details>

Let's add another fleck to our project:

<InstallPackage pkg="@flecks/docker" />

Configure `build/flecks.yml`:

```yml title="build/flecks.yml"
'@db_test/tags:./packages/tags': {}
'@flecks/core': {}
'@flecks/db': {}
// highlight-start
'@flecks/db/server':
  database: db
  dialect: postgres
  password: THIS_PASSWORD_IS_UNSAFE
  username: postgres
'@flecks/docker': {}
'@flecks/server':
  up:
    - '@flecks/docker'
    - '@flecks/db'
    - '@db_test/tags'
// highlight-end
```

Notice that we configured `@flecks/server.up` to make sure to enforce a specific order in which
our server flecks come up: first `@flecks/docker` to spin up the database, then
`@flecks/db` to connect to the database, and finally our `@db_test/tags` fleck to interact with
the database. This is important!

Now `npm start` will reveal the following message in the logs:

```
  @flecks/server/entry Error: Please install pg package manually
```

Pretty straightforward how to proceed:

<InstallPackage pkg="pg" />

Remember, **small core**! :smile: Now `npm start` again and you will see some new lines in the
logs:

```
  @flecks/docker/container creating datadir '/tmp/flecks/flecks/docker/sequelize' +0ms
  @flecks/docker/container launching: docker run --name flecks_sequelize -d --rm -p 5432:5432 -e POSTGRES_USER=postgres -e POSTGRES_DB=db -e POSTGRES_PASSWORD=THIS_PASSWORD_IS_UNSAFE -v /tmp/flecks/flecks/docker/sequelize:/var/lib/postgresql/data postgres +0ms
  @flecks/docker/container 'sequelize' started +372ms
  @flecks/db/server/connection config: { database: 'db', dialect: 'postgres', host: undefined, password: '*** REDACTED ***', port: undefined, username: 'postgres' } +0ms
  @flecks/db/server/connection synchronizing... +2s
  @flecks/db/server/connection synchronized +3ms
```

and of course, we see:

```
There were 0 tags.
There are 2 tags.
```

because we just created a new postgres database from scratch just then! Kill the application and
run `npm start` one more time and then you will see what you expect:

```
There were 2 tags.
There are 2 tags.
```

Awesome, we have a connection to a real postgres database!

<details>
  <summary>Where is the docker container?</summary>

  You may notice that the first time you start your application there is a delay while Docker spins
  up your new database server. This is normal. You may have also noticed that subsequent runs speed
  up to near-instantaneous. This is also normal!

  `@flecks/docker` runs your container in a manner that outlives your application. If you kill your
  application and then run:

  ```
  docker ps
  ```

  You will see a line for your postgres database looking something like:

  ```
  CONTAINER ID   IMAGE     COMMAND                 CREATED        STATUS        PORTS                                      NAMES
  <SOME_ID>      postgres  "docker-entrypoint.s…"  5 minutes ago  Up 5 minutes  0.0.0.0:5432->5432/tcp, :::5432->5432/tcp  flecks_sequelize
  ```

  You'll see under the `NAMES` column heading, there is an entry called `flecks_sequelize`. That's
  our database! You can always

  ```
  docker kill flecks_sequelize
  ```

  to free up any resources being used. flecks keeps the containers running so that you get a nice
  fast application start.

  :::note

  The container name is based off of `@flecks/core.id` which by default is `flecks`. If you change
  your application's ID, the container name will be different.

  :::

</details>

## Production

Sure, spinning up a database like magic is spiffy for development, but you probably want to be a
little less freewheeling on your production server.

Build the application we've built so far:

```
npm run build
```

Then, take a look in the `dist` directory. You'll see a file there called `docker-compose.yml`.
`@flecks/docker` automatically emits this file when you build your application for production to
make container orchestration easier. Let's take a look:

```yml title="dist/docker-compose.yml"
version: '3'
services:
  flecks_app:
    build:
      context: ..
      dockerfile: dist/Dockerfile
    environment:
      FLECKS_ENV_FLECKS_DOCKER_SERVER_enabled: 'false'
      // highlight-next-line
      FLECKS_ENV_FLECKS_DB_SERVER_host: sequelize
    volumes:
      - ../node_modules:/var/www/node_modules
  // highlight-start
  sequelize:
    image: postgres
    environment:
      POSTGRES_USER: postgres
      POSTGRES_DB: db
      POSTGRES_PASSWORD: THIS_PASSWORD_IS_UNSAFE
  // highlight-end
```

Notice our database container is included and already prepopulated with the configuration we
specified!

You can run (after you [install Docker Compose](https://docs.docker.com/compose/install/) if
necessary):

```
docker-compose -f dist/docker-compose.yml up
```

This demonstrates that your application is now being orchestrated by Docker Compose and is
chugging right along!