Evan Dontje

Entity Framework Code First makes generating database migrations dead simple. They can then be run from the commandline with the migrate.exe executable included in the Entity Framework nuget package. However, in some continuous deployment scenarios it’s easier to execute the migration from C# code. It turns out this process is pretty simple too, but not as well documented.

Running migrations from code

The three necessary classes are DbConnectionInfo, your specific implementation of DbMigrationsConfiguration, and DbMigrator. First, the DbConnectionInfo has to be configured

string sqlConnectionString = "YOUR_CONNECTION_STRING_HERE";
var connectionInfo = new DbConnectionInfo( sqlConnectionString, "System.Data.SqlClient" );

The only tricky part is the second parameter. It’s the “InvariantName” of the SQL provider that should be used with the provided connection string. "System.Data.SqlClient" is the provider for SQL Server according to the MSDN documentation.

Next, the DbMigrationsConfiguration must be set up

var configuration = new MyDbMigrationConfiguration
                    {
                       TargetDatabase = connectionInfo,
                       AutomaticMigrationsEnabled = false,
                       AutomaticMigrationsDataLossAllowed = false
                    };

In the code above I disabled automatic migrations. For my continuous deployment scenario I want to be able to control the migrations explicitly, so I’ve disabled the AutomaticMigration features.

Finally, run the migrations using a DbMigrator instance

var migrator = new DbMigrator( configuration );

// Roll back all migrations
migrator.Update( "0" );

// Roll back to a specific migration
migrator.Update( "MigrationNameToRollBackTo" );

// Apply all migrations up to a specific migration
migrator.Update( "MigrationNameToApply" );

// Apply all migrations
migrator.Update();

As you can see above the Update method gives you a lot of flexibility to control the migrations applied. One thing worth noting is that the string argument of the Update method is the class name of the migration, not the file name. This is important because they are usually different. The filename usually has a timestamp prepended to the name, while the class name does not.