Entity Framework : Sharing entities across different DbContexts

You can try using views, declare the user as a view in PluginDataContext and when you perform the migration, type the method "create User view as ...", this allows you to relate the book to the user.

This solution could help you:Entity Framework 6 Code First Migrations with Multiple Data Contexts. However, in this case, both context are in the same project. I don't know if works with contexts that are in two different projects (I think it should if you are using the same class to map User). As the blog said, you need to comment the generated code related to the Users table when you run the Add-Migration command for the PluginX Context.

When you add the Booking entity, don't use the DbSet.Add() method. Instead use the DbSet.Attach() method and set the DbContext.Entry(Entity).State property for the Booking to EntityState.Added and make sure the DbContext.Entry(Entity).State for User stays EntityState.Unchanged.

So for example instead of doing this:

pluginDataContext.dbBooking.Add(myNewBooking);

Do this:

pluginDataContext.dbBooking.Attach(myNewBooking);
pluginDataContext.Entry(myNewBooking).State = EntityState.Added;

This is because the Add() method marks all entities in the object graph as EntityState.Added which will cause inserts without checking if the entity already exists in the database. The Attach() method simply makes the context begin tracking the entity.

This is why I almost never use DbSet.Add().

When you work with multiple contexts you have two options:

1. Treat each context like they were separate applications. Imagine your user is an external resource that you get from a web service. You won't be able to add a foreign key to that. What you would do in this is either add only the userId in your tables and when you need the user details call the external service to get them or have a local light copy of the user in the Bookings context that you would update every now and then from the Users context. This approach is good when you work with a large system and you want to isolate the parts (read about DDD and bounded contexts)
2. Apart from your 2 contexts, create a third context with the whole model (users, bookings, etc). You will use the complete context to create the migrations and maintain the DB structure, but in the application you will use the smaller contexts. This is a very simple solution. It's easy to maintain the migrations with a single context and it still allows you to isolate the DB operation in smaller contexts that don't have access to unrelated entities.