Identity Server: Deploy to Azure

This post is going to cover taking the existing set of applications we have been using to learn about Identity Server and deploying them to Azure. The starting point of the code can be found here.

Prep Work

The applications as they stand from the link above are not ready to be pushed to Azure most due to some configuration changes that are needed. We will go through each of the applications and take the hard-coded values and move them to  appsettings.json.

API Application Configuration

The API application needs two configuration values for the address of the Identity Application and the address of the Client Application. The following two lines need to be added to the application’s  appsettings.json file.

Then in the  Startup class, the values need to be used. The Identity Server address is used in the JWT Bearer setup.

Then the Client address is used in the CORS setup.

Identity Application Configuration

The Identity application needs a configuration value for the address of the address of the Client Application. The following line needs to be added to the application’s  appsettings.json file.

Next, the  Config class needs a reference to configuration passed into the  GetClients function.

Next, the references to  http://localhost:5002 need to be replaced with the value from the configuration. The following is one example.

Identity Application Entity Framework

As part of publishing this set of applications, this example is going to use Azure SQL and right now the application is set up to use SQLite. In the  Startup class replace  UseSqlite with  UseSqlServer. The following is an example of one of the needed replacements.

When switching database providers make sure to delete and recreate your database migrations. I didn’t to begin with and it cost me a lot of time in changing down a strange error which this post covers.

Client Application Configuration

The Client application needs two configuration values for the address of the Identity Application and the address of the API Application. The following two lines need to be added to the application’s  appsettings.json file.

Then in the  Startup class, the Identity Server Address needs to be used in the  AddOpenIdConnect call.

Next, the configuration values need to be passed to the Angular application. This process ended up being harder to figure out that I had anticipated and turned into a full blog post on its own. See this post for the details. The code for all the changes will also be on GitHub in case you need to the the diff for the client application.

Publish to Azure

Right-click on the Identity Application and select Publish.

This will show the Publish screen which provides the option to publish to Azure. We don’t have an existing App Service so we are going to create a new one. This page in the official docs explains all the options available on the publish screen. Click the publish button to continue.

The next screen that shows is the Create App Service Screen. I used all the default values and created a new Resource Group and App Service Plan. Keep in mind that the resource group and plan will be reused for the remaining two applications we are looking deploy. The only thing that will change between the applications on this screen will be the App Name.

The services tab looks like the following.

Next in the additional resources box lets hit the plus button next to SQL Database since our group of applications is going to need somewhere to store data. This will take us to the Configure SQL Database screen.

Since I don’t already have a SQL Server setup I am going to hit the New button to add one. That results in the next screen where you enter a name for the server as well as a username and password. After entering the required information click OK.

This will put you back on the configure database screen with most of it filled out. Make sure to set the database name you want to use.

Finally back on the Create App Service screen, you will see all the resources that you selected and configured. When you are satisfied with what you see click the Create button and let Azure do its magic.

When it is done you will see the profile now listed on the Publish page.

The above needs to be repeated for both the API and Client Applications, but using the Resource Group and App Service plan created above. Each profile should use a unique application name.

Identity Application Azure Configuration

The Identity Application needs access to the database that we created above. This means we need to set the  DefaultConnection. The first step is to determine what the connection string should be. On the Azure Portal in your list of resources select the SQL database that we created above.

On the next page copy the provided connection string. Now navigate to the Identity App Service and under the Settings section select Application settings. Scroll down and find the Connection strings section and enter the copied value as the DefaultConnection.

Just above the Connection strings section we also need to enter a few values in the App settings section. For the Identity Application, we need the Twitter key and secret as well as the address of the client application. The following is a screenshot minus the actual values.

For the ClientAddress use the URL found in the Overview of the Client App’s App Service page.

API Application Azure Configuration

From the list of resources select the API App’s App Service page and in the Settings section select Application settings. In the App settings section add values for  IdentityServerAddress and  ClientAddress. As with the  ClientAddress above the URLs for each application can be found on their respective App Service pages.

Client Application Azure Configuration

From the list of resources select the Client App’s App Service page and in the Settings section select Application settings. In the App settings section add values for  IdentityServerAddress and  ApiAddress.

Wrapping Up

At this point, you should be able to load up the application at the client address provided by Azure and have a working application. Overall the deployment to Azure was pretty easy. Getting the applications prepared to be deployed was a bit more challenging and sent me down a couple of rabbit holes. The code in its final state can be found here.

All Migrations are not Created Equal

While trying to deploy my sample Identity Server set of applications to Azure I got the following error when the Entity Framework migrations attempted to run.

This was not something I would get when attempting to run locally, but it failed every time when using SQL Azure. Long store short is that the migrations that were trying to be applied were created when I was using Sqlite as a backing store ( UseSqlite).

I deleted all the migrations and recreated them with the app being aware that it would be running on SQL Server ( UseSqlServer) and all worked as it should. It makes total sense that the migrations would vary based on the data provider being used, but not something I had really thought about. Not something I will forget again.

Unable to create an object of type ‘ApplicationDbContext’. Add an implementation of ‘IDesignTimeDbContextFactory

Forgive the long title, but this is an issue I have been fighting trying to upgrade an Identity Server 4 project to ASP.NET Core 2. There is an issue on GitHub dedicated to this exact error which can be found here. Before you go down the route of trying all the suggestions in the issue take a moment and make sure that nothing in the  Startup class is doing anything that would try to hit the database with Entity Framework.

There is a nice section in the official migration docs titled “Move database initialization code” which I seemed to have missed. So before you head down any rabbit holes like I did make sure this isn’t what is causing your need to add an implementation of IdesignTimeDbContextFactory.

As stated in the migration docs move database related code out of the  Configure function of the  Startup class and into the  Main function. The following is the example of this from the docs.

This will keep Entity Framework tooling from accidentally running code you didn’t expect.  With version 2 all the code in the  Configure function gets run.

Migration from ASP.NET Core 1.1.x to 2.0

On August 14th .NET Core 2.0 was released including corresponding versions of ASP.NET Core 2.0 and Entity Framework Core 2.0 which got with the finalization of .NET Standard 2.0. The links take you to the release notes for each item.

In this post, I will be covering taking the project used for the ASP.NET Basics series from 1.1.x to the 2.0 release. The starting point of the code can be found here. This post is only going to cover conversion of the Contacts project.

Installation

If you are a Visual Studio user make sure you have the latest version of Visual Studio 2017, which can be found here and at a minimum should be version 15.3.

Next, install the SDK for your operating system. The list of installs can be found here. For development, it is key that you install the SDK, not just the runtime. The following is a preview of what to expect on the download page.

Csproj

The  csproj file of the project being upgraded is the best place to start the conversion. The TargetFramework needs to be changed to 2.0.

Next,  PackageTargetFallback changed to  AssetTargetFallback.

There is a new Microsoft.AspNetCore.All package that bundles up what used to be a huge list of individual packages. Those individual packages still exist, but this new one wraps them and makes it much easier to get started. The following is the package list before and after.

Last change in this file is to change the  DotNetCliToolReference versions to 2.0.0.

Program.cs

Program.cs is another area that has been simplified by creating a default builder that does all the same things that were happening before but hide the details. Keep in mind the old version still works and is valid to use if you use case needs it.

Identity

The remaining changes I had to make were all related to Identity. In the  Startup class’s  Configure function the following change was needed.

Next, in the  ManageLoginsViewModel class, the type of the  OtherLogins property changed.

The  SignInManager dropped the  GetExternalAuthenticationSchemes function in favor of  GetExternalAuthenticationSchemesAsync. This caused changes in a couple of files. First, in the  ManageController the following change was made.

The second set of changes were in the  Login.cshtml file. First the function change.

Then the change to deal with the changed property names.

Wrapping up

With the changes in the Contacts project now works on ASP.NET Core 2.0!  Make sure to check out Microsoft’s regular migration guide. as well as their identity migration guide. A full list of breaking changes for this release can be found here.

There is a lot more to explore with this new release and I have a lot of projects to update. Don’t worry I won’t be doing a blog post on all of them, but if I do hit any issues I will create a new post of update this one with the fixes. The finished code can be found here.

Identity Server: Introduction

In the SPA based sample applications, this blog has used so far user authentication has either been completely ignored in order to keep the examples simpler or the sites have used ASP.NET Core’s built in identity to encapsulate the whole SPA. In this post (or series of posts) I am going to share what I learn along the way of creating an Angular (2+) application that utilizes ASP.NET Core as its host/API/backend.

This post isn’t going to cover any code it is just going to be a lot of the information I gathered in the process of learning more about Identity Server.

Following are all the post in this series.

Identity Server: Introduction (this post)
Identity Server: Sample Exploration and Initial Project Setup
Identity Server: Interactive Login using MVC
Identity Server: From Implicit to Hybrid Flow
Identity Server: Using ASP.NET Core Identity
Identity Server: Using Entity Framework Core for Configuration Data
Identity Server: Usage from Angular

Identity Server

According to their docs IdentityServer4 is an OpenID Connect and OAuth 2.0 framework for ASP.NET Core which enables Authentication as a Service, Single Sign-on, API Access Control and a Federation Gateway.

Obviously, that covers a lot of scenarios. The two that I am interested in are Authentication as a Service and the API Access Control which has driven my research which means that the other aspects of IdentityServer4 will not be included.

Official Samples

The IdentityServer GitHub account has a samples repo that contains a ton of examples. I have found the quickstart area of the repo to be the most helpful when starting out.

Based on all the quickstarts samples it looks like a typical setup involves a minimum of three projects. One for the API, one for the client and one for Identity Server. As you go through the samples the number of projects increase, but that is because of a wider range of scenarios that the sample is trying to cover.

References for learning

Damienbod has a whole series of blog posts related to IdentityServer4 and code to go along with it which can be found here. As a side note if you are interested in ASP.NET Core and you aren’t following damienbo you should be he has a ton of great content.

Blog posts
Videos

Identity Server Alternatives

Identity Server isn’t the only way to go there is a number of Software as a Service options that cover a lot of same scenarios. The following are some examples.

Auth0 – Check out the .NET Core related blogs by Jerrie Pelser
Stormpath
Amazon Cognito

Wrapping up

Obviously, I didn’t get a lot covered on how to actually do something with IdentityServer, but I wanted to share my starting point. This is an area I am going to continue digging it to and sharing information about as I learn more.

If you have any resources in this area please leave a comment below.

Entity Framework Core with SQLite Migration Limitations

This is part of what has turned into a series on Entity Framework Core with SQLite. The other parts can be found below.

Entity Framework Core with SQLite
Entity Framework Core Errors Using Add-Migration
Entity Framework Core with SQLite Scaffolding

The starting point of the code for this post can be found here.

Migration Limitations when using SQLite

SQLite’s ALTER TABLE is limited which in turn limits what Entity Framework Core can do via a migration. The official docs on the subject can be found here. These limitations are on the Entity Framework Team’s list of issues as an open enhancement and can be tracked here.

As long as you are just adding new tables or columns you would never notice the limitation, but if you have spelling problems like I do then the need to rename a column can be important. Thankfully things like ReSpeller (link is to the pro page, but a free version is available in ReSharpers extension manager) help with my spelling issues.

Unsupported example with a column rename

As an example of how to handle a migration that isn’t supported, we are going to rename the State property of the Contact class to Subregion.

Rename property on the model

Open the Contact class which can be found in the Models directory and make the following change.

Add a migration

With the property name change using the following command in the Package Manager Console to create a new migration.

Which produces the following migration class.

Error trying to apply the migration

As expected when an attempt to apply the above migration results in the following exception.

System.NotSupportedException: SQLite does not support this migration operation (‘RenameColumnOperation’). For more information, see http://go.microsoft.com/fwlink/?LinkId=723262.

Modify migration to manually rename the column

Searching for how to rename a column in SQLite will turn up a lot of results including this from the official docs and answers like this on StackOverflow. The gist of the how to do a rename is to create a new table with the desired schema, copy the data from the original table, drop the old table, and finally rename the new table to match the original name.

Now knowing the process the migration above can be modified to apply SQL directly instead of using Entity Framework Core to generate the SQL. This can be done by using the Sql function of the MigrationBuilder class. The following is the resulting migration.

You will notice that I didn’t bother doing the Down function, but the same idea would apply when trying to undo a migration. SQLiteStudio or similar tools can be used to generate the SQL above if SQL isn’t something you want to deal with.

Fix other references to the renamed field

This isn’t really the topic of this post, but I wanted to throw in a reminder that after a rename like this there are places that will need to be updated that the tooling may not have picked up. For example, make sure all your views are using the new column as well as any bind statements in your controllers.

Wrapping up

The first time I hit the need to rename a column and it resulted in an exception it was extremely frustrating. Over time as I learned what the tooling around SQLite provides it has become less of an issue. I look forward to seeing what the Entity Framework team does in the future around this issue. The finished code can be found here.

Entity Framework Core with SQLite Scaffolding

This is the third in what is turning into a series of post about using SQLite with Entity Framework Core. This post is going to cover adding a migration, scaffolding a controller and related views, and a few things that are harder to do using SQLite. The following are the first two post.

Entity Framework Core with SQLite
Entity Framework Core Errors Using Add-Migration

Adding Model, DbContext, Controller, and Views

If you have any experience with Entity Framework Core or have read any of my past entries on the subject this section is going to repeat some of the same information, but I am including it so someone who is looking for a full example will have it.

Model

In the Models folder add a Contact class similar the following.

DbContext

In the Data folder add a ContactsDbContext that inherits from DbContext. The following is an example that auto applies migrations to a database, if you don’t need that functionality it can be dropped out.

Now that the application has a model and a related DbContext the following can be used to add a migration that will create a Contacts in the SQLite database. Run from the Package Manager console.

Add-Migration is a Powershell command to add a migration (surprise!), AddContacts is the name of the migration and -Context ContactsDbContext is an argument that lets the command know which  DbConext to use. The Context is only needed if your application has more than one DbContext.

Controller and Views

With the above complete Visual Studio provides some tooling that makes it very fast to create a controller with views for listing, adding, editing, and deleting items. To begin right-click on the Controllers folder and select Add > New Scaffolded Item.

Select the MVC Controller with views, using Entity Framework option and click Add.

On the next dialog use the drop downs to select a model class and a data context class. Then verify the controller name and click add.

When the process completes the following items will have been added to your project.

Add to nav bar

To add a link to the new section of the app to the nav bar open the _Layout.cshtml in the Views/Shared/ directory. The following is the section of the file that needs to be changed to add an item to the nav bar.

Specifically, the following line was added to provide access to the contact list page.

Wrapping up

With the above, the application will be runnable. The code for this post can be found here. The next post in this series will cover the limitations of migrations when using SQLite with Entity Framework Core.

 

Entity Framework Core Errors Using Add-Migration

I started off trying to expand my sample from last week’s post and hit some issues when trying to add a migration for a new DbContext.

The Setup

I added the following DbContext that only has one DbSet and auto applies migrations in the constructor.

The command

Using Visual Studio’s Package Manager Console I ran the following command.

Error 1 – No parameterless constructor

The above command resulted in the following error.

No parameterless constructor was found on ‘ContactsDbContext’. Either add a parameterless constructor to ‘ContactsDbContext’ or add an implementation of ‘IDbContextFactory<ContactsDbContext>’ in the same assembly as ‘ContactsDbContext’.

I read the first sentence and added a parameterless constructor to  ContactsDbContext. I did think it was strange that a parameterless constructor wasn’t required the other contexts I had written in the past, but the error said to add a parameterless constructor so that is what I did.

Error 2 – System.InvalidOperationException: No database provider has been configured for this DbContext

Now having a parameterless constructor I ran the Add-Migration command again and was greeted with the following error.

System.InvalidOperationException: No database provider has been configured for this DbContext. A provider can be configured by overriding the DbContext.OnConfiguring method or by using AddDbContext on the application service provider. If AddDbContext is used, then also ensure that your DbContext type accepts a DbContextOptions<TContext> object in its constructor and passes it to the base constructor for DbContext.

The second error forced me to step back and think more about what the problem was as it didn’t have an action I could take as the first sentence, which is, of course, my fault for not fully digesting what the error was saying.

The fix

The bit I was missing was the fact that I hadn’t added the following to the  ConfigureServices function of the project’s  Startup class.

With the above added I removed the parameterless constructor from  ContactsDbContext and was able to successfully run the add migration command again.

Wrapping up

The moral of the story is to actually read the full error message before running off and trying to fix the problem. The second error message saying “using AddDbContext on the application service provider” is what triggered me to head in the right direction.

This was also a good reminder that tools like the ones used by Add-Migration can/do compile the project they are being used on in order to have enough context to perform their tasks.

Entity Framework Core with SQLite

All the applications used as examples for ASP.NET Core and Entity Framework Core from this site so far used database running SQL Server/SQL Express. In addition to the Microsoft-based SQL databases, Entity Framework Core has support for a number of other database providers. This post is going to look at using SQLite. A full list of the support database providers can be found here.

Starting point

Using Visual Studio 2017 I started with a new ASP.NET Core project using Individual User Accounts which ensured all the Entity Framework Core bits were present. The template in RC 4 used packages based on the Core 1.0.3 which I upgraded to 1.1.0. The project at this point can be found here.

Just a side note this project was created when Visual Studio 2017 was at the RC 4 stage. This code associated with this post will be updated when Visual Studio 2017 is released.

Naming warning

As you will be able to see with the structure of the solution I started this work using the project name SQLite. With this project name, it was impossible to get the SQLite package to install. If you see something like the following renaming your project should get you running.

This issue is where I found out what the problem was.

Add SQLite Packages

Right-click on the project file and click Manage NuGet Packages.

Select Browse and in the search box enter “Microsoft.EntityFramework.Sqlite” and install the two packages that are found.

Remove SqlServer Packages

While still in the Manage NuGet Packages screen click on the Installed tab. Select and uninstall the following packages.

Configuration changes

Open appsettings.json and in the ConnectionStrings delete the line for DefaultConnection. Next, in the same section add a line for a SQLite connection string. The following is the result.

The above will expect the database file to be in the same location as the application is running. For a debug build the database file can be found in the \bin\Debug\netcoreapp1.0\ directory of the project.

Startup changes

The final location to change is in the ConfigureServices function of the Startup class. The following shows the addition of the application DB context before and after the changes.

Wrapping up

The application is now runnable using SQLite as its backing data store. At this point, the only thing using data access is related to identity. The first time an attempt is made to access the database you may be prompted to apply migrations.

I have been using SQLite Studio to view the data in my database if you have that need outside of the application it does a good job.

The code in its final state can be found here.

ASP.NET Core Conversion to csproj with Visual Studio 2017 and update to 1.1.1

On March 7th Visual Studio 2017 was released bring the ASP.NET Core tools preview. ASP.NET Core 1.1.1 was also released. This post is going to cover converting the project from my MailGun post from being project.json based to csproj as well as migrating from the project from ASP.NET Core 1.0.2 to 1.1.1. Here is the project as it stood before I made any changes.

Visual Studio 2017

The first step is to get a version of Visual Studio 2017 (VS 2017) installed. The download page can be found here. Make sure to grab the community edition if you are looking for a free fully-featured IDE option. Check out this blog post from Microsoft on the many new features Visual Studio 2017 brings.

The installer for VS 2017 has changed a lot from previous versions. The way it works now is you select the workload you use and it only installs the bit it has to to keep the size of install down. The following is a screen shot from my install. I have a lot more workloads checked that is needed for just an ASP.NET Core application. At a minimum make sure the “ASP.NET and web development” workload gets installed. If you are interested in cross-platform development scroll to the bottom and also check “.NET Core cross-platform development”.

Project conversion

When you open the solution in VS 2017 it will prompt you to do a one-way upgrade.

After the conversion is complete a migration report will open. Below is mine. I had no issues, but if there were any this report should give you some idea of how they should be addressed.

As part of the conversion process, the following file changes happened.

That is all there is to the conversion. The tooling takes care of it all and your project should keep work just as before. The sample project post conversion can be found here.

Migration from 1.0.x to 1.1.1

The migration is almost as simple as the project conversion. In the solution explorer right click on the project to be migrated and select Properties.

Find the Target framework selection and select  .NETCoreApp 1.1. Then save your solution.

Next, open the NuGet Package Manager. It can be by right click on the project and selecting Manage NuGet Packages or from the Tools > NuGet Package Manager > Manage NuGet Packages for Solution.

Select the Updates tab and update all the related packages to 1.1.1 and click the Update button.

If you want a specific list of all the package changes check out the associated commit.

The only other change needed is in the constructor of the  Startup class.

Wrapping up

After all the changes above your solution will be on the latest released bits. Having been following releases since beta 4 I can tell you this is one of the easiest migration so far. I may be partial, but .NET and Microsoft seem to be getting better and better over the last couple of years.

I am going to leave you with a few related links.

ASP.NET Core 1.1.1 Release Notes
Announcing New ASP.NET Core and Web Dev Feature in VS 2017
Project File Tools – Extension for IntelliSense in csproj
Razor Language Services – Extension for tag helper Intellisense