Question

I develop an asmx web service (i.e. ASP.NET 2.0).

There’s a piece of code that may read the contents of the HTTP request (via HttpContext.Current.Request.InputStream) while processing it. I realise that InputStream may only be read once for a request, and I make sure I never try to read it more than once.

The problem seems to be that if InputStream happens to be read during the early stages of the application’s lifecycle (e.g. after pskill w3wp, during Application_Start), the HTTP request fails with a HTTP 400 – Bad Request error, with no explanation given, no exception thrown and no entry in the httperr log. If it is read later (e.g. within the web method itself), requests run fine whether InputStream is read or not. Application_Start runs fine if InputStream isn’t read.

Is this some sort of ASP.NET bug? IIS bug? Or am I doing something wrong by daring to read InputStream? And if so, is there another way to get a look at the “raw” contents of the request without disturbing the inner workings of IIS/ASP.NET?

In short, adding this code within Application_Start is enough to reproduce this error:

using (StreamReader reader = new StreamReader(HttpContext.Current.Request.InputStream))
    reader.ReadToEnd();

Answer 1

Couldn’t find a way to read request contents during Application_Start without disturbing inner workings of ASP.NET/IIS. Instead, ended up making sure this doesn’t happen until Application_Start is over (and also doesn’t happen from the moment Application_End starts, which also turned out to be problematic and created access violations).