I'm writing a web api attached to a database that I created using Model First.
Being new I was confusing Model First with Code First.
As it's my first EF/Web app I didn't realize that migration wasn't yet implemented by Model First and I followed the procedure to implement it. When it ran I received some errors, looked them up and only then realized I'd need to remove it again.
Following somebody's instructions I deleted the migrations folder, plus the _migrationHistory table from the database.
But now I receive the following exception whenever the _migrationhistory table is missing from the database.
protected void Application_Start()
{
// Call initializer
Database.SetInitializer(new PulseDbInitializer());
new ApplicationDbContext().Database.Initialize(true); <- error here
AreaRegistration.RegisterAllAreas();
GlobalConfiguration.Configure(WebApiConfig.Register);
FilterConfig.RegisterGlobalFilters(GlobalFilters.Filters);
RouteConfig.RegisterRoutes(RouteTable.Routes);
BundleConfig.RegisterBundles(BundleTable.Bundles);
}
"An exception of type 'System.NotSupportedException' occurred in EntityFramework.dll but was not handled in user code
Additional information: Model compatibility cannot be checked because the database does not contain model metadata. Model compatibility can only be checked for databases created using Code First or Code First Migrations."
The following is a partial description of my initializer
public partial class PulseDbInitializer : DropCreateDatabaseIfModelChanges<ApplicationDbContext>, IPulseDbInit
//public partial class PulseDbInitializer :DropCreateDatabaseAlways<ApplicationDbContext>, IPulseDbInit
{
private ApplicationDbContext _context;
protected override void Seed(ApplicationDbContext context)
{
_context = context;
var pid = new PulseDbInitializionData(context);
pid.Init(this);
}
}// class PulseDbInitializer
It seems that the app is trying to use migrations (because it's creating the _migrationhistory table) even though I've removed the migrations subfolder.
Being new at this, I'm not sure where to go. I'm aware that I've created my initializer, see above, so am I missing something?
EDIT - Too long for a comment
Im not sure we're understanding each other.
Very simply, I don't know what's generating the _MigrationHistory table in my database as afaik if I remove the Migration subdirectory from my project and delete the database, then I thought that's all I need to do to drop migrations.
In my model first, the DropCreateDatabaseIfModelChanges is what I've always used for generating the database and is working as expected. It creates a new database, or recreates when the model changes.
Until I added the migration I didn't notice this _MigrationHistory table and I don't have a clue if the add-migration command added any code to my project that I need to delete or change (I really hate things happening and not being told exactly what was done in some kind of log. Please microsoft, for future reference)
Or am I totally wrong and the _MigrationHistory table was always there but I never noticed? It all seems to work, as long as I don't remove the table.
Your initializer is inheriting from DropCreateDatabaseIfModelChanges, which checks for model changes. Since you are not using code first you can't check for model changes in code (since your code doesn't define the model -that'd be code-first-... the model defines the code -that's model-first-)... so you need to implement your own IDatabaseInitializer which only checks (for example) that the database exists:
public partial class PulseDbInitializer : IDatabaseInitializer<ApplicationDbContext>, IPulseDbInit
{
public void InitializeDatabase(ApplicationDbContext context)
{
var exists = new DatabaseTableChecker().AnyModelTableExists(context.InternalContext);
if (exists == DatabaseExistenceState.Exists)
{
// maybe check if certain data exists and call the `Seed` method if
// it doesn't
return;
}
// Throw some error if it doesn't exist
}
protected override void Seed(ApplicationDbContext context)
{
_context = context;
var pid = new PulseDbInitializionData(context);
pid.Init(this);
}
}
Related
I am trying to use EF Core 3.0 migrations with a hybrid of an existing tables and new tables built with code first. To prevent the scaffolding of the existing tables I would like to decorate the model class with an attribute (fluently or annotations) so that the migration code generation for those tables is skipped but the model is still built into the DbContext class.
The approach I'm taking is to add the following lines to OnConfiguring
optionsBuilder.ReplaceService<IMigrationsSqlGenerator, SkipMigrator>();
And then creating a SkipMigrator with the following code
public class SkipMigrator:SqlServerMigrationsSqlGenerator
{
public SkipMigrator(
MigrationsSqlGeneratorDependencies dependencies,
IMigrationsAnnotationProvider migrationsAnnotations)
: base(dependencies, migrationsAnnotations){}
protected override void Generate(
MigrationOperation operation,
IModel model,
MigrationCommandListBuilder builder)
{
if (operation.FindAnnotation("SkipMigrations")!=null)
{
Console.WriteLine("Skipping table:");
}
else
{
base.Generate(operation,model,builder);
}
}
}
I assumed that the Generate method was what triggered the creation of the migration code file but it never gets called. Id there a different place I should be intercepting the code generation?
If there a different/simpler way to tell migrations to skip tables yet still keep them in my DbContext?
Your means of trying to create your own IMigrationsSqlGenerator was correct, and I've used that approach before to alter the SQL that is generated as part of a migration.
services.AddDbContext<MyDbContext>(opt =>
{
opt.UseSqlServer();
opt.ReplaceService<IMigrationsSqlGenerator, SkipMigrator>();
});
However, as of EF Core 5.0 it's now much easier to exclude specific tables from migrations using the ExcludeFromMigrations() method on a TableBuilder:
public class ReportingContext : DbContext
{
public DbSet<User> Users { get; set; }
protected override void OnModelCreating(ModelBuilder modelBuilder)
{
modelBuilder.Entity<User>().ToTable(nameof(Users), t => t.ExcludeFromMigrations());
}
}
More info: https://devblogs.microsoft.com/dotnet/announcing-entity-framework-core-efcore-5-0-rc1/#exclude-tables-from-migrations
If there a different/simpler way to tell migrations to skip tables yet still keep them in my DbContext?
Yes, but it requires a different approach.
Instead of using your current DbContext class for creating migrations, create a SecondDbContext class only for the very purpose of creating migrations. This new SecondDbContext class will hold the DbSets<T> that you want EF to do its migrations on.
Then simply specify the second context when calling add-migration UpdateTable -c SecondDbContext and then update-database -c SecondDbContext.
My Question was originally: DropCreateDatabaseIfModelChanges fails to update database?
But I changed it because the role of these two methods is more ambiguous for me
.
...................
What I know so far that using DropCreateDatabaseIfModelChanges will recreate the database whenever the schema changes (IfModelChanges). That mean that I don't have to worry about migration, since it will recreate the database from scratch anyway, right?
I first created a class DataInitializer inheriting from DropCreateDatabaseIfModelChanges, and implemented the Seed method with the initial data, and called the Database.SetInitializer(new DataInitializer());
from the Main method, but I still get the error about model changing and I should add a migration (since I dropped a property from the entity to test the data initializing).
this is the initializer class:
public class DataInitializer : DropCreateDatabaseIfModelChanges<BlogContext>
{
protected override void Seed(BlogContext context)
{
var blogs = new List<Blog>
{
new Blog {FollowersCount=456, Name="ABC", Url="abc.com" },
new Blog {FollowersCount=789, Name="DEF", Url="def.com" },
new Blog {FollowersCount=246, Name="GHI", Url="ghi.com" },
new Blog {FollowersCount=135, Name="JKL", Url="jkl.com" },
new Blog {FollowersCount=258, Name="MNO", Url="mno.com" },
};
blogs.ForEach(x => context.Blogs.Add(x));
}
}
These are my questions: Does it work if migration is enabled, and what happens if I'm implementing the Seed method in the Configuration class with initial data- which one will be called (and I have a Seed method initialized with data already in the Configuration class)
After I added a migration, it throws this error (whether I called SetInitializer or not):
There is already an object named 'Blogs' in the database.
You have to tell the web.config file to use your initializer class whenever Model changes. Add the following code in the entityFramework tag in web.config file. Hope it will work..
<contexts>
<context type="YourProjectName.BlogContext,YourProjectName">
<databaseInitializer type="YourProjectName.DataInitializer , YourProjectName" />
</context>
</contexts>
Following is a part of code in my DbContext that creates data and seeds data.
public class labSchedulerDbContext :DbContext
{
static abDbContext()
{
Database.SetInitializer(new abDatabaseInitializer());
}
public abDbContext() :base (nameOrConnectionString:"abDbContext")
{
}
protected override void OnModelCreating(DbModelBuilder modelBuilder)
{
modelBuilder.Conventions.Remove<PluralizingTableNameConvention>();
}
}
Rest of the code is just declaring models.
In my setinitializer class with seed
public class abDatabaseInitializer : DropCreateDatabaseIfModelChanges<abDbContext>
{
.....
}
Code is working fine. Problem I have is when I created site in IIS and ran it. It runs fine but after few hours it will delete the database and initialize and seed data again. Thus we lose the data added. I tried to comment out lines where setinitilizer file is being called. Still didn't work. I refreshed apppool and site maybe it was cached files. But still no luck.
Can you please tell me how I can disable the whole initializer file once a database is created so no matter what it doesn't go and drops db and create a new one.
Cod
You can use this other initializer:
CreateDatabaseIfNotExists: This is default initializer. As the name suggests, it will create the database if none exists as per the configuration. However, if you change the model class and then run the application with this initializer, then it will throw an exception
The only thing that you have to do is to drop the database yourself before you deploy a new version with a different model. (Provided that's what you want to do).
If I build an application and let code first figure out where to put the database and the user inserts data via the application, will that data get lost on a click once update ? If so, how do I approach this problem ?
Thanks
No there is No need to "lose" any data when using automatic migrations.
You migration config class should state no data loss allowed
You will need to build custom scripts/ or tweak the generated scripts when dealing with changes that result in data loss.
public override void MigrateDb() {
Database.SetInitializer(new MigrateDatabaseToLatestVersion<MYDbContext, MYSECIALMigrationConfiguration>());
// Context = GetDefaultContext(); // check if a new context is really needed here
Context.Database.Initialize(true);
}
public class MYSPECIALMigrationConfiguration : MYBaseMigrationConfiguration<MYDbContext>{ }
public abstract class MYBaseMigrationConfiguration<TContext> : DbMigrationsConfiguration<TContext>
where TContext : DbContext{
protected MYBaseMigrationConfiguration() {
AutomaticMigrationsEnabled = true; // you can still chnage this later if you do so before triggering Update
AutomaticMigrationDataLossAllowed = true; // you can still chnage this later if you do so before triggering Update
}
how to approach migrations.
.. Thats actual a big question.
EF6 Migrations - new features and options
Great info on Migrations when working in teams.
This covers many scenarios you may recognise and thus help you understand what approach suits you best.
I created a DbContext like so :
public class myDB : DbContext
{
public DbSet<Party> Parties { get; set; }
public DbSet<Booking> Bookings { get; set; }
}
This generated my DataBase and the two tables above..Great
I then decided to add another DbSet into the mix & I got an error:
the model backing the 'Party' context has changed since the database was created
I'm aware of the fixes for this using modelbuilder.IncludeMetadataInDatabase = false; and Database.SetInitializer<ClubmansGuideDB>(null);
1) What's the correct way to add my new classes to the context and have them generated in the DataBase?
2) In my futile attempts to solve this myself I deleted the tables and tried to re-run the app with no success I then deleted the full database and re-run and it doesn't re-generate the DB. How can I start from scratch - is there some cache somewhere?
I believe you're looking for:
Database.SetInitializer<ClubmansGuideDB>(new DropCreateDatabaseIfModelChanges<ClubmansGuideDB>());
As of EF 4.1 and above.
Note that this assumes you have permission to even drop your database. I do this locally for ease of development but disable it in staging/production (I do a manual Schema Compare and push my changes).
By the way, for testing, if you need to force recreate the database, you can do:
using (var context = new ClubmansGuideDB()) {
context.Database.Initialize(force: true);
}
(using if you don't already have a reference to your DB)
You can let the Entity Framework recreate the whole database by using a Database Initializer or if you want to migrate data you can look at the Code First Migrations
The following would recreate your database when the model changes:
Database.SetInitializer<myDB>(new DropCreateDatabaseIfModelChanges<myDB>());
Try putting this in your Global.asax.cs Application_Start() method.
Database.SetInitializer<DatabaseContext>(null);
To reset the database from scratch on app run make a class like this
//Where myDB is your context
public class EntityBase: DropCreateDatabaseAlways<myDB>
{
}
Then in your Application_Start() method you can do this
Database.SetInitializer(new EntityBase());