How to speed up a change column migration using Sequelize and MySQL - mysql

Question

I have several models with the fields: createdAt, updatedAt and deletedAt. These are set to the type DATE which results in a timestamp that is accurate to one second. I want these fields to be precise to a millisecond and the way to do that is to set their type to DATE(6) using Sequelize. This is the migration that I am using:
module.exports = {
up: (queryInterface, Sequelize) => {
return Promise.all([
queryInterface.changeColumn('transactions', 'createdAt', {
type: Sequelize.DATE(6),
allowNull: true,
}),
queryInterface.changeColumn('transactions', 'updatedAt', {
type: Sequelize.DATE(6),
allowNull: true,
}),
queryInterface.changeColumn('transactions', 'deletedAt', {
type: Sequelize.DATE(6),
allowNull: true,
}),
])
},
down: (queryInterface, Sequelize) => {
return Promise.all([
queryInterface.changeColumn('transactions', 'createdAt', {
type: Sequelize.DATE,
allowNull: true,
}),
queryInterface.changeColumn('transactions', 'updatedAt', {
type: Sequelize.DATE,
allowNull: true,
}),
queryInterface.changeColumn('transactions', 'deletedAt', {
type: Sequelize.DATE,
allowNull: true,
}),
])
}
};
The transactions table has 93 million rows and this migration ran for 10 hours before an internet problem caused it to time out. As such, is there any way to speed up it up? This is a MySQL database, the Sequelize version is 5.8.6 and the sequelize-cli version is 4.0.

Answer 1

This migration runs the ALTER TABLE command on the given database table. Under the hood, ALTER TABLE creates a copy of the original table, makes the given changes and copies over the original data. For a large database, this will take time. The problem is that the table gets locked whilst this operation is running so any updates on it will not go through. For this situation, there are two tools available (that I am aware of):
pt-online-schema-change
gh-ost
These tools also create a copy of the table but they don't lock up the table so we can continue updating it whilst the operation runs. They also record any updates to the original table during the operation and update the new table accordingly.