Timeout Strategy

Ensure operations complete within a specified time limit or get cancelled

Timeout Strategy

The timeout proactive resilience strategy cancels the execution if it does not complete within the specified timeout period. If the execution is canceled by the timeout strategy, it throws a TimeoutRejectedException.

It is crucial that your callback respects the cancellation token. If it does not, the callback will continue executing even after cancellation, thereby ignoring the timeout.

When to Use Timeout

Use the timeout strategy when:

• Operations should not take longer than a specific duration
• You want to prevent indefinite waiting for responses
• Dealing with potentially slow or unresponsive services
• Implementing SLA requirements with strict time limits
• Protecting against resource exhaustion from long-running operations

Installation

dotnet add package Polly.Core

Usage

Basic Timeout

// Simple timeout of 3 seconds
var pipeline = new ResiliencePipelineBuilder()
    .AddTimeout(TimeSpan.FromSeconds(3))
    .Build();

try
{
    await pipeline.ExecuteAsync(async ct => 
    {
        // Your operation that should complete within 3 seconds
        await CallExternalServiceAsync(ct);
    }, cancellationToken);
}
catch (TimeoutRejectedException)
{
    // Operation took too long
    Console.WriteLine("Operation timed out");
}

Using TimeoutStrategyOptions

var options = new TimeoutStrategyOptions
{
    Timeout = TimeSpan.FromSeconds(30),
    OnTimeout = args =>
    {
        Console.WriteLine($"Timeout occurred after {args.Timeout}");
        // Log, send metrics, etc.
        return default;
    }
};

var pipeline = new ResiliencePipelineBuilder()
    .AddTimeout(options)
    .Build();

Dynamic Timeout

var options = new TimeoutStrategyOptions
{
    TimeoutGenerator = static args =>
    {
        // Adjust timeout based on context
        var isHighPriority = args.Context.Properties
            .TryGetValue(new ResiliencePropertyKey<bool>("IsHighPriority"), out var priority) 
            && priority;
        
        var timeout = isHighPriority 
            ? TimeSpan.FromSeconds(60) 
            : TimeSpan.FromSeconds(10);
        
        return new ValueTask<TimeSpan>(timeout);
    }
};

Timeout with Events

var options = new TimeoutStrategyOptions
{
    Timeout = TimeSpan.FromSeconds(5),
    OnTimeout = static args =>
    {
        var operationKey = args.Context.OperationKey;
        Console.WriteLine($"{operationKey}: Execution timed out after {args.Timeout.TotalSeconds} seconds.");
        
        // Log to monitoring system
        // Send alert
        
        return default;
    }
};

Configuration Options

Timeout · TimeSpan · default: "30 seconds"

Defines a fixed period within which the delegate should complete, otherwise it will be cancelled.

TimeoutGenerator · Func<TimeoutGeneratorArguments, ValueTask<TimeSpan>> · default: "null"

Dynamically calculates the timeout period using runtime information. If set, the Timeout property is ignored.

OnTimeout · Func<OnTimeoutArguments, ValueTask> · default: "null"

Invoked after the timeout occurred, just before throwing TimeoutRejectedException.

Important Considerations

Respecting Cancellation Tokens

The timeout strategy relies on co-operative cancellation. Your callbacks must honor the cancellation token:

Good Example

var pipeline = new ResiliencePipelineBuilder()
    .AddTimeout(TimeSpan.FromSeconds(1))
    .Build();

await pipeline.ExecuteAsync(
    static async innerToken => 
    {
        // ✅ Use the innerToken provided by the pipeline
        await Task.Delay(TimeSpan.FromSeconds(3), innerToken);
    },
    outerToken);

If the cancellation token is not respected, the callback will continue executing after the timeout, and the strategy will have to wait for it to complete before throwing TimeoutRejectedException.

Best Practices

Always use the provided cancellation token

Pass the cancellation token provided by Polly to all async operations. This ensures proper timeout behavior.

await pipeline.ExecuteAsync(async ct => 
{
    await httpClient.GetAsync(url, ct); // ✅ Pass ct
    await Task.Delay(1000, ct);        // ✅ Pass ct
}, cancellationToken);

Set appropriate timeout values

• Too short: operations fail unnecessarily
• Too long: defeats the purpose of timeout

Consider your service’s typical response times and add buffer for variance.

Use dynamic timeouts for varying scenarios

Use TimeoutGenerator when different operations or contexts require different timeout values.

Combine with retry for comprehensive resilience

Place timeout inside retry to ensure each retry attempt has its own timeout:

var pipeline = new ResiliencePipelineBuilder()
    .AddRetry(new RetryStrategyOptions())
    .AddTimeout(TimeSpan.FromSeconds(5))
    .Build();

Log timeout events

Use OnTimeout to log or send metrics when timeouts occur. This helps identify slow operations.

Examples

HTTP Request with Timeout

var httpClient = new HttpClient();

var pipeline = new ResiliencePipelineBuilder<HttpResponseMessage>()
    .AddTimeout(new TimeoutStrategyOptions
    {
        Timeout = TimeSpan.FromSeconds(10),
        OnTimeout = args =>
        {
            Console.WriteLine($"HTTP request timed out after {args.Timeout}");
            return default;
        }
    })
    .Build();

try
{
    var response = await pipeline.ExecuteAsync(async ct =>
    {
        return await httpClient.GetAsync("https://api.example.com/data", ct);
    }, cancellationToken);
    
    Console.WriteLine($"Status: {response.StatusCode}");
}
catch (TimeoutRejectedException)
{
    Console.WriteLine("Request took too long");
}

Database Query with Timeout

var pipeline = new ResiliencePipelineBuilder<List<Customer>>()
    .AddTimeout(TimeSpan.FromSeconds(5))
    .Build();

var customers = await pipeline.ExecuteAsync(async ct =>
{
    await using var connection = new SqlConnection(connectionString);
    await connection.OpenAsync(ct);
    
    await using var command = new SqlCommand("SELECT * FROM Customers", connection);
    // Ensure the command respects the cancellation token
    ct.Register(() => command.Cancel());
    
    var results = new List<Customer>();
    await using var reader = await command.ExecuteReaderAsync(ct);
    
    while (await reader.ReadAsync(ct))
    {
        results.Add(new Customer
        {
            Id = reader.GetInt32(0),
            Name = reader.GetString(1)
        });
    }
    
    return results;
}, cancellationToken);

Per-Operation Dynamic Timeouts

var operationTimeoutKey = new ResiliencePropertyKey<TimeSpan>("OperationTimeout");

var options = new TimeoutStrategyOptions
{
    TimeoutGenerator = args =>
    {
        // Get timeout from context or use default
        var timeout = args.Context.Properties.GetValue(
            operationTimeoutKey, 
            TimeSpan.FromSeconds(30));
        
        return new ValueTask<TimeSpan>(timeout);
    }
};

var pipeline = new ResiliencePipelineBuilder().AddTimeout(options).Build();

// Use with custom timeout
var context = ResilienceContextPool.Shared.Get();
context.Properties.Set(operationTimeoutKey, TimeSpan.FromSeconds(5));

try
{
    await pipeline.ExecuteAsync(
        async (ctx, ct) => await SlowOperationAsync(ct),
        context,
        cancellationToken);
}
finally
{
    ResilienceContextPool.Shared.Return(context);
}

Timeout with Retry

// Each retry attempt has its own 5-second timeout
// Total time could be up to (3 attempts × 5 seconds) + retry delays
var pipeline = new ResiliencePipelineBuilder<HttpResponseMessage>()
    .AddRetry(new RetryStrategyOptions<HttpResponseMessage>
    {
        MaxRetryAttempts = 3,
        Delay = TimeSpan.FromSeconds(1),
        ShouldHandle = new PredicateBuilder<HttpResponseMessage>()
            .Handle<TimeoutRejectedException>()
            .Handle<HttpRequestException>()
    })
    .AddTimeout(TimeSpan.FromSeconds(5))
    .Build();

var response = await pipeline.ExecuteAsync(async ct =>
{
    return await httpClient.GetAsync(url, ct);
}, cancellationToken);

Overall Timeout for Retries

// Entire retry sequence must complete within 15 seconds
var pipeline = new ResiliencePipelineBuilder<HttpResponseMessage>()
    .AddTimeout(TimeSpan.FromSeconds(15))  // Outer timeout
    .AddRetry(new RetryStrategyOptions<HttpResponseMessage>
    {
        MaxRetryAttempts = 5,
        Delay = TimeSpan.FromSeconds(2)
    })
    .AddTimeout(TimeSpan.FromSeconds(3))   // Inner timeout per attempt
    .Build();

OnTimeout vs Try-Catch

Using OnTimeout

var pipeline = new ResiliencePipelineBuilder()
    .AddTimeout(new TimeoutStrategyOptions
    {
        Timeout = TimeSpan.FromSeconds(2),
        OnTimeout = args =>
        {
            Console.WriteLine("Timeout occurred");
            // Access Context and Timeout values
            return default;
        }
    })
    .Build();

await pipeline.ExecuteAsync(Operation, cancellationToken);

OnTimeout is called before the exception is thrown. This is useful when combining with retry strategy, as the retry might handle the exception. If you use try-catch, you might not catch the exception if retry succeeds on a subsequent attempt.

What happens if my operation ignores the cancellation token?

The timeout strategy will wait for your operation to complete before throwing TimeoutRejectedException. This defeats the purpose of the timeout. Always respect the cancellation token in your callbacks.

Can I use different timeouts for different operations?

Yes! Use TimeoutGenerator to dynamically calculate timeouts based on context properties, operation key, or any other runtime information.

Does timeout affect the original cancellation token?

The timeout strategy wraps the incoming cancellation token with a new one. If the original token is canceled, the timeout strategy honors it without throwing TimeoutRejectedException.

Should timeout be inside or outside retry?

It depends:

• Inside retry (retry → timeout): Each attempt gets its own timeout
• Outside retry (timeout → retry): The entire retry sequence must complete within the timeout

You can also use both for maximum control.