After upgrading to ServiceStack to 4.0.24.0, I started receiving this below error when trying to login using Google OAuth.
The same works locally on my dev machine.
Has anything changed since the 4.0.21.0?
The site is hosted on Azure as you can see from the address bar.
I've the licensed version of SS
Google OAuth Callback url is correctly registered http://somesite.azurewebsites.net/auth/GoogleOAuth
Using ServiceStack.Authentication.OAuth2 4.0.24.0
This is a pretty strange problem, but recently did some digging and here is what I found. It seems that Azure Websites doesn't like the HttpResult returning from the method Authenticate on the OAuth2Provider.
Important to note: This problem is only present when hosted on Azure Websites. ServiceStack OAuth2Providers (4.0.24 to 4.0.40) work as expected when hosted else where.
Specifically, it doesn't like the Content-Length header being set when returning a HttpResult. I'm not sure exactly why this is the case as it works locally and when deployed to my own IIS host on AWS.
I've created a repository you can deploy yourself that highlights this issue.
Azure Websites seem to be interpreting this response differently than on a standard IIS setup (eg, installed via Web Installer on a clean VM) which seems to be causing the problem.
One (not very nice) work around would be to use a GlobalResponseFilter stripping out Content-Length header of the only the GoogleOAuth2Provider response.
I've tested this out and indeed worked as expected and I was able to login using Google OAuth. Here is an example of using a GlobalResponseFilter to work around this problem.
GlobalResponseFilters.Add((req, res, dto) =>
{
if (!req.PathInfo.EndsWith("/auth/GoogleOAuth", true, CultureInfo.InvariantCulture))
{
return;
}
HttpResult httpResult = dto as HttpResult;
if (httpResult != null && httpResult.Headers.ContainsKey("Content-Length"))
{
httpResult.Headers.Remove("Content-Length");
}
});
Would be great if someone can point out why Azure websites does this.
Hope this helps.
Related
I created a simple Blazor WASM webapp using C# .NET5. It connects to some Functions which in turn get some data from a SQL Server database.
I followed the tutorial of BlazorTrain: https://www.youtube.com/watch?v=5QctDo9MWps
Locally using Azurite to emulate the Azure stuff it all works fine.
But after deployment using GitHub Action the webapp starts but then it needs to get some data using the Functions and that fails. Running the Function in Postman results in a 503: Function host is not running.
I'm not sure what I need to configure more. I can't find the logging from Functions. I use the injected ILog, but can find the log messages in Azure Portal.
In Azure portal I see my 3 GET functions, but no option to test or see the logging.
With the help of #Aravid I found my problem.
Because I locally needed to tell my client the URL of the API I added a configuration in Client\wwwroot\appsettings.Development.json.
Of course this file doesn't get deployed.
After changing my code in Program.cs to:
var apiAddress = builder.Configuration["ApiAddress"] ?? $"{builder.HostEnvironment.BaseAddress}/api/";
builder.Services.AddHttpClient("Api",(options) => {
options.BaseAddress = new Uri(apiAddress);
});
My client works again.
I also added my SqlServer connection string in the Application Settings of my Static Web App and the functions are working as well.
I hope somebody else will benefit from this. Took me several hours to figure it out ;)
Here's what I'm doing.
Install latest Azure SDK (by the date)
Open Visual Studio 2013
Create a new Azure Mobile App project. The simple service is created with TodoItem DataObject and 2 simple controllers - TodoItemController and ValuesController
Do not change anything
Start the project
The service is started and hosted in local IISExpress on url http://localhost:50993/ (the port may vary).
The "This mobile app is up and running" web page is opened in the browser. But http 400 error is returned when I try to invoke some GET-actions: for example http://localhost:50993/api/values or http://localhost:50993/tables/TodoItem.
Any ideas? Is something wrong in my environment or is that me doing something wrong?
Thanks.
I guess you can opt out of version checking by setting a value of true for the app setting MS_SkipVersionCheck. Specify this either in your web.config or in the Application Settings section of the Azure Portal.
This generally happens when you don't add a ZUMO-API-VERSION header to the request. This is required when making requests from a REST client, but the mobile client SDKs add the header automatically.
To fix, add the header ZUMO-API-VERSION with value of 2.0.0.
To learn more, see https://azure.microsoft.com/en-us/documentation/articles/app-service-mobile-client-and-server-versioning/.
Adding
http://localhost:50993/tables/Location?ZUMO-API-VERSION=2.0.0
at the end will do the trick.
I think I have a similar issue as this post-
Webactivator doesn't run on IIS 7
Swashbuckle running with katana-owin based web api is able to work with IIS Express.
assume all the urls below have an "http" prefix.
I can browse to something like localhost:8085/swagger Which redirects me correctly to localhost:8085/swagger/ui/index
But when published to IIS redirects to localhost/swagger/ui/index instead of localhost/myapp/swagger/ui/index and this results in a 404. Notice that even though an application name is specified while publishing somehow swashbuckle/swagger doesn't know and only uses the pathbase without the application name.
basically it just picks up the site root URL instead of the application URL that was used to call /swagger
Any solution to this problem?
I got the same issue. Solved this using the following code (c.RootUrl)
config.EnableSwagger(c =>
{
c.IncludeXmlComments(GetXmlCommentsPath());
c.SingleApiVersion("v1", "yourAPI");
c.RootUrl(req => req.RequestUri.GetLeftPart(UriPartial.Authority) + VirtualPathUtility.ToAbsolute("~/").TrimEnd('/'));
});
Reference : Relative path for UI request URL
I created a new Web API project and created the following routing spec (actually I have simplified, looking for the bug):
// Web API configuration and services
config.Routes.MapHttpRoute(
name: "DefaultApi",
routeTemplate: "api/{controller}/{action}/{id}"
);
When I make a call via ajax with the route:
/api/account/GetSUID/0
everything works fine. When I just use a different controller:
/api/tile/GetTileSet/0
it returns a status of 200 but instead of hitting my controller, it just returns the contents of the default page in jqXHR.responseText! It is as if it is just skipping any API routing like I am requesting the default site page.
I am baffled by this one as I have written literally hundreds of web API functions over the past few years in several other projects. I have never had any issue making calls to multiple controllers. I have looked high and low for what could be happening here and am hoping that someone here might have an idea.
Here is a sample method on the controller:
[HttpGet]
public HttpResponseMessage CheckRequestedID(int id, [FromUri]string Search)
{
if (!BSDIUtil.HasAllAcceptableCharacters(Search))
return Request.CreateResponse(HttpStatusCode.BadRequest);
if (FolderModel.IDAlreadyExists(DAL, Search)) // We can check this because this function is only called when staff members are creating accounts for other people (participants always use their email).
return Request.CreateResponse(HttpStatusCode.OK, false);
else
return Request.CreateResponse(HttpStatusCode.OK, true);
}
This will work if on the account controller but not on the tile controller.
One other thing, I am using the "Community" edition of Visual Studio and Windows 8.1
This is not a problem that is likely to occur often but I have solved it and figured I would post it here in case anyone else has the same issue.
I am using web api in the context of a standard web forms app (although I am only using webforms for my reporting pages). In the web.config for a web forms app, you can declare the paths that the user has access to before authenticating. I was only providing access to the account controller: all others were not permitted due to my authentication mechanism. Once I authenticate (e.g. the forms authentication call) or if I change the location path to include only "api", the problem goes away.
I was facing same problem but in different context. Had many controllers and respective routing templates. Only one controller was responding to requests.
Later i realized my other controller classes were not public!!
Just starting off with Breeze and Azure. I downloaded the sample app and added my own controllers. On my local machine it works just fine. When I deploy to my Azure site, I keep getting the following error in the logging section of the page: Query failed: Metadata query failed for: api/CoreData/Metadata; An error has occurred.
I have a controller named Products that simply returns a view. The view contains the js calls to my CoreData ApiController. Follows the sample.
The only thing logged on the server is the following:
At first I was using EF to access data, but have changed to just creating an in-memory object and returning it so as to eliminate db issues.
The URL is http://acapella.azurewebsites.net/products. I've spent quite a bit of time trying to diagnose and research, but feel that I'm overlooking something simple.
Any help or suggestions would be appreciated.
An Azure web site is a free, relatively easy way to deploy a Breeze application where you can test it on a variety of mobile devices and experience your app's responsiveness (or lack thereof).
I have had good success pushing a Breeze app to an Azure web site but there were many stumbles including some that looked like yours.
Two thoughts. First, you'll want to turn "Custom errors off" in your Web.config.
<configuration>
<system.web>
<customErrors mode="Off"/>
</system.web>
</configuration>
My second recommendation: take a look at John Papa's post on deploying an MVC Web API app to an Azure web site. It is excellent. And I know it works.
I hope to have a video on the process very soon.
The first thing you can do is return the exception. In your Breeze controller do this:
[HttpGet]
public string Metadata()
{
try
{
return _contextProvider.Metadata();
}
catch (Exception e)
{
return e.Message + Environment.NewLine + Environment.NewLine + e.StackTrace;
}
}
That made it clear to me something was wrong with my connection string. The exception I got was:
Format of the initialization string does not conform to specification starting at index 0.
My connection string locally seemed ok, but the problem was on Azure. My web.release.config had a correct transform, but when publishing, Visual Studio would add another connection string. You can check this by connecting with FTP to your Azure site and looking at your web.config. I fixed it by selecting the correct connection string when publishing:
After that, there still seem to be two connection strings in my uploaded web.config, but at least it works.