WCF Callback Service with netTcp Binding timeout after 10 mins - wcf

Question

I'm creating a chat application with WCF(using callback contract) and netTcpBinding.
I'm hosting the service as a windows service and accessing it from other computers
via the client application.
The problem that i'm facing now is the clients connection comes to a Fault state after
10 mins which seems to be some kind of timeout that occur.
I already tried increasing the received timeout and send timeout in both service and client but didn't work.
which setting should i change to increase this timeout period and in which application, service or client?
Following are my configuration files,
Service
<system.serviceModel>
<services>
<service behaviorConfiguration="PeerTalk.Service.ChatServiceBehavior"
name="PeerTalk.Service.ChatService">
<endpoint address="" binding="netTcpBinding" bindingConfiguration=""
contract="PeerTalk.Service.ServiceContracts.IChat">
<identity>
<dns value="localhost" />
</identity>
</endpoint>
<endpoint address="mex" binding="mexTcpBinding" bindingConfiguration=""
contract="IMetadataExchange" />
<host>
<baseAddresses>
<add baseAddress="http://localhost:7920/ChatService" />
<add baseAddress="net.tcp://localhost:7921/ChatService" />
</baseAddresses>
</host>
</service>
</services>
<behaviors>
<serviceBehaviors>
<behavior name="PeerTalk.Service.ChatServiceBehavior">
<serviceMetadata httpGetEnabled="false" />
<serviceDebug includeExceptionDetailInFaults="false" />
</behavior>
</serviceBehaviors>
</behaviors>
<bindings>
<netTcpBinding>
<binding name="tcpBinding"
maxBufferSize="67108864"
maxReceivedMessageSize="67108864"
maxBufferPoolSize="67108864"
transferMode="Buffered"
closeTimeout="00:01:00"
openTimeout="00:01:00"
receiveTimeout="00:00:10"
sendTimeout="00:00:10"
maxConnections="100">
<readerQuotas maxDepth="64"
maxStringContentLength="67108864"
maxArrayLength="67108864"
maxBytesPerRead="67108864"
maxNameTableCharCount="16384"/>
<security mode="Transport">
<transport clientCredentialType="Windows" protectionLevel="EncryptAndSign" />
<message clientCredentialType="Windows"/>
</security>
<reliableSession enabled="false" inactivityTimeout="00:01:00"/>
</binding>
</netTcpBinding>
</bindings>
</system.serviceModel>
Client
<system.serviceModel>
<bindings>
<netTcpBinding>
<binding name="NetTcpBinding_IChat" closeTimeout="00:01:00" openTimeout="00:01:00"
receiveTimeout="00:10:00" sendTimeout="00:00:10" transactionFlow="false"
transferMode="Buffered" transactionProtocol="OleTransactions"
hostNameComparisonMode="StrongWildcard" listenBacklog="10" maxBufferPoolSize="67108864"
maxBufferSize="67108864" maxConnections="10" maxReceivedMessageSize="67108864">
<readerQuotas maxDepth="32" maxStringContentLength="67108864"
maxArrayLength="67108864" maxBytesPerRead="67108864" maxNameTableCharCount="16384" />
<reliableSession ordered="true" inactivityTimeout="00:01:00"
enabled="false" />
<security mode="Transport">
<transport clientCredentialType="Windows" protectionLevel="EncryptAndSign" />
<message clientCredentialType="Windows" />
</security>
</binding>
</netTcpBinding>
</bindings>
<client>
<endpoint address="net.tcp://10.10.10.45:7921/ChatService" binding="netTcpBinding"
bindingConfiguration="NetTcpBinding_IChat" contract="PeerTalkService.IChat"
name="NetTcpBinding_IChat">
</endpoint>
</client>
</system.serviceModel>
Thanks.

Answer 1

The timeout in this case is defined by both receiveTimeout in the binding and inactivityTimeout in reliable session which is used for duplex messaging. The correct solution is not increasing timeout but implementing some ping / keep alive messages. The reason is that increasing timeout will keep connections open for failed clients.

Answer 2

Can you post client call sample (service call example). What might happening here is that you are not closing client correctly and you reach maximum sessions on service side.
You must be aware that using net.tcp binding is different than http.
You can use System.ServiceModel performance counters (http://msdn.microsoft.com/en-us/library/ms750527.aspx) and see after 10 minutes what is happening (number of calls outstanding, number of service instances, etc..)
http://dkochnev.blogspot.com/2011/06/wcf-framework-40-monitoring-service.html