Why would WCF fail to call a SOAP service when a 302 response is encountered?

问题

I have written an application that starts with making a WCF call to login. I generated the client code with a service reference. It works fine for clients who have their services installed locally to their network. However, there is a saas environment, as well, where these same services are controlled by the corporate powers that be. In the saas environment, I was informed that the login was failing. Investigating using Fiddler, I found that the service call to login is returning HTML, specifically, the web page listing all the available methods from the .asmx.

The saas environment has one little quirk which may be causing the problem here, but I don't know how to verify that this is the problem, nor how to solve it if it is the problem. The quirk is that the server redirects (302) the call.

The client code:

    client.Endpoint.Address = new EndpointAddress("http://" + settings.MyURL + "/myProduct/webservices/webapi.asmx");
    client.DoLogin(username, password);

The raw data sent to the server, before the redirect, includes the s:Envelope XML tag. Notice the missing s:Envelope XML tag when sending to the redirected server:

    GET https://www.myurl.com/myProduct/webservices/webapi.asmx HTTP/1.1
    Content-Type: text/xml; charset=utf-8
    VsDebuggerCausalityData: uIDPo7TgjY1gCLFLu6UXF8SWAoEAAAAAQxHTAupeAkWz2p2l3jFASiUPHh+L/1xNpTd0YqI2o+wACQAA
    SOAPAction: "http://Interfaces.myProduct.myCompany.com/DoLogin"
Accept-Encoding: gzip, deflate
    Host: www.gotimeforce2.com
    Connection: Keep-Alive

How do I get this silly thing working?

Edit: It is worth noting that I am using WCF/svcutil.exe/service-reference rather than the older ASMX/wsdl.exe/web-reference. Otherwise, for future readers of this topic, the wsdl solution suggested by Raj would have been a great solution. If you are seeing this issue and are using the wsdl technique, see Raj's excellent answer.

Edit2: After doing a bunch of research into WCF and 302, it sounds like they just don't play well together, nor does there appear to be a simple way of giving the WCF api custom code to handle the situation. As I have no control over the server, I have sucked it up and re-generated my api as a web-reference and am using Raj's solution.

Edit3: Updated the title to better reflect the solution, now that the cause of the issue is understood. Original title: Why would WCF not include s:Envelope on a redirect?

回答1:

Ok, So I did some digging on this and tried to replicate the issue on my side. I was able to replicate the issue and find a solution to it as well. However I'm not sure how well this will apply in your case since it is dependent on interfacing with the server team that manages the load balancer. Here are the findings.

Looking at http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html you notice the following addendum in the explanation for HTTP Status Codes 302 and 303.

302 Found

  Note: RFC 1945 and RFC 2068 specify that the client is not allowed
  to change the method on the redirected request.  However, most
  existing user agent implementations treat 302 as if it were a 303
  response, performing a GET on the Location field-value regardless
  of the original request method. The status codes 303 and 307 have
  been added for servers that wish to make unambiguously clear which
  kind of reaction is expected of the client.

303 See Other

  Note: Many pre-HTTP/1.1 user agents do not understand the 303
  status. When interoperability with such clients is a concern, the
  302 status code may be used instead, since most user agents react
  to a 302 response as described here for 303.

Further looking at http://en.wikipedia.org/wiki/List_of_HTTP_status_codes you notice the following explanation for the HTTP status codes 302, 303 and 307.

302 Found : This is an example of industry practice contradicting the standard. The HTTP/1.0 specification (RFC 1945) required the client to perform a temporary redirect (the original describing phrase was "Moved Temporarily"), but popular browsers implemented 302 with the functionality of a 303 See Other. Therefore, HTTP/1.1 added status codes 303 and 307 to distinguish between the two behaviors. However, some Web applications and frameworks use the 302 status code as if it were the 303.

303 See Other (since HTTP/1.1): The response to the request can be found under another URI using a GET method. When received in response to a POST (or PUT/DELETE), it should be assumed that the server has received the data and the redirect should be issued with a separate GET message. Here is the basic flow in a normal Client/Server Interaction

307 Temporary Redirect (since HTTP/1.1): In this case, the request should be repeated with another URI; however, future requests should still use the original URI. In contrast to how 302 was historically implemented, the request method is not allowed to be changed when reissuing the original request. For instance, a POST request should be repeated using another POST request.

So according to this, we are able to explain the behavior of the WCF call which sends a GET request without the s:Envelope on the 302 redirect. This will undoubtedly fail on the client side.

The easiest way of fixing this is to have the server return a 307 Temporary Redirect instead of a 302 Found status code in the response. Which is where you need the help of the Server Team that manages the redirect rules on the load balancer. I tested this out locally and the client code consuming the service with the Service Reference seamlessly executes the call even with the 307 Temporary Redirect.

In fact you could test this all out with the solution I've uploaded to Github Here. I've updated this to illustrate the utilization of a service reference instead of a wsdl generated proxy class to consume the asmx service.

However if the change from 302 Found to 307 Temporary Redirect is not feasible in your environment, then I would suggest using either Solution 1 (which shouldn't have a problem whether it is a 302 or 307 status code in the response) or using my original answer which would resolve this by directly accessing the service at the correct URL based on the setting in the config file. Hope this helps!

Solution 1

If you do not have access to the config files on production or if you just plain don't want to use the multiple URLs in the config file, you could use this following approach. Link to Github repo containing sample solution Click Here

Basically, if you notice the file auto generated by wsdl.exe you will notice that the service proxy class derives from System.Web.Services.Protocols.SoapHttpClientProtocol. This class has a protected method System.Net.WebRequest GetWebRequest(Uri uri) that you can override. In here you could add a method to check to see if a 302 temporary redirect is the result of HttpWebRequest.GetResponse() method. If so, you can set the Url to the new Url returned in the Location header of the response as follows.

this.Url = new Uri(uri, response.Headers["Location"]).ToString();

So create a class called SoapHttpClientProtocolWithRedirect as follows.

public class SoapHttpClientProtocolWithRedirect :
    System.Web.Services.Protocols.SoapHttpClientProtocol
{
    protected override System.Net.WebRequest GetWebRequest(Uri uri)
    {
        if (!_redirectFixed)
        {
            FixRedirect(new Uri(this.Url));
            _redirectFixed = true;

            return base.GetWebRequest(new Uri(this.Url));
        }

        return base.GetWebRequest(uri);
    }

    private bool _redirectFixed = false;
    private void FixRedirect(Uri uri)
    {
        var request = (HttpWebRequest)WebRequest.Create(uri);
        request.CookieContainer = new CookieContainer();
        request.AllowAutoRedirect = false;
        var response = (HttpWebResponse)request.GetResponse();

        switch (response.StatusCode)
        {
            case HttpStatusCode.Redirect:
            case HttpStatusCode.TemporaryRedirect:
            case HttpStatusCode.MovedPermanently:
                this.Url = new Uri(uri, response.Headers["Location"]).ToString();
                break;
        }
    }
}

Now comes the part that illustrates the advantage of using a proxy class manually generated using wsdl.exe instead of a service reference. In the manually created proxy class. modify the class declaration from

public partial class WebApiProxy : System.Web.Services.Protocols.SoapHttpClientProtocol

to

public partial class WebApiProxy : SoapHttpClientProtocolWithRedirect

Now invoke the DoLogin method as follows.

var apiClient = new WebApiProxy(GetServiceUrl());
//TODO: Add any required headers etc.
apiClient.DoLogin(username,password);

You will notice that the 302 redirect is handled smoothly by the code in your SoapHttpClientProtocolWithRedirect class.

One other advantage is that, by doing this, you will not have to fear that some other developer is going to refresh the service reference and lose the changes that you made to the proxy class since you had manually generated it. Hope this helps.

Original Answer

Why don't you just include the entire url for production/local service in the config file? That way you can initiate the call with the appropriate url in the appropriate location.

Also, I would refrain from using a service reference in any code destined for production. One way of using the asmx service without a service reference would be to generate the WebApiProxy.cs file using the wsdl.exe tool. Now you can just include the WebApiProxy.cs file in your project and instantiate as shown below.

var apiClient = new WebApiProxy(GetServiceUrl());
//TODO: Add any required headers etc.
apiClient.DoLogin(username,password);

Here is the GetServiceUrl() method. Use a Configuration Repository to further decouple and improve testability.

private string GetServiceUrl()
    {
        try
        {
            return
            _configurationRepository.AppSettings[
                _configurationRepository.AppSettings["WebApiInstanceToUse"]];
        }
        catch (NullReferenceException ex)
        {
            //TODO: Log error
            return string.Empty;
        }
    }

Then your config file can contain the following information in the section.

<add key="StagingWebApiInstance" value="http://mystagingserver/myProduct/webservices/webapi.asmx "/>
<add key="ProductionWebApiInstance" value="https://www.myurl.com/myProduct/webservices/webapi.asmx"/>
<!-- Identify which WebApi.asmx instance to Use-->
<add key="WebApiInstanceToUse" value="ProductionWebApiInstance"/>

Also I would refrain from concatenating strings using the + overload. When doing it once, it doesn't come across as too much of a performance impact but if you have many concatenations like this throughout the code, it would lead to a big difference in execution times compared to using a StringBuilder. Check http://msdn.microsoft.com/en-us/library/ms228504.aspx for more information on why using a StringBuilder improves performance.

来源：https://stackoverflow.com/questions/17152385/why-would-wcf-fail-to-call-a-soap-service-when-a-302-response-is-encountered