This is an emulator for the serverless driver
of PlanetScale and TiDB, which are serverless MySQL services.
By using this, you can proxy HTTP requests to requests for a MySQL server like Docker.
The goal of this project is to enable local development in the same way as connecting to PlanetScale or TiDB, without actually connecting to them.
There may be times when you don't want to connect to the PlanetScale from your local development environment for reasons such as:
- There is a risk of unintended billing (to reduce unnecessary spending)
- In large-scale development, issuing id and password to connect to the production service can be cumbersome
- Want to reduce development time and improve product velocity by developing and testing locally, like LocalStack in AWS or Firebase Local Emulator Suite in Firebase
It seems that there are challenges to developing in a local environment while connected to PlanetScale.
By using this service, requests to PlanetScale are proxied to a MySQL server in your local environment. This means that you can complete development in your local environment without having to connect to PlanetScale.
- Before: <your server> ++(HTTP)++ <PlanetScale>
- After: <your server> ++(HTTP)++ <Serverless MySQL HTTP(SMH)> ++(MySQL protocol)++ <MySQL server(ex. Docker)>
How to use with the @planetscale/database
SDK
Simply set the REST URL to the location where SMH is running. For example,
import { connect } from '@planetscale/database';
const config = { host: '<host>', username: '<user>', password: '<password>' };
if (import.meta.env.DEV) config.url = 'http://localhost:6306';
const conn = connect(config);
const results = await conn.execute('select 1 from dual where 1=?', [1]);
console.log(results);
If you only want to make this setting in your local environment, it's convenient to use import.meta.env.DEV
from Env Variables if you're using Vite. And you can probably implement it in the same way with process.env.NODE_ENV
or similar.
※The connection information to the MySQL server is set by SMH environment variables (see Configuration section), so host
, user
, and password
passed to connect(config)
are not used. In other words, when you are developing in a local environment, you can specify anything in host
, name
, and password
.
How to use with the @tidbcloud/serverless
First, with @tidbcloud/serverless
, requests to the domain have the http-
prefix, so set http-localhost
to resolve to 127.0.0.1
.
On Linux, you can modify /etc/hosts
as follows:
127.0.0.1 http-localhost
or
$ sudo echo "127.0.0.1 http-localhost" | sudo tee -a /etc/hosts
Then, just specify localhost
as the host to initialize as follows. This allows development in a local environment just like connecting to TiDB.
import { connect } from '@tidbcloud/serverless';
if (import.meta.env.DEV) {
// For using SMH,
// it is necessary to allow self-signed server certificates during HTTPS communication.
// https://github.com/orgs/nodejs/discussions/44038
const { Agent, setGlobalDispatcher } = await import('undici');
const agent = new Agent({
connect: {
rejectUnauthorized: false
}
});
setGlobalDispatcher(agent);
}
const conn = connect({
host: 'localhost',
username: '<username>',
password: '<password>',
database: '<database>'
});
const results = await conn.execute('select * from users where id=?', [1]);
console.log(results);
If you only want to make this setting in your local environment, it's convenient to use import.meta.env.DEV
from Env Variables if you're using Vite. And you can probably implement it in the same way with process.env.NODE_ENV
or similar.
※The connection information to the MySQL server is set by SMH environment variables (see Configuration section), so user
, and password
passed to connect(config)
are not used. In other words, when you are developing in a local environment, you can specify anything in name
, and password
.
Note
SMH uses a self-signed SSL certificate, so there will be errors in fetch
requests used internally by @tidbcloud/serverless
. Therefore, it is necessary to use the configuration with undici.
To start SMH, a MySQL server that can be connected from SMH must also be started at the same time. Do not forget to start the MySQL server.
Will the MySQL server exist separately or will it be started at the same time with docker compose
? The method differs depending on that.
If you have a MySQL server locally, you can start an SMH container that connects to it. In this example, SMH runs on port 6306.
$ docker run \
-d -it -p 6000:6000 -p 6443:6443 \
-e MYSQL_DATABASE=<your_database_name> \
-e MYSQL_HOST=<your_server_host_here> \
yutak23/serverless-mysql-http:latest
See Configuration for information on setting up a connection to the MySQL server.
Link to the Docker hub: https://hub.docker.com/r/yutak23/serverless-mysql-http
PlanetScale's timezone is fixed at UTC. The local MySQL server timezone must also be set to UTC.
If it is docker compose, it can be configured as follows.
version: '3.9'
services:
mysql:
image: mysql:8.0.32
container_name: mysql
environment:
MYSQL_ROOT_PASSWORD: ''
MYSQL_ALLOW_EMPTY_PASSWORD: 'yes'
TZ: 'UTC' # PlanetScale's timezone is fixed at UTC
ports:
- 3306:3306
serverless-mysql-http:
image: yutak23/serverless-mysql-http:latest
container_name: serverless-mysql-http
ports:
- 6000:6000
- 6443:6443
environment:
MYSQL_DATABASE: <your database name>
MYSQL_HOST: mysql # Using `mysql` hostname since they're in the same Docker network.
SMH allows you to change the connection settings to the MySQL server with the following environment variables.
Environment variable name | Default Value | Notes |
---|---|---|
MYSQL_HOST | 127.0.0.1 |
127.0.0.1 is the localhost in the container and is assumed to be overwritten when SMH is used. Be sure to specify the host of your own server. |
MYSQL_PORT | 3306 |
|
MYSQL_USER | root |
|
MYSQL_PASSWORD | '' |
empty string |
MYSQL_DATABASE | sample_db |
SMH can also be used in a CI environment such as GitHub Actions.
This is because SMH is provided as a Docker image, so it can be started as a service container for a job; just start a MySQL server and put SRH next to it.
SMH does not connect to MySQL until a request comes in, so you don't have to worry about a race condition where the MySQL container is not ready. If you are worried, just make sure to check the MySQL server startup as shown in the following example.
jobs:
test:
runs-on: ubuntu-latest
services:
mysql:
image: mysql:8.0
env:
MYSQL_ALLOW_EMPTY_PASSWORD: yes
TZ: UTC
ports:
- 3306:3306
options: --health-cmd="mysqladmin ping" --health-interval=10s --health-timeout=5s --health-retries=3
smh:
image: yutak23/serverless-mysql-http:latest
env:
MYSQL_DATABASE: sample_db
MYSQL_HOST: mysql
ports:
- 6000:6000
- 6443:6443
SMH has been tested to produce the same results as when connecting to PlanetScale. However, there are some differences in the behavior with PlanetScale in the following parts:
-
It does not support query caching by PlanetScale Boost
Even if you implement code like
await conn.execute('SET @@boost_cached_queries = true')
, it will not result in an error, but settingboost_cached_queries = true
will not cache the query results.
However, we do not think this is a critical issue as you would not be troubled by not being able to use query caching when developing in a local environment. -
There may be discrepancies in the
fields
dataThere may be discrepancies in the properties of the fields(Field) returned when connecting to PlanetScale and the properties of the fields returned by SMH. Specifically, differences like the following may occur.
PlanetScale SMH name: 'count(*)',
type: 'INT64',
columnLength: 21,
charset: 63,
flags: 32897name: 'count(*)',
type: 'INT64',
table: '',
orgTable: '',
database: 'sample_db',
orgName: 'count(*)',
columnLength: 21,
charset: 63However, we believe that there is few problem with local environment construction due to this reproducibility (compatibility)(If you wish, please raise the issue and we will consider modifications to improve reproducibility).