A basic ruby tool for managing database migrations.
Two different types of migration are supported:
- Simple SQL files
- Any type of executable file
SQL migrations (any file with a .sql extension) simply get read and their contents executed against the database as a single statement. Any files which do not have a .sql extension but are marked as executable (chmod +x) will be invoked with the connection string to the database passed as their first and only parameter.
All migrations which are successfully applied are tracked in an _versions table. The _versions table is created on first execution of the tool. Only missing migrations (migrations not already tracked in _versions) will be applied any time the tool is run.
All migrations are executed within a single database transaction so either all the specified migrations apply successfully or they all get rolled back (assuming the database supports this).
To run the tool simply execute the migrate command passing the database connection string as the first parameter followed by a list of migration files to apply. e.g.
migrate postgres://localhost/mydb migrations/*.*Add the following to your Gemfile:
gem 'migrate', git: 'https://github.com/garrettheaver/migrate.git'Currently the following databases are supported by the tool. Pull requests to support others are always welcome.
- PostgreSQL (
postgres://<user>:<pass>@<host>/<database>)
Since executable migrations (chmod +x files) must establish their own connection, any changes they make cannot be rolled back as part of the migrate command transaction should an error occur. Furthermore, they must set an appropriate transaction isolation level (such as READ UNCOMMITTED) to be able to see changes performed but not yet committed in an earlier migration.