Database Migrations & Seeds

Creating a Migration

rails generate migration IssueNumberShortDescription

The generator template automatically includes the system_init_done guard (return if !Setting.exists?(name: 'system_init_done')), which skips data-migration logic on fresh installations where seeds handle setup. Zammad migrations can include data migrations, not just schema changes.

• Call Model.reset_column_information after adding/removing columns in the same migration

Schema Changes

When modifying table structure, also update the base migrations so fresh installations get the same schema (migrations are cumulative, but base files are the canonical schema source for new installs):

• db/migrate/20120101000001_create_base.rb
• db/migrate/20120101000010_create_ticket.rb

Conventions:

• Use limit: 3 for timestamp precision
• Specify type: :integer for polymorphic references
• Use id: :integer in create_table calls (Zammad uses integer primary keys, not bigint)
• Add indexes for frequently queried columns

Settings

New or updated settings must be added in both the migration and the seeds. See existing examples in db/seeds/settings.rb for the correct structure.

Seeds

Seeds live in db/seeds/ and use idempotent methods like create_if_not_exists to avoid duplicates. When adding user-facing fields, also update db/seeds/object_manager_attributes.rb.

Helpers

MigrationHelper (lib/migration_helper.rb) provides utilities for renaming custom object attributes and handling reserved SQL words.

Testing

rake db:migrate              # Test the migration
rake db:rollback STEP=1      # Undo last migration (schema only — data migrations are not reversed)