Azure 服务总线配对命名空间 - 模拟故障转移

Question

I am working with azure service bus paired namespace and need to be able to simulate a failover to the secondary namespace. I did kind of have this working by entering an incorrect connection string for the primary namespace and saw it fail over and send the message to the secondary namespace. This no longer seems to do the trick. I can not find a way through the azure management portal or anywhere else to take a namespace offline. Anyone any ideas how to do this?

Here is my code for reference

var pairedNamespaceConfiguration = this.pairedNamespaceConfigurationDictionary[configurationKey];
MessagingFactory factory = MessagingFactory.CreateFromConnectionString(pairedNamespaceConfiguration.PrimaryNamespace.ConnectionString);
MessagingFactory secondaryMessagingFactory = MessagingFactory.CreateFromConnectionString(pairedNamespaceConfiguration.SecondaryNamespace.ConnectionString);
NamespaceManager secondaryNamespaceManager = NamespaceManager.CreateFromConnectionString(pairedNamespaceConfiguration.SecondaryNamespace.ConnectionString);

SendAvailabilityPairedNamespaceOptions sendAvailabilityOptions = new SendAvailabilityPairedNamespaceOptions(secondaryNamespaceManager, secondaryMessagingFactory, pairedNamespaceConfiguration.BacklogQueueCount, TimeSpan.FromSeconds(pairedNamespaceConfiguration.FailoverIntervalSeconds), false);
factory.PairNamespaceAsync(sendAvailabilityOptions).Wait();
MessageSender messageSender = factory.CreateMessageSender(pairedNamespaceConfiguration.PathName);
string messageContent = JsonConvert.SerializeObject(message);

using(BrokeredMessage brokeredMessage = new BrokeredMessage(messageContent))
{
    messageSender.Send(brokeredMessage);
}

Answer 1

Modify your \\Windows\\system32\\drivers\\etc\\hosts file to point the original namespace to something like 127.0.0.1. This will make the original namespace connection fail.

Answer 2

I'm using this example Geo-replication with Service Bus Relayed Messages to implement the same think. Maybe it's useful for you also.

All Service Bus entities reside in a namespace. A namespace is affiliated to a datacenter. To allow for a failover between datacenters, the user must create one Service Bus and ACS namespace (in case ACS is used) per datacenter. Any Service Bus relay that needs to remain accessible in the presence of datacenter failures must be created in both namespaces.

The server opens two NetTcp relay endpoints, one in each of the two namespaces. The server processes any request that is received via one of these endpoints. Note that the two relays have to have different names (.eg, address of primary relay is sb://myPrimaryNamespace.servicebus.windows.net/myService-primary and b://mySecondaryNamespace.servicebus.windows.net/myService-secondary).

The client considers one of the two replicated relays as the active relay and the other one as a backup. It opens a channel to the active relay and invokes methods on the service. If the invocation fails with any exception that is not part of the service contract, the client abandons the channel, opens a channel to the backup relay, and invokes the service method again. The client will consider the new channel to be the active channel and continues to use that channel until the next fault occurs.