I have a problem with my MVC 4 application which used to work fine, but stopped for some reason, and I cannot find out why. I use simple memebrship provider and code first approach. This is my Index action method in the home controller
[Authorize]
public class HomeController : Controller
{
private IActivityRepository repo;
public HomeController(IActivityRepository activityRepository)
{
repo = activityRepository;
}
//Allow anonymous to allow to create database if there isn't one yet
[AllowAnonymous]
public ActionResult Index()
{
repo.InitializeDatabase(); //!!!!!!!!!!!!!!!!!!!!!
return RedirectToAction("ManageActivities");
}
The whole concept of mine is that if database doesn't exist it gets created in InitializeDatabase Method. Then user is redirected to ManageActivities action method which is decorated with [Authorize] attribute, what in effect takes user to login action method in AccountCotroller (out of the box in MVC4). This controller is decorated with [InitializeSimpleMembership], what fires InitializeSimpleMembershipAttribute filter.
This logic worked fine for me a while ago. Today I wanted to create a new database for testing purposes. When I create data context I call the base class with a custom name for the database like so:
public class ActivityLogContext : DbContext
{
public ActivityLogContext() : base("ActivitiesConnection")
{
}
So I've changed details for my connection string and run the application. Unfortunatelly, for some reason the code hits InitializeSimpleMemebership filter before running Index method from the home controller (even though its decorated with [AllowAnonymous]). In effect simple membership is initialized but database does not yet exist, what runs me into error.
My question is, why InitializeSimpleMemebership filter is getting released on application start if Index method doesn't require authorization?
I would eliminate the use of the InitializeSimpleMembership as discussed in this article. Move initialization to the Global.asax Application_Start method and do your initialization there also, so that it happens in the correct sequence.
Related
I'm writing an ASP.NET Core API in which the dependency injection container provides me with instances of my DbContext class. I set the service lifetime of the DbContext to transient (which means that the container will create a new instance for every class and every request).
builder.Services.AddDbContext<ComparisonInfoContext>(dbContextOptions =>
dbContextOptions.UseSqlServer("Connection string"), ServiceLifetime.Transient, ServiceLifetime.Transient);
Here is my DbContext class:
public class ComparisonInfoContext : DbContext
{
public DbSet<ComparisonInfo> Comparisons { get; set; } = null!;
public ComparisonInfoContext(DbContextOptions<ComparisonInfoContext> options)
: base(options)
{
}
protected override void OnModelCreating(ModelBuilder modelBuilder)
{
modelBuilder.Entity<ComparisonInfo>().OwnsMany(c => c.DiscrepancyInfos);
}
}
I also have a BackgroundService class called JobChecker which I added by calling:
builder.Services.AddHostedService<JobChecker>();
In JobChecker, I have a function that runs every minute and the first thing that function does is call context.Comparisons.ToListAsync(). The first time the function runs everything works fine, but the second time, this call returns outdated values for some reason (The values it returns used to be in the database but were changed prior to this call being made).
What I think is happening is the following:
JobChecker (the BackgroundService class) receives it's instance of DbContext at the very start of the program.
The function inside JobChecker calls context.Comparisons.ToListAsync() and gets the correct values from the database (no problems here).
Some controller somewhere changes the values in the database using it's own instance of the DbContext.
The function inside JobChecker runs and calls context.Comparisons.ToListAsync(). It sees that it is already tracking instances with the same id which came from the previous call (step 2). These instances were not effected by the changes made by the controller in step 3. It does not materialize the values it got from the database, instead it returns the instances that already exist. This is explained more here: The life of a query
Does anyone know whether this is in fact what is happening? Is it even correct to set the service lifetime of the DbContext class to transient?
Your DB Context might be transient, but the BackgroundService is not. In effect you only have a single instance of the context in the BackgroundService.
I would look at injecting in an DB Context Factory:
services.AddDbContextFactory<ApplicationDbContext>(...)
Then inject one into your background service:
public class MyBackgroundService(IDbContextFactory<ApplicationDbContext> contextFactory)
Then every iteration of your loop in the background service (it sounds like it is running on a timer), you can use this to instantiate a new context.
using (var context = _contextFactory.CreateDbContext())
{
// ...
}
Docs:
https://learn.microsoft.com/en-us/ef/core/dbcontext-configuration/#using-a-dbcontext-factory-eg-for-blazor
I once faced the same problem with you ,and finally fixed it follow the document:
public SomeBackgroundService( IServiceProvider services)
{
......
}
using (var scope = Services.CreateScope())
{
var context = scope.ServiceProvider.GetService<BackgroundServiceProjContext>();
var personlist1 = context?.Person.ToList();
}
I modified the name and the context injected directly into backgroundservice still read the former value:
I am just getting to grips with Asp.net Core and I'm trying to set up a basic site.
I want to build an admin panel that is under a subdirectory.
I have a simple controller which was scaffolded by the EF crud feature.
So it seems that from the examples I should just be able to add a [Route()] attribute to the controller and it will prefix everything. Something like this:
[Route("Admin/Subfolder/[controller]")]
public class EventsController : Controller
{
}
But when I do that I just get an error page saying "multiple actions matched" and it lists index, details, create, etc.
I can get it working if I then go through every method and put a [Route()] attribute on it but this doesn't seem to be in line with the documentation.
It feels like I should be able to just add a prefix to the controller route without having to take over management of every route within the controller. Case in point, the POSTS are not working now and I'm not sure what the format of the route attribute should be for them.
What am I doing wrong?
You are doing it correctly. Default route attribute can be applied at the controller level. “Placing a route attribute on the controller makes all actions in the controller use attribute routing.”
Can you post complete code of your controller? There must be something else going on in there. Make sure you use HttpPost/HttpGet attribute for actions with the same name, like so:
[Route("Admin/Subfolder/[controller]")]
public class EventsController : Controller
{
[HttpGet]
public IActionResult NewEvent()
{ }
[HttpPost]
public IActionResult NewEvent()
{ }
}
Good explanation on routing can be found here
I have a legacy Application that we're converting to use the MVC 5 Application template. We have a custom API method, to keep the example simple let's just say it's signature is:
bool Login(username, password);
How can I set the User as logged in, so that I can use things like the [Authorize] attribute? For the moment we want the simplest method possible just to get us started developing the site.
I tried implementing this to set User.Identity manually. But this is then reset on every subsequent request.
In the end I extracted out the logic to the Account controller. This handles the Login and stores the result in the Session. Then I just needed to override the System.Web.Mvc.AuthorizeAttribute class and AuthoriseCore method as follows:
using System.Web;
using System.Web.Mvc;
namespace HomeHealth.Web.Infrastructure
{
public class HomeHealthAuthorizeAttribute : AuthorizeAttribute
{
protected override bool AuthorizeCore(HttpContextBase httpContext)
{
return WebSession.SdkContext.IsAuthenticated;
}
}
}
It has some helper code to clean up accessing the Context from the session, but that's irrelevant. The point is that this is the Attribute/Method you probably want. You can then mark Controllers/Methods with the following:
[HomeHealthAuthorize]
public class PatientController : BaseController
Then all the checking/redirecting is done for you.
I am rewriting an ASP.NET webforms app in MVC4 and was wondering how to solve the following problem. It is a multi-tenant app, so part of the URL has the tenant NAME in it:
http://mysite/tenant/controller/action
But tenant is an abbreviation representing the tenant, but I'd like to always convert that to the corresponding integer id and use that throughout the code. What is the best way to write that convert code once and have some variable/property available to all controller methods.
public class DivisionController : Controller
{
//
// GET: /Division/
public ActionResult Index()
{
// I want this.TenantId to be available in all controller methods
FetchDivisions(this.TenantId);
return View();
}
Is a base controller the best way to handle this or filters or attributes?
Yes a base controller will handle this just fine. If you need to perform a database lookup to convert the abbreviation to the integer value you can use the OnActionExecuting event like so:
protected override void OnActionExecuting(ActionExecutingContext filterContext)
{
base.OnActionExecuting(filterContext);
// Lookup code here.
}
MVC 4 does present me some strange behaviour at the moment.
Imagine the following Code:
TestController.cs
public class TestController : Controller
{
public ActionResult Index(Function function, string action)
{
return View();
}
public class Function
{
public string Action { get; set; }
}
}
It seems, that when I call the URL directly through the browser (localhost:PORT/Test), the Action-Property gets automatically filled with "Index".
If the Action would be named "MySuperDuperActionWhichGetsInvokedQuiteOften", exactly this Methodname would be in the property.
Can somebody explain what MVC is doing here?
The Problem is, that I of course want to fill that stuff myself, for example through an AJAX-Query. But if MVC is filling in this property all by itself, this breaks some behaviour.
I could, of course, just rename my property and it would be working, but it would still be quite interesting what's going on.
EDIT
I would understand it that my second parameter, string action, get's filled with the method-name. But why on earth would MVC bind any property/parameter that is named the same to the request-value of it?
It is problem with default model binder. It "maps" request fields to properties in your class. There is an article of MSDN describing how does it works but to simply this situation the code will be like this:
Action = Request["action"] //where of course Request["action"] equals to name of your action