Franz.Common.Mediator.Polly 1.6.14

Franz.Common.Mediator.Polly

Franz.Common.Mediator.Polly extends Franz.Common.Mediator with Polly-based resilience pipelines. It gives you retry, circuit breaker, advanced circuit breaker, timeout, and bulkhead isolation for Mediator requests — all with enriched Serilog logging, context-awareness, and resilience telemetry built-in.

⚡ No extra wiring. No boilerplate. Just plug it in.

• Current Version: 1.6.14
• Target Frameworks: .NET 9+
• Dependencies: Polly, Serilog, Franz.Common.Mediator

✨ Features

• 🔁 Retry Pipeline — automatic retries with backoff & correlated telemetry.

• 🚦 Circuit Breaker Pipeline — prevents cascading failures under load.

• 📊 Advanced Circuit Breaker Pipeline — trips based on failure ratio in rolling window.

• ⏱ Timeout Pipeline — cancels long-running requests automatically.

• 📦 Bulkhead Pipeline — limits concurrent requests and queue pressure.

• 🧠 ResilienceContext — shared state across all resilience pipelines:

  • RetryCount, CircuitOpen, TimeoutOccurred, BulkheadRejected, Duration

• 👁 IResilienceObserver — extensibility hooks for external telemetry, alerts, or dashboards.

• 📝 Enriched Serilog Logging — all logs include:

  • Correlation ID
  • Request type
  • Policy name
  • Pipeline name
  • Execution time
  • Health indicators

📦 Installation

dotnet add package Franz.Common.Mediator.Polly

⚙️ Setup

1. Config-driven Entry Point (v1.6.2+)

From v1.6.2, resilience is now config-driven. Define your policies in appsettings.json:

"Resilience": {
  "RetryPolicy": {
    "Enabled": true,
    "RetryCount": 3,
    "RetryIntervalMilliseconds": 500
  },
  "CircuitBreaker": {
    "Enabled": true,
    "FailureThreshold": 0.5,
    "MinimumThroughput": 10,
    "DurationOfBreakSeconds": 30
  },
  "TimeoutPolicy": {
    "Enabled": true,
    "TimeoutSeconds": 5
  },
  "BulkheadPolicy": {
    "Enabled": true,
    "MaxParallelization": 10,
    "MaxQueueSize": 20
  }
}

Then just call:

builder.Services.AddFranzResilience(builder.Configuration);

✅ That’s it — retry, circuit breaker, timeout, and bulkhead are auto-registered from config and wired into Mediator pipelines. ✅ Each policy injects structured logs and updates the ResilienceContext.

2. Manual Registration (pre-1.6.2 style)

If you prefer explicit registration:

using Franz.Common.Mediator.Polly;

builder.Services.AddFranzPollyPolicies(options =>
{
    options.AddRetry("DefaultRetry", retryCount: 3, intervalMs: 500);
    options.AddCircuitBreaker("DefaultCircuitBreaker", 0.5, 10, 30);
    options.AddTimeout("DefaultTimeout", 5);
    options.AddBulkhead("DefaultBulkhead", 10, 20);
});

builder.Services
    .AddFranzPollyRetry("DefaultRetry")
    .AddFranzPollyCircuitBreaker("DefaultCircuitBreaker")
    .AddFranzPollyTimeout("DefaultTimeout")
    .AddFranzPollyBulkhead("DefaultBulkhead");

🧠 Observability Enhancements (v1.6.14)

Version 1.6.14 introduces a resilience-awareness layer across all Mediator pipelines.

🧩 ResilienceContext

Carries runtime state between pipelines:

public sealed class ResilienceContext
{
    public string PolicyName { get; init; } = string.Empty;
    public int RetryCount { get; set; }
    public bool CircuitOpen { get; set; }
    public bool TimeoutOccurred { get; set; }
    public bool BulkheadRejected { get; set; }
    public TimeSpan Duration { get; set; }
    public DateTimeOffset Timestamp { get; } = DateTimeOffset.UtcNow;

    public bool IsHealthy => !CircuitOpen && !TimeoutOccurred && !BulkheadRejected;
}

Each pipeline updates this context and emits structured logs through Serilog.

👁 IResilienceObserver

Observers can listen to policy outcomes globally:

public interface IResilienceObserver
{
    void OnPolicyExecuted(string policyName, ResilienceContext context);
}

You can implement custom observers for metrics or telemetry (e.g., Application Insights, Prometheus, Elastic APM).

Example:

public sealed class ElasticResilienceObserver : IResilienceObserver
{
    private readonly ILogger<ElasticResilienceObserver> _logger;

    public ElasticResilienceObserver(ILogger<ElasticResilienceObserver> logger)
        => _logger = logger;

    public void OnPolicyExecuted(string policyName, ResilienceContext context)
        => _logger.LogInformation("🧠 {Policy} -> Healthy={Healthy} Duration={Duration}ms Retries={RetryCount}",
            policyName, context.IsHealthy, context.Duration.TotalMilliseconds, context.RetryCount);
}

Register it once:

builder.Services.AddSingleton<IResilienceObserver, ElasticResilienceObserver>();

📊 Pipelines Overview

All pipelines automatically participate in Franz’s logging & correlation system.

🧩 Example Logs (v1.6.14)

Success

[12:01:22 INF] ▶️ Executing GetBookQuery [abc123] with RetryPolicy
[12:01:22 INF] ✅ GetBookQuery [abc123] succeeded after 47ms (policy RetryPolicy, retries=0)

Retry + Timeout

[12:01:25 WRN] 🔁 GetBookQuery [abc123] retry attempt 2 (policy RetryPolicy)
[12:01:25 ERR] ⏱️ GetBookQuery [abc123] timed out after 5s (policy TimeoutPolicy)

Circuit Breaker Open

[12:01:27 ERR] ❌ GetBookQuery [abc123] failed after 3 retries (policy RetryPolicy)
[12:01:27 WRN] 🚦 Circuit opened for 30s (policy CircuitBreaker)

🛠 Benefits

• 🧩 Composability-first — pipelines remain orthogonal yet share context.
• 🧠 Self-aware architecture — logs know what policies were triggered.
• 📈 Observer hooks — tap into resilience events for monitoring or dashboards.
• ⚡ Zero boilerplate — configured in <20 lines.
• 🏢 Enterprise-ready — deterministic, auditable, and DI-safe.

🗺 Roadmap

• FranzResilienceSummaryPipeline — emits aggregated resilience telemetry per request.
• OpenTelemetry integration via Activity tags.
• Prebuilt “DefaultSets” (HTTP, Database, Messaging).

📜 Changelog

v1.6.14

• 🧠 Introduced ResilienceContext — unified runtime state for all pipelines.
• 👁 Added IResilienceObserver for external resilience monitoring.
• 🧾 Upgraded all pipelines to emit context-rich Serilog logs.
• 🔗 Added correlation ID propagation across all resilience policies.
• 🚀 Internal optimizations to reduce policy lookup overhead.

v1.6.2

• ✨ Added AddFranzResilience(IConfiguration) for config-driven resilience.
• ♻️ Unified policy registry and Mediator pipelines.
• 🛡 Simplified startup — <20 lines bootstraps resilience + mediator.

⚡ With Franz.Common.Mediator.Polly, resilience is first-class, observable, and deterministic. Configure it once — and your Mediator pipelines automatically enforce retries, timeouts, bulkheads, and breakers with total visibility.