Tag: ServiceBusClient

When working with Microsoft Azure Service Bus, you may encounter the following exception:

“Cannot allocate more handles. The maximum number of handles is 4999.”

This issue typically arises due to improper dependency injection scope configuration for the ServiceBusClient. In most cases, the ServiceBusClient is registered as Scoped instead of Singleton, leading to the creation of multiple instances during the application lifetime, which exhausts the available handles.

In this blog post, we’ll explore the root cause and demonstrate how to fix this issue by using proper dependency injection in .NET applications.

Understanding the Problem

Scoped vs. Singleton

1. Scoped: A new instance of the service is created per request.
2. Singleton: A single instance of the service is shared across the entire application lifetime.

The ServiceBusClient is designed to be a heavyweight object that maintains connections and manages resources efficiently. Hence, it should be registered as a Singleton to avoid excessive resource allocation and ensure optimal performance.

Before Fix: Using Scoped Registration

Here’s an example of the problematic configuration:

public void ConfigureServices(IServiceCollection services)
{
    services.AddScoped(serviceProvider =>
    {
        string connectionString = Configuration.GetConnectionString("ServiceBus");
        return new ServiceBusClient(connectionString);
    });

    services.AddScoped<IMessageProcessor, MessageProcessor>();
}

In this configuration:

• A new instance of ServiceBusClient is created for each HTTP request or scoped context.
• This quickly leads to resource exhaustion, causing the “Cannot allocate more handles” error.

Solution: Switching to Singleton

To fix this, register the ServiceBusClient as a Singleton, ensuring a single instance is shared across the application lifetime:

public void ConfigureServices(IServiceCollection services)
{
    services.AddSingleton(serviceProvider =>
    {
        string connectionString = Configuration.GetConnectionString("ServiceBus");
        return new ServiceBusClient(connectionString);
    });

    services.AddScoped<IMessageProcessor, MessageProcessor>();
}

In this configuration:

• A single instance of ServiceBusClient is created and reused for all requests.
• Resource usage is optimized, and the exception is avoided.

Sample Code: Before and After

Before Fix (Scoped Registration)

public interface IMessageProcessor
{
    Task ProcessMessageAsync();
}

public class MessageProcessor : IMessageProcessor
{
    private readonly ServiceBusClient _client;

    public MessageProcessor(ServiceBusClient client)
    {
        _client = client;
    }

    public async Task ProcessMessageAsync()
    {
        ServiceBusReceiver receiver = _client.CreateReceiver("queue-name");
        var message = await receiver.ReceiveMessageAsync();
        Console.WriteLine($"Received message: {message.Body}");
        await receiver.CompleteMessageAsync(message);
    }
}

After Fix (Singleton Registration)

public void ConfigureServices(IServiceCollection services)
{
    // Singleton registration for ServiceBusClient
    services.AddSingleton(serviceProvider =>
    {
        string connectionString = Configuration.GetConnectionString("ServiceBus");
        return new ServiceBusClient(connectionString);
    });

    services.AddScoped<IMessageProcessor, MessageProcessor>();
}

public class MessageProcessor : IMessageProcessor
{
    private readonly ServiceBusClient _client;

    public MessageProcessor(ServiceBusClient client)
    {
        _client = client;
    }

    public async Task ProcessMessageAsync()
    {
        ServiceBusReceiver receiver = _client.CreateReceiver("queue-name");
        var message = await receiver.ReceiveMessageAsync();
        Console.WriteLine($"Received message: {message.Body}");
        await receiver.CompleteMessageAsync(message);
    }
}

Key Takeaways

1. Always use Singleton scope for ServiceBusClient to optimize resource usage.
2. Avoid using Scoped or Transient scope for long-lived, resource-heavy objects.
3. Test your application under load to ensure no resource leakage occurs.