I want to terminate a httpwebrequest when it takes too long time in connection.
Here is just a simaple code that I wrote:
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(URL);
request.Timeout = 5000;
request.ReadWriteTimeout = 5000;
request.Proxy = new WebProxy("http://" + proxyUsed + "/", true);
request.UserAgent = "Mozilla/4.0 (compatible; MSIE 6.01; Windows NT 5.0)";
using (WebResponse myResponse = request.GetResponse())
{
using (Stream s = myResponse.GetResponseStream())
{
s.ReadTimeout = 5000;
s.WriteTimeout = 5000;
using (StreamReader sr = new StreamReader(s, System.Text.Encoding.UTF8))
{
result = sr.ReadToEnd();
httpLink = myResponse.ResponseUri.AbsoluteUri;
sr.Close();
}
s.Close();
}
myResponse.Close();
}
However, sometimes the connection will take a about 15minutes to get the response.
The situation is after 15minutes I still can get the response but not the full source code of the URL.
I guess that the connection is too slow that the URL response me a bit data within the timeout, just say for example receive 1byte in 5seconds, so it doesn't expire the timoue but it's very long.
How can I terminate the connection?
Thanks:)
You might find that the timeout is actually working, but the thread hangs when trying to close the stream. I don't know why it happens, but sometimes it does. I've never used ReadToEnd
, but I've run across this when using Read
.
I fixed the problem by calling Abort
on the request, before I close the stream. It's a bit of a kluge, but it's effective. The abbreviated code below shows the technique.
HttpWebResponse response = null;
StreamReader sr = null;
try
{
response = (HttpWebResponse)request.GetResponse(...);
Stream s = response.GetResponseStream();
sr = new StreamReader(s, Encoding.UTF8);
// do your reading here
}
finally
{
request.Abort(); // !! Yes, abort the request
if (sr != null)
sr.Dispose();
if (response != null)
response.Close();
}
What I've found is that the ReadTimeout
and ReadWriteTimeout
work as expected. That is, when the read times out, execution really does go to the finally
block. And if the request.Abort
isn't there, the call to sr.Dispose
will hang.
Break down the stream reading, and abort if the total time has been too long.
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(URL);
request.Timeout = 5000;
request.ReadWriteTimeout = 5000;
request.Proxy = new WebProxy("http://" + proxyUsed + "/", true);
request.UserAgent = "Mozilla/4.0 (compatible; MSIE 6.01; Windows NT 5.0)";//ahem! :)
DateTime giveUp = DateTime.UtcNow.AddSeconds(5);
using (WebResponse myResponse = request.GetResponse())
{
httpLink = myResponse.ResponseUri.AbsoluteUri;
using (Stream s = myResponse.GetResponseStream())
{
s.ReadTimeout = 5000;
s.WriteTimeout = 5000;
char[] buffer = new char[4096];
StringBuilder sb = new StringBuilder()
using (StreamReader sr = new StreamReader(s, System.Text.Encoding.UTF8))
{
for(int read = sr.Read(buffer, 0, 4096); read != 0; read = sr.Read(buffer, 0, 4096))
{
if(DateTime.UtcNow > giveUp)
throw new TimeoutException();
sb.Append(buffer, 0, read);
}
result = sb.ToString();
}
}
}
Make sure to also close the connection on exception. In that case the WebResponse object is on the WebException:
try
{
using (WebResponse myResponse = request.GetResponse())
// do stuff
}
catch (WebException webEx)
{
webEx.Response.Close();
}