Working across teams, we always need to keep the database in sync with code changes, with migrations across branches and more.

To sync up, or when setting up a new database, we run the following commands:

# Create Database
rails db:create 
# Populate Seed Data
rails db:seed  
# Sync up migrations on current branch
rails db:migrate

Rails 6 adds rake db:prepare to take care of the above for us, and prepare the database for use of the required database.

Problems with just migrate

Running rails db:migrate when the database doesnt exist, throws ActiveRecord::NoDatabaseError.

➜  blog git:(master) ✗ rails db:migrate
rails aborted!
ActiveRecord::NoDatabaseError: FATAL:  database "blog_development" does not exist

The task db:prepare internally catches ActiveRecord::NoDatabaseError and then runs rails db:setup.

In other words, if the database is correctly configured, db:prepare will just run the migration, if the database already exists. If the database does not exist, db:prepare ensures that it is created, migrated and seeded with data.

Behaviour depending on presence of database

When database is present:

➜  blog git:(master) ✗ rails db:prepare
== 20190513112934 CreateArticles: migrating ===================================
-- create_table(:articles)
   -> 0.0103s
== 20190513112934 CreateArticles: migrated (0.0104s) ==========================

This executed migrations on the existing database.

When a database is not present:

➜  blog git:(master) ✗ rails db:prepare
Created database 'blog_development'

This created the database and populated seed data.

The important bit is that db:prepare always runs in an idempotent way. This is useful when being utilized in CI scripts, staging deployments, and more.