Saltar al contenido principal

Database migration

Casdoor updates its database schema on its own when it starts. This page explains how that works and how Casdoor developers add a migration for a change that can't be automated.

Automatic schema updates​

Casdoor accesses the database through XORM. At startup, XORM compares the tables with the Go structs and adds missing tables and columns. For the details, see the XORM documentation on schema operations.

XORM doesn't rename columns and doesn't transform existing data. Those changes need a migration.

Migrations​

A migration is a function that runs once against the database. Casdoor writes migrations with the XORM migrate package.

The following migration is part of renaming the column p_type to ptype. XORM adds the new column ptype, and the migration fills it. Another step removes the old column.

migrations := []*migrate.Migration{
        {
            ID: "CasbinRule--fill ptype field with p",
            Migrate: func(tx *xorm.Engine) error {
                _, err := tx.Cols("ptype").Update(&xormadapter.CasbinRule{
                    Ptype: "p",
                })
                return err
            },
            Rollback: func(tx *xorm.Engine) error {
                return tx.DropTables(&xormadapter.CasbinRule{})
            },
        },
    }
    m.Migrate()

Each migration has an ID. Casdoor stores the IDs of the migrations that it has run in the database and skips them on later starts.

See also​