Multi tier application with ASP.NET Core WebAPI, WebAPI service layer and Entity Framework Core backend. The frontend is made with Web Component and @nyaf.
Currently running ASP.NET Core 3.1 and EF 3.1 Core and build against Core 3.1 (Assemblys are .NET Standard 2.1).
The Project is no longer maintened! A new version using .NET 6 is available on Azure DevOps: https://dev.azure.com/joergkrause/Core3%20Microservice%20Boilerplate Despite the name, it's .NET 6 for sure!
For Windows users this is your preferred environment:
- Windows 10 Pro
- Docker Community Edition (nice to have, it works wothout Docker)
- Visual Studio 2019 16.6 (latest version strongly recommended)
- .NET Core SDK 3.1
I assume that your docker installation setup supports Linux containers. We don't need Windows here.
For Linux users, this is what you'll need:
- Ubuntu 16+ (recommended)
- Docker Community Edition
- Visual Studio Code (and some extensions to support C#)
- .NET Core SDK 3.1
The next sections below for details for each step.
- Restore packages and build
- Adjust connection strings for database
- Run "ConsoleApps/SetupConsole" app to create database
- Start the service layers - it runs on port 5001 (Authentication Service) and 5005 (Demo Service) by default
- Go to FrontEnd project's folder and execute
npm run build- This creates the Web app
- Now build the WebFrontEnd's API
- The service layers can output their OpenAPI definition files, but these are included in the package for convenience.
- Start the front end API - it's on port 5000 by default.
- Start the front end host - it's on port 8080 by default. This requires CORS to be enabled (it is in the project, of course).
If you use Visual Studio the Solution is set to "Multiple Startup Projects" and hence requires no multiple manual start actions, just hit F5.
This section guides you through the elementary setup steps.
The database environment needs a connection string. The test environment can make use of the SQLLocalDB, which comes with Visual Studio.
The connection string goes to (localdb)\MSSQLLocalDb. You can change this in SetupConsole and the ServiceLayers (micro services).
If you want to use another instance, just create a DB instance on the console like that:
sqllocaldb create JoergIsAGeek
Locate the connection strings in the appsettings.json file, one for the setup and one for each service runtime.
This is an example for the sqllocaldb:
"ConnectionStrings": {
"MachineDataContext": "Data Source=(localdb)\\MSSQLLocalDB;Initial Catalog=MachineDataDatabase;Integrated Security=True;MultipleActiveResultSets=True;Connect Timeout=30"
}
- git-workspace\Sources\MicroServices\Authentication\JoergIsAGeek.Workshop.Enterprise.ServiceLayer.Authentication\appsettings.json
- git-workspace\Sources\MicroServices\MachineData\JoergIsAGeek.Workshop.Enterprise.ServiceLayer.MachineData\appsettings.json
- git-workspace\JoergIsAGeek.Workshop.Enterprise.SetupConsole\appsettings.json
The WebFrontEnd has no direct DB access.
As an alternative way you can use a Docker container running MS SQL Server Developer Edition on Linux.
The project creates two databases, one for authentication and one for demo data. That's a common scenario, as usually you'll handle your identity stuff elsewhere. Also, in a more advanced scenario, you'll replace the internal identity stuff with external one (IdentityServer, OAuth federation service), so keeping it separate fromthe beginning is a good architectural choice.
The WebAPI consumes two services. That demonstrates a Microservice approach. To have a CI/CD flow, the build process includes pulling of the OpenAPI 3 descriptions. This process includes these steps:
- Clear the folder "Connected Services"
- Check if NSwag is available in the path. Nwsag is the proxy generator. Pull from npm if not yet installed.
- Create a docker image from Dockerfile available in the services root path.
- Run this particular image
- Execute a
Web-Getto pull the OpenAPI (Swagger) file and place it in the Setup folder of the API project. - Repeat steps 3 to 5 for each service
- Cal Nswag to create the proxies. That places the C# proxy files in the folder "Conntected Services"
- Build the project. Run!
The build script is currently placed in a pre-build event of the WebApi project. Check the *.csproj file!
The service layer exposes the API as OpenAPI 3 (f.k.a. Swagger, powered by NSwag). The frontend uses NSwag to create the proxy. Changes require to recreate the proxy. This is part of the build step.
The frontend project contains the Angular files. It works as usual.
npm i
npm run build
This makes a production build and copies the final bundle to the wwwroot folder of the WebApi project for delivery.
The easiest way to run all parts is Docker. There are no specific settings required, you can skip installing SQL Server and Core SDK.
There are 4 containers:
- Linux with MSSQL 2017 for database --> Port 1433
- Linux with ASPNET Core for micro service 1 (AuthService) --> Port 5001
- Linux with ASPNET Core for micro service 2 (MachineDataService) --> Port 5005
- Linux with ASPNET Core for web front end (Web App) --> Port 5000, exposed externally as Port 8080
The composer file runs all of them in production mode. The final solution runs on http://localhost:8080 on your host system or any other external IP. The 500X ports are for inter-container communication only.
The setup is in one batch file (Powershell), and it has some distinct parts you have to know.
The default ports are provided by runtime settings. All services start with a line such as this:
webBuilder.UseStartup<Startup>().UseUrls(args);
That means, you can provide the bindings as runtime parameters.
After a first build runs fine it's time to create the images. The appropriate Dockerfile files are in the folder Sources:
- DockerFile-Auth
- DockerFile-MachineData
The build process includes the whole process, compiling and publishing and can take a while. Give your machine a few minutes. After that, you can quickly start the containers from these images.
docker build --tag authservice:1.0 -f DockerFile-Auth .
docker build --tag machinedataservice:1.0 -f DockerFile-MachineData .
To quickly start all parts, the API, SQL Server, and Frontend just run the build in Visual Studio. Then, launch all parts using Docker Compose:
$ docker-compose --project-name workshop up
Assure you're in the solution rool and a command prompt, that's where the docker-compose.yml file is located. It will take roughly 5 minutes to start all containers.
Open the browser at location http://localhost:5000, then.
There are three demo users and some demo data available. The 'TestConsole' app adds these data:
- Weird Guest (Login: weird@guest.com, Password: p@ssw0rd)
- Doris Demo (Login: doris@demo.com, Password: p@ssw0rd)
- Andy Admin (Login: andy@admin.com, Password: p@ssw0rd)
There are three roles defined, each role has one member, except "User", which has also the Admin as a member:
- Guest (Member: Weird Guest)
- User (Member: Doris Demo, Andy Admin)
- Admin (Member: Andy Admin)
There is one Policy as a User claim defined:
{ "role": "api_access" }
The members "Doris Demo" and "Andy Admin" have this claim.
Weird Guest has no claim and hence cannot access the backend API.
Login, Register and this Info page can be opened by all users. A Guard (in Angular) is protecting the dashboard from user not properly logged on. If you click on "Dashboard" as an anonymous user your're are going to be redirected to the login form.
The demo data is a shop floor definition, we have a machine, which has devices and these devices have values. Only the user in the Role "User" can access these data.
This is made on Web Components, which is a HTML 5 native technology all current browsers support 100%. It's made with TypeScript, WebPack, and my @nyaf framework.
Have a look at https://github.com/joergkrause/nyaf for more informatio n regarding the underlying codes. It's much smaller than Angular but nonetheless extremely powerful fore SPA apps.