Develop Reference

c# Socket send receive network

I'm totally confused and stunned about my problem. I wrote a multi client server using sockets.
I created a first GUI that receives clients and display them in a gridview, in the click gridview event it opens a new GUI with the socket between the clicked client and the server. So everything about the connection between server and the clients go well and fast
But my problems are :
the send information between them like sending full process it sometimes shown full and it sometimes less result .
send files management for example when the server request full folders/files in a directory sometimes it shows them all and sometimes less result .
But the command such open a window open an url, send a message , commands like those works so perfectly and instantly .
I am so confused right now what's the problem
Notice 1 : I used the extern IP for connection between server/client.
Notice 2 : internet connection is perfect (definitely not slow)
I tried to use different buffer sizes, but what I'm really confused about is that sometimes the result comes full with a specific buffer size and sometimes not with the same buffer size .
Thank you for your time !

I would also recommend you use wireshark so that you are able to get the full information and transactions that are really going on in the network.
This will help you to rule out where the issue is coming from.
To determine the packet size see the image
Here I am not talking about how the TCP handshaking protocol works as I assume it is working. However do note that in the transmission if you see something like RST then maybe somebody issued a reset command implying that some error occurred in the transmission. Such as something due to checksum for example. Although it is generally a problem for someone working directly on TCP/IP protocol
This should help you confirm that you sent and recieved the correct length.

telnet - how to know when to give control back to the user

To begin with, why I've written a Telnet client is not important. It's my job and I have to do it. So, on to my question...
My Telnet client is simple and allows for both interactive communication (used by a human) as well as automated communication (used by a program). However, in order for my client to know when the server is done sending data so it can give control back to the user, I've had to hard code all of the possible command prompts for a given server into the client application. For instance, I create a list of prompts and then pass this list to the client so it can watch for these in the data stream from the server. When it sees that a data stream ends with one, it then gives control back to the user, who will enter a response or a command and send it. The client then watches again for any of these prompts and so on. The list could look like the following for a Telnet server running on CentOS:
private static void SetTerminalPrompts()
{
prompts.Add("login: ");
prompts.Add("Password: ");
prompts.Add("]$ ");
prompts.Add("]# ");
}
This would be used like:
response = telnetClient.ReceiveUntilPrompt(prompts);
And public string ReceiveUntilPrompt(List<string> prompts) simply reads the data stream watching for a prompt from the list.
The problem is watching for these prompts. In a general case, the Telnet client will not know the server to which it is connecting. And it seems silly to attempt to create a complete list of possible prompts by guessing all servers to which the client may ever connect (even though I've found several examples and posts that say just that).
What I would like to know is how do the Telnet clients that come with Windows or Linux know when to give control back to the user? How do they know when the Telnet server is done sending data?

C# An established connection was aborted by the software in your host machine

These errors are getting more and more frequent on my Game Server. They are causing the server to keep closing and restarting...
System.Net.Sockets.SocketException (0x80004005): An established connection was aborted by the software in your host machine
at System.Net.Sockets.Socket.BeginSend(Byte[] buffer, Int32 offset, Int32 size, SocketFlags socketFlags, AsyncCallback callback, Object state)
at iRP.Game.Sessions.Session.SendData(Byte[] Data)
This is the code from which these errors are generated:
public void SendData(byte[] Data)
{
try
{
if (mSocket == null)
{
//Output.WriteLine("[SND] Socket has a null exception, which means it is now invalid. Remove this socket!", OutputLevel.CriticalError);
}
else
{
mSocket.BeginSend(Data, 0, Data.Length, SocketFlags.None, sendCallback, mSocket);
}
}
catch (Exception e)
{
string WhatToWrite = "Error handled (SESSION): " + e.ToString() + "\n\n" + e.Message + "\n\nStack: " + e.StackTrace + Environment.NewLine + "\n\n";
File.AppendAllText(Environment.CurrentDirectory + "\\data\\fatal.txt", WhatToWrite);
Program.Stop();
}
}
The buffer sizes are correctly set, we are using KeepAlive on the socket and were using Send and Receive Timeouts.
People suggested that disabling the firewall would help, but whenever I do this our Game Server (Dedicated Server) restarts itself as if it's under attack, so the firewall must remain enabled.
Anyone else got any other solutions for this?
PS: We are behind DDoS Mitigation Services which may be limiting the number of connections...

An established connection was aborted by the software in your host machine
That is a boiler-plate error message, it comes out of Windows. The underlying error code is WSAECONNABORTED. Which really doesn't mean more than "connection was aborted". You have to be a bit careful about the "your host machine" part of the phrase. In the vast majority of Windows application programs, it is indeed the host that the desktop app is connected to that aborted the connection. Usually a server somewhere else.
The roles are reversed however when you implement your own server. Now you need to read the error message as "aborted by the application at the other end of the wire". Which is of course not uncommon when you implement a server, client programs that use your server are not unlikely to abort a connection for whatever reason. It can mean that a fire-wall or a proxy terminated the connection but that's not very likely since they typically would not allow the connection to be established in the first place.
You don't really know why a connection was aborted unless you have insight what is going on at the other end of the wire. That's of course hard to come by. If your server is reachable through the Internet then don't discount the possibility that you are being probed by a port scanner. Or your customers, looking for a game cheat.

This problem appear if two software use same port for connecting to the server
try to close the port by cmd according to your operating system
then reboot your Android studio or your Eclipse or your Software.

Could be related to the maximum number of concurrent requests. I was able to fix it with two solutions:
Increase the default limit of the max concurrent connections (by default it is set to 2):
ServicePointManager.DefaultConnectionLimit = 25
Wrap sending requests around a buffer: could use ConcurrentQueue to limit the rate, or implement a simple wait as the following:
while (activeRequests >= maxConcurrentRequests)
Thread.Sleep(1000);
Interlocked.Increment(ref activeRequests);
var response = await _client.GetStreamAsync(endpoint);
Interlocked.Decrement(ref activeRequests);

While the answer from Hans is an excellent high level summary of the error that worked great for getting me started, I ended up finding a page that explained the root cause well enough that I was able to recreate it with a python script.
The page presents a couple different descriptions of this error that are more detailed than the "connection was aborted" paraphrasing from Hans's answer but still pretty cryptic and not very informative.
The post then explains this scenario that would produce the error:
An HTTP POST is to be sent to an HTTP server.
The server begins reading the POST and notices that the HTTP request header is invalid.
It immediately sends an HTTP response (with an error status, perhaps status=400) and closes the connection without trying to continue reading the remainder of the HTTP request that is forthcoming.
Meanwhile, the client is still happily writing the remainder of the HTTP request to the socket. (Remember a TCP/IP socket connection needs to be closed from both sides. In this case, the server has closed its side, but the client is still pumping data into the half-open connection.)
The client finishes writing the HTTP POST to the socket — meaning that data has been buffered to Winsock. The client application then tries to read the HTTP response, but it cannot because the outgoing retransmission (of the buffered data by WinSock) failed and the socket connection was shutdown on the client side (by Winsock). Even though the HTTP server sent the response, it is lost and cannot be retrieved. The error your application will receive when trying to read the HTTP response on the socket is WSAECONNABORTED
Wikipedia also has a page to explain what winsock is, but all you really need to know for this scenario is that its the Windows socket API.

I had "ManageEngine Agent" installed on my computer, which was blocking my connection to the databases, so uninstalling it was the solution.
Check which application could block the connection to your DB.

C# Cannot load XML document from server

I had an XMLDocument loading a document from a server with no problems till, almost randomly, I started getting a connection refused error.
It also doesn't matter what host I put in, whether it's a legit one or unresolvable. It always gives the same result.
Here's the code:
XmlDocument doc = new XmlDocument();
doc.Load("http://doesnotmatterifIresolveornot.com");
And here is the error:
{"No connection could be made because the target machine actively refused it 127.0.0.1:8888"}
I've turned off any applicable firewalls I can find in Win7, but it's weird cause it happened while I was testing it.

Find out why it's trying to go to 127.0.0.1:8888.
My guess is that for some reason, it thinks that's your HTTP proxy. Did you run something like Fiddler recently? Fiddler runs on 8888 and changes your default proxy settings - maybe they got stuck incorrectly?
Look in Control Panel, or in the Internet Explorer settings.

Are you serving your XML document using IIS? If so, you may need to add a mime-type definition to IIS to serve XML files. This article should help with that (if it is indeed the problem).
you may also try the HTTP loader to get a more detailed picture of what the server is responding with (HTTP headers, in particular, could be useful for troubleshooting).
I suspect the primary issue is that you're trying to connect to a socket (server + port) that the server isn't configured to listen on -- that means you'll get this error regardless of whether or not the URL resolves, since the server isn't configured to deal with a socket connection of the sort you're sending it.

Ping always succeeds, cant easily check uptime using ping on a url

Im getting frustrated because of OpenDNS and other services (ie: roadrunner) that now always returns a ping even if you type any invalid url ie: lkjsdaflkjdsjf.com --- I had created software for my own use that would ping a url to verify if the site was up or not. This no longer works. Does anyone have any ideas about this?
Requirements:
It should work with any valid web site, even ones i dont control
It should be able to run from any network that has internet access
I would greatly appreciate to hear how others now handle this. I would like to add, im attempting to do this using System.Net in c#
Thank you greatly :-)
New addition: Looking for a solution that i can either buy and run on my windows machine, or program in c#. :-)
Update:
Thank you all very much for your answers. Ultimately i ended up creating a solution by doing this:
Creating a simple webclient that downloaed the specified page from the url (may change to just headers or use this to notify of page changes)
Read in xml file that simply lists the full url to the site/pages to check
Created a windows service to host the solution so it would recover server restarts.
On error an email and text message is sent to defined list of recipients
Most values (interval, smtp, to, from, etc) are defined in the .config for easy change
I will be taking some of your advice to add 'features' to this later, which includes:
AJAX page for real-time monitoring. I will use WCF to connect to the existing windows service from the asp.net page
Download Headers only (with option for page change comparison)
make more configurable (ie: retries on failure before notification)

Wget is a nice alternative. It will check not only whether the machine is active, but also whether the HTTP server is accepting connections.

You could create a simple web page with an address bar for the website and some javascript that uses AJAX to hit a site. If you get any HTTP response other than 200 on the async callback, the site isn't working.
<html>
<head>
<script language="javascript" type="text/javascript">
<!--
var ajax = new XMLHttpRequest();
function pingSite() {
ajax.onreadystatechange = stateChanged;
ajax.open('GET', document.getElementById('siteToCheck').value, true);
ajax.send(null);
}
function stateChanged() {
if (ajax.readyState == 4) {
if (ajax.status == 200) {
document.getElementById('statusLabel').innerHTML = "Success!";
}
else {
document.getElementById('statusLabel').innerHTML = "Failure!";
}
}
}
-->
</script>
</head>
<body>
Site To Check:<br />
<input type="text" id="siteToCheck" /><input type="button" onclick="javascript:pingSite()" />
<p>
<span id="statusLabel"></span>
</p>
</body>
This code depends on the browser not being IE and I haven't tested it, but it should give you a really good idea.

To see if a service is up, not only should you ping, but it's good to have scripts that will hit a service, such as a website, and get back a valid response. I've used What's Up Gold in the past, rather than write my own. I like all the features in products like that. such as sending me a page when a service is down.

For the record, lkjsdaflkjdsjf.com is a hostname (which at the moment is not registered to anyone). ping does not work with URLs, ping works with hostnames. hostnames are looked up using the Domain Name System. DNS is supposed to fail when hostnames are not registered.
The problem is that some services (apparently your ISP, and definitely OpenDNS) do NOT fail DNS requests for hostnames that aren't registered. Instead they return the IP address of a host on their network that presents a search page to any http request.
You appear to want to know two things: Is the name real (that is, is there a host with this name registered to some actual machine)? and Is that machine functioning?
If you already know that the name in question is real (for instance, you want to know if www.google.com is up), then you can use ping because you know that the name will resolve to a real address (the ISP can't return their IP for a registered name) and you'll only be measuring whether that machine is in operation.
If you don't know whether the name is real, then the problem is harder because your ISP is returning false data to your DNS request. The ONLY solution here is to find a DNS server that is not going to lie to you about unresolved names.
The only way I can think of to differentiate between your ISP's fake records and real ones is to do a reverse lookup on the IP you get back and see if that IP is in your ISP's network. The problem with this trick is that if the ISP doesn't have reverse DNS info for that IP, you won't know whether it's the ISP or just some knucklehead who didn't configure his DNS properly (I've made that mistake many times in the past).
I'm sure there are other techniques, but the fact that DNS lies underneath everything makes it hard to deal with this problem.

Don't directly know of any off the shelf options in c#, although I'd be very suprised if there aren't a few available.
I wrote something similar a few years ago, don't have the code anymore cos it belongs to someone else, but the basic idea was using a WebClient to hit the domain default page, and check for a http status code of 200.
You can then wire any notification logic around the success or fail of this operation.
If bandwidth is a concern you can trim it back to just use a HEAD request.
For more complex sites that you control, you can wire up a health monitoring page that does some more in depth testing before it sends the response, eg is DB connection up etc.
Often a machine that is dead on port 80 will still respond to a ping, so testing port 80 (or whatever other one you are interested in) will be a much more reliable way to go.

I've found ping to be very unreliable just because of all the hops you're having to jump through, and something in between can always interfere.
Trying to open up an http connection with a web server is probably the best way to go.

You could try running 'httping' if you have cygwin available or
http://freshmeat.net/projects/httping/

As far as I can see, the problem here that OpenDNS resolves invalid domains back to themselves to forward you on to something close to what you're after (so if you typo ggooggllee.com you end up at the right place via a bounce from the OpenDNS servers). (correct me if I'm wrong)
If that's the case, you should just be able to check whether the IP you've resolved == any of the IPs of OpenDNS? No more ping - no protocol (HTTP) level stuff - just checking for the exceptional case?

If you tend toward the sys-admin solution rather than the programming solution you could install a local name server and tell it not to accept anything but NS records for delegation only zones. This was the fix I (and I assumed everyone else on the internet) used when Network Solution/Verisign broke this last time. I installed BIND on a couple of local machine, told my DHCP servers to hand out those addrs as the local name servers, and set up something like the following for each of the delegation only zones that I cared about:
zone "com" { type delegation-only; };
zone "net" { type delegation-only; };
Come to think of this, I think this might be turned on by default in later BIND versions. As an added bonus you tend to get more stable DNS resolution than most ISPs provide, you control your own cache, a little patching and you don't have to rely on your ISP to fix the latest DNS attack, etc.

i like CALM for this sort of thing as it logs to a database and provides email notifications as well as a status dashboard
you can set up a test page on the site and periodically do a GET on it to receive either a 'true' result (all is well) or an error message result that gets emailed to you
caveat: i am the author of CALM