I have a very simple client/server solution that uses a wellknown singleton and a TcpChannel. Basically just copy/paste from the docs.

I noticed that when a client exists, the server throws twice an exception System.Net.Sockets.SocketException. I guess this is because it keeps a proxy to a client that does not exist anymore. These exceptions do not cause a crash, they are trapped inside remoting I think. They only appear in the debug output window.

This issue has been posted before, but no solution has been given.

So basically, if a client does Activator.GetObject(), how should it "disconnect" from the server so that the latter does not throw these exceptions? Personally I think this should automatically be handled when the proxy is garbage collected.

We are having the same problem. Does anyone have an answer to this thread?

// unregister the channel

ChannelServices.UnregisterChannel(this.clientChannel);

// if you have a delegate event handler for communication purposes:

_CallbackSink.OnHostToClient -= callbackSinkOnClient;

// set the object that hosted the remoting object to null.

We are having the same problem. Does anyone have an answer to this thread?

The OP said that he was getting this exception printed to the debug console but wasnt causing any negative side effects, is that true for you as well? If so, what would you like us to solve here? It sounds like its working correctly in this case.

We are having the same problem. Does anyone have an answer to this thread?

// unregister the channel

ChannelServices.UnregisterChannel(this.clientChannel);

// if you have a delegate event handler for communication purposes:

_CallbackSink.OnHostToClient -= callbackSinkOnClient;

// set the object that hosted the remoting object to null.

Above solution not working for me for any of the given MS samples.


I guess what i'm after is a correct understanding of whats happening.

In my server's trace viewer in debug mode i see 2: System.Net.Sockets.SocketException instances.

This always happens when the client disconnects after invoking instances of remotable objects (Register client activated type )

Should I be worried about this log msg or is it what .NET is expected to do? i.e, it deliberately uses.NETexception handling to detect when a client has shutdown/discarded its connection resources.

I did enable the handling of first chance exception handling for the above exception type and managed to gleam one line of additional debug trace - "The client or remote hostwas forcibly disconnected" or something like that. That sounds like the client threw away its connection and was very naughty, but not to worry, .NET detected this and handled it nicely for you.

Some clarrification on this issue would be nice ( if there is any ). I would certainly prefer not to encounter any first chance exceptions in this scenarioby having the ability to cleanly shutdown a remoted object's connection.