mysql-migrations
Table of contents
Prerequisites
A node project with mysql used for database.
Install
It can be installed using npm.
npm install mysql-migrations
Setup
- Create a directory where you wish to maintain all your migrations. We call it
migrations
. - Instantiate mysql-migrations by passing a mysql pool and the migrations directory path.
# migration.js
var mysql = require('mysql');
var migration = require('mysql-migrations');
var connection = mysql.createPool({
connectionLimit : 10,
host : 'localhost',
user : 'root',
password : 'password',
database : 'your_database'
});
migration.init(connection, __dirname + '/migrations');
Advanced Setup
If you want to execute something at the end of any migration, you can add third parameter as callback function. Example:
# migration.js
var mysql = require('mysql');
var migration = require('mysql-migrations');
var connection = mysql.createPool({
connectionLimit : 10,
host : 'localhost',
user : 'root',
password : 'password',
database : 'your_database'
});
migration.init(connection, __dirname + '/migrations', function() {
console.log("finished running migrations");
});
Adding Migrations
Initiate a migration
Run
node migration.js add migration create_table_users
Now open the migrations folder. Locate the newest file with greatest timestamp as it predecessor. The file will have the name which was specified in the command such as 12213545345_create_table_users.js
Add migrations
Write the query in up
key of the json created for the forward migration. As a part of good practice, also write the script to rollback the migration in down
key. Ex.
module.exports = {
"up": "CREATE TABLE users (user_id INT NOT NULL, UNIQUE KEY user_id (user_id), name TEXT )",
"down": "DROP TABLE users"
}
Add seed
Run
node migration.js add seed create_table_users
to add a seed.
module.exports = {
"up": "UPDATE users SET name = 'John Snow' WHERE name = ''",
"down": "UPDATE users SET name = '' WHERE name = 'John Snow'"
}
Initate and Add migration in single command
Run
node migration.js add migration create_table_users "CREATE TABLE mysql_migrations_347ertt3e (user_id INT NOT NULL, UNIQUE KEY user_id (user_id) )"
Locate the newest file with greatest timestamp as it predecessor and open it. Query will be automatically added as up
key. However down
key needs to be filled manually.
Custom migrations
You may initiate the migration file and add a function.
module.exports = {
'up' : function (conn, cb) {
conn.query ("UPDATE users set name = 'alen'", function (err, res) {
cb();
});
},
'down' : ""
}
Executing Migrations
There are few ways to run migrations.
- Run
node migration.js up
. Runs all the pending up
migrations. - Run
node migration.js up 2
. Runs 2 pending up
migrations from the last position. - Run
node migration.js down
. Runs only 1 down
migrations. - Run
node migration.js refresh
. Runs all down migrations followed by all up.
Example Output:
UP: "CREATE TABLE users2 (user_id INT NOT NULL, UNIQUE KEY user_id (user_id), name TEXT )"
UP: "CREATE TABLE users (user_id INT NOT NULL, UNIQUE KEY user_id (user_id), name TEXT )"
UP: "CREATE TABLE users1 (user_id INT NOT NULL, UNIQUE KEY user_id (user_id), name TEXT )"
No more "UP" migrations to run
Execute anonymous migrations
At times, few migrations need to run again or anonymously. There could be variety of reasons old migrations need to be executed or rollbacked. It can be done this way.
Up migration
node migration.js run 1500891087394_create_table_users.js up
Down migration
node migration.js run 1500891087394_create_table_users.js down
Since these are anonymous executions, no records are maintained for any executions.
Executing backdated migrations
Suppose there are few migrations which were merged late into the main branch. Technically, they should not run because they are old migrations and should already have been run, but it happens that someone has been working on a branch for long and once merged into master, the older migrations do not work because the latest migration timestamp is greater. There is a flag which can help in running those migrations. There are two ways:
- Single Run
node migration.js up --migrate-all
- For all runs, you can configure in the init part itself.
# migration.js
var mysql = require('mysql');
var migration = require('mysql-migrations');
var connection = mysql.createPool({
connectionLimit : 10,
host : 'localhost',
user : 'root',
password : 'password',
database : 'your_database'
});
migration.init(connection, __dirname + '/migrations', function() {}, ["--migrate-all"]);
and then run migrations normally
node migration.js up
Saving Schema and Loading from Schema
Having schema.sql is good because it will help you review the schema changes in PR itself. The schema is stored in migrations folder under the file name schema.sql
. There are two ways to generate schema.
- Single Run
node migration.js up --update-schema
- For all runs, you can configure in the init part itself.
# migration.js
var mysql = require('mysql');
var migration = require('mysql-migrations');
var connection = mysql.createPool({
connectionLimit : 10,
host : 'localhost',
user : 'root',
password : 'password',
database : 'your_database'
});
migration.init(connection, __dirname + '/migrations', function() {}, ["--update-schema"]);
and then run migrations normally
node migration.js up
node migration.js down 2
..
Updated Schema will be stored in migrations folder after each run.
Loading Directly from Schema
Suppose you setup your project and you want to avoid running the entire migrations and simply want to load it from schema generated in the above process. You can do it via:
node migration.js load-from-schema
The above command will create all the tables and make entry in the logs table. It is helpful when setting up projects for newbies or environments without running entire migration process.
Pending
Test cases: Will pick up when I get time.
Help and Support
Will be more than happy to improve upon this version. This is an over night build and needs to be improved certainly. Will welcome everyone who wants to contribute.
Credits and other stuff
It is my first contribution to npm and I am sort of happy over it. I made this when I was really looking for a suitable tool with barebone settings allowing me to maintain database structure. I could not find a basic one and hence wrote my own and finally decided to publish. It took me around 2 hours to write the first version which barely works. But it still does my job.
Credits to my parents.