Different performance between Java and c# code when testing URL

Question

When running the following Java code, I get very accurate and consistent results in determining if the web page I'm testing is up.

protected synchronized boolean checkUrl(HttpURLConnection connection){
 boolean error = false;
 //HttpURLConnection connection = null;
 GregorianCalendar calendar = new GregorianCalendar();

 try{
  if(connection != null){
   connection.connect();

   //200 is the expected HTTP_OK response
   error = processResponseCode(connection.getResponseCode());

   connection.disconnect();
  } else{
   error = false;
  }

 }catch(java.net.UnknownHostException uhe){
  ...  } 
 catch(Exception e){
  ...  }

 return error;
}

The closest match to the Java pattern in c# has much higher results of false positives (mostly due to timeouts - which has a default period of 100000ms).

protected bool connectedToUrl = false;
        response = null;

        HttpWebRequest webreq = (HttpWebRequest)WebRequest.Create(this.getUri());
        webreq.Credentials = CredentialCache.DefaultCredentials;
        WebResponse res = null;// webreq.GetResponse();

        try
        {
            WebRequest request = WebRequest.Create(this.getUri()) as WebRequest;
            request.Credentials = CredentialCache.DefaultCredentials;

            if (request != null)
            {
                // Get response 
                res = webreq.GetResponse();

                connectedToUrl = processResponseCode(res);
            }
            else
            {
                logger.Fatal(getFatalMessage());

                string error = string.Empty;
            }
        }
        catch (Exception e)
        {
            throw e;
        }

        return connectedToUrl;
    }

I have tried various patterns in c# to match the effectiveness of the quoted Java code, to no avail.

Any ideas?

Answer 1

I believe this is because you're not closing any of the request objects.

Answer 2

I think you're missing the GregorianCalendar in the C# version :-)

Why do you have two Request Objects in the C# version?

Answer 3

Also this:

   catch (Exception e)
   {
      throw e;
   }

Does nothing but destroy the stack trace on an exception that's been bubbled upwards. If you have error handling elsewhere in your code I suggest removing the try catch block. Otherwise you should log the exception and move on. Don't just catch it to throw it.

Answer 4

Simply change this:

res = webreq.GetResponse();
connectedToUrl = processResponseCode(res);

to

using (WebResponse res = webreq.GetResponse()) 
{
    connectedToUrl = processResponseCode(res);
}

(Remove the declaration from earlier.)

Until you haven't closed/disposed the response (or it's been finalized), it's holding onto the connection. You can only have a certain number (2 by default, I believe) of connections to any one host at a time, hence the timeouts. When you dispose the response, it allows another request to use the same connection.