Cannot connect to a WCF service hosted in a Worker Role - wcf

Question

I've created a WCF service and hosted it in cloud through a worker role. Unfortunately when I try to connect to the worker role service I get an exception with the message:
"No DNS entries exist for host 3a5c0cdffcf04d069dbced5e590bca70.cloudapp.net."
3a5c0cdffcf04d069dbced5e590bca70.cloudapp.net is the address for the worker role deployed in azure staging environment.
The workerrole.cs has the following code to expose the WCF service:
public override void Run()
{
using (ServiceHost host = new ServiceHost(typeof(MyService)))
{
string ip = RoleEnvironment.CurrentRoleInstance.InstanceEndpoints["tcppoint"].IPEndpoint.Address.ToString();
int tcpport = RoleEnvironment.CurrentRoleInstance.InstanceEndpoints["tcppoint"].IPEndpoint.Port;
int mexport = RoleEnvironment.CurrentRoleInstance.InstanceEndpoints["mexinput"].IPEndpoint.Port;
// Add a metadatabehavior for client proxy generation
// The metadata is exposed via net.tcp
ServiceMetadataBehavior metadatabehavior = new ServiceMetadataBehavior();
host.Description.Behaviors.Add(metadatabehavior);
Binding mexBinding = MetadataExchangeBindings.CreateMexTcpBinding();
string mexlistenurl = string.Format("net.tcp://{0}:{1}/MyServiceMetaDataEndpoint", ip, mexport);
string mexendpointurl = string.Format("net.tcp://{0}:{1}/MyServiceMetaDataEndpoint", RoleEnvironment.GetConfigurationSettingValue("Domain"), 8001);
host.AddServiceEndpoint(typeof(IMetadataExchange), mexBinding, mexendpointurl, new Uri(mexlistenurl));
// Add the endpoint for MyService
string listenurl = string.Format("net.tcp://{0}:{1}/MyServiceEndpoint", ip, tcpport);
string endpointurl = string.Format("net.tcp://{0}:{1}/MyServiceEndpoint", RoleEnvironment.GetConfigurationSettingValue("Domain"), 9001);
host.AddServiceEndpoint(typeof(IMyService), new NetTcpBinding(SecurityMode.None), endpointurl, new Uri(listenurl));
host.Open();
while (true)
{
Thread.Sleep(100000);
Trace.WriteLine("Working", "Information");
}
}
}
The tcppoint and mexinput are configured with the ports 8001 and 9001. Also Domain is configured with worker role deployment url:3a5c0cdffcf04d069dbced5e590bca70.cloudapp.net
On the client part(a console app), we are using the following configuration in app.config::
<bindings>
<netTcpBinding>
<binding name="NetTcpBinding_IMyService" closeTimeout="00:01:00"
openTimeout="00:01:00" receiveTimeout="00:10:00" sendTimeout="00:01:00"
transactionFlow="false" transferMode="Buffered" transactionProtocol="OleTransactions"
hostNameComparisonMode="StrongWildcard" listenBacklog="10"
maxBufferPoolSize="524288" maxBufferSize="65536" maxConnections="10"
maxReceivedMessageSize="65536">
<readerQuotas maxDepth="32" maxStringContentLength="8192" maxArrayLength="16384"
maxBytesPerRead="4096" maxNameTableCharCount="16384" />
<reliableSession ordered="true" inactivityTimeout="00:50:00"
enabled="false" />
<security mode="None">
<transport clientCredentialType="Windows" protectionLevel="EncryptAndSign" />
<message clientCredentialType="Windows" />
</security>
</binding>
</netTcpBinding>
</bindings>
<client>
<endpoint address="httpp:\\3a5c0cdffcf04d069dbced5e590bca70.cloudapp.net:9001/MyServiceEndpoint" binding="netTcpBinding"
bindingConfiguration="NetTcpBinding_IMyService" contract="ServiceReference1.IMyService"
name="NetTcpBinding_IMyService" />
</client>
<behaviors>
<serviceBehaviors>
<behavior name="behave">
<serviceMetadata httpGetEnabled="false"/>
</behavior>
</serviceBehaviors>
</behaviors>
</system.serviceModel>
<system.net>
<defaultProxy useDefaultCredentials="true">
<proxy autoDetect="False" usesystemdefault="False" bypassonlocal="True" />
</defaultProxy>
The following code is built using the sample code available in msdn as background. Locally it is working fine. Unfortunately when i deploy it to cloud, the exception occurs. Moreover, when i use the virtual ip instead of the url, a connection time out occurs with the exception the remote machine did not respond.

Answer 1

Looks like you have your service setup to listen on net.tcp (TCP) and your client using http bindings. I would not expect that to work even locally. I am assuming you have actually opened port 9000 in the ServiceDefinition. Remember that will be a load-balanced endpoint. Are you trying to communicate to this instance from within the deployment (inter-role) or from outside the cloud?
I have found it is a lot easier to setup the host and client (when communicating within a role) through code. Try this:
http://dunnry.com/blog/2010/05/28/HostingWCFInWindowsAzure.aspx
If you are trying to hit the service from a client outside the deployment, this still applies, but for the client building part. You will need to use the external DNS name and port defined in ServiceDefinition.
I have also seen DNS errors if you try to hit the endpoint too soon before the role was ready. It can take a bit to propogate the DNS and you should try not to resolve it until it is ready, lest you cache a bogus DNS entry. If you can resolve that DNS name however to your VIP address, that is not the issue.

Answer 2

public void CallWebService(string data)
{
try
{
string uri = "url"+data;
HttpWebRequest request = (HttpWebRequest)HttpWebRequest.Create(uri);
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
Stream str = response.GetResponseStream();
StreamReader sr = new StreamReader(str);
String IResponse = sr.ReadToEnd();
}
catch (Exception ex)
{
System.Console.WriteLine("Message: "+ex.Message);
}
}
Hope it helps you.