Database migration tool

This is used to keep track of the database structure, content etc, and update it when need be via deploys.

Supported databases:

• MariaDB (and MySQL)
• Elasticsearch

A table/index by default called db_version will be created, containing a single integer.

Scripts will be placed by default in process.cwd()/dbmigration/.js

Each migration script will be ran, and the db_version increased, until no more migration scripts exists.

Installation

npm i larvitdbmigration

Usage

Application startup script

In your application startup script, do something like this:

MariaDb and MySQL

'use strict';

const DbMigration = require('larvitdbmigration');
const Db = require('larvitdb');
const dbDriver = new Db({
    host: '127.0.0.1',
    user: 'foo',
    password: 'bar',
    database: 'baz'
});
const dbMigration = new DbMigration({
    dbType: 'mariadb',
    dbDriver,
    tableName: 'db_version', // Optional - used as index name for elasticsearch
    migrationScriptPath: './dbmigration', // Optional
    log // Optional, will use log.silly(), log.debug(), log.verbose(), log.info(), log.warn() and log.error() if given.
});

dbMigration.run().then(() => {
    // Now database is migrated and ready for use!
}).catch(err => {
    throw err;
});

Elasticsearch

'use strict';

const DbMigration = require('larvitdbmigration');
const dbMigration = new DbMigration({
    dbType: 'elasticsearch',
    url: 'http://127.0.0.1:9200',
    indexName: 'db_version', // Optional
    migrationScriptPath: './dbmigration', // Optional
    got // Optional, will use default got instance if not specified.
    log // Optional, will use log.silly(), log.debug(), log.verbose(), log.info(), log.warn() and log.error() if given.
});

dbMigration.run().then(() => {
    // Now database is migrated and ready for use!
}).catch(err => {
    throw err;
});

Retry behavior can be configured in the got instance that is passed to the constructor.

Example migration scripts

Lets say the current database have a table like this:

CREATE TABLE bloj (nisse int(11));

And in the next deploy we'd like to change the column name "nisse" to "hasse". For this you can do one of two methods:

MariaDB / MySQL, Javascript

Create the file process.cwd()/migrationScriptPath/1.js with this content:

'use strict';

// Always make the function async (or explicitly return a promise, see elasticsearch example below)
exports = module.exports = async function (options) {
    const {db} = options;

    await db.query('ALTER TABLE bloj CHANGE nisse hasse int(11);');
};

Elasticsearch

Create the file process.cwd()/migrationScriptPath/1.js with this content:

'use strict';

const got = require('got');

exports = module.exports = async function (options) {
    const {url, log} = options;

    log.info('Some script-specific logging');

    await got.put(`${url}/some_index/_mapping`, {
        json: {
            properties: {
                names: {
                    type: 'string',
                    position_increment_gap: 100
                }
            }
        }
    });
};

SQL

IMPORTANT! SQL files will be ignored if a .js file exists.

Create the file process.cwd()/migrationScriptPath/1.sql with this content:

ALTER TABLE bloj CHANGE nisse hasse int(11);

Summary

Tadaaa! Now this gets done once and the version will be bumped to 1. If you then create a script named "2.js" or "2.sql" you might guess what happends. :)

Changelog

6.0.0

• Removed locking mechanism for Elasticsearch migrations, there is no support for it in Elasticsearch.
• Rewrote library in TypeScript.
• Updated all dependencies to latest version (and replaced request with got).
• The Elasticsearch driver now only sends url and log instance to the running migration script (instead of the driver instance as it were before).