Add HTTP resilience with retry policy for Azure OpenAI rate limiting

Directory.Packages.props

@@ -36,6 +36,7 @@
     <PackageVersion Include="Microsoft.EntityFrameworkCore.Sqlite" Version="8.0.10" />
     <PackageVersion Include="Microsoft.EntityFrameworkCore.SqlServer" Version="8.0.10" />
     <PackageVersion Include="Microsoft.EntityFrameworkCore.Tools" Version="8.0.10" />
+    <PackageVersion Include="Microsoft.Extensions.Http.Resilience" Version="10.2.0" />
     <PackageVersion Include="Microsoft.NET.Test.Sdk" Version="18.0.1" />
     <PackageVersion Include="Microsoft.SemanticKernel" Version="1.60.0" />
     <PackageVersion Include="Microsoft.SemanticKernel.Connectors.PgVector" Version="1.60.0-preview" />

EssentialCSharp.Chat.Shared/EssentialCSharp.Chat.Common.csproj

@@ -6,6 +6,7 @@
 
   <ItemGroup>
     <PackageReference Include="Azure.Identity" />
+    <PackageReference Include="Microsoft.Extensions.Http.Resilience" />
     <PackageReference Include="Microsoft.SemanticKernel" />
     <PackageReference Include="Microsoft.SemanticKernel.Connectors.PgVector" />
     <PackageReference Include="ModelContextProtocol" />

EssentialCSharp.Chat.Shared/Extensions/ServiceCollectionExtensions.cs

@@ -4,8 +4,10 @@
 using EssentialCSharp.Chat.Common.Services;
 using Microsoft.Extensions.Configuration;
 using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Http.Resilience;
 using Microsoft.SemanticKernel;
 using Npgsql;
+using Polly;
 
 namespace EssentialCSharp.Chat.Common.Extensions;
 
@@ -38,6 +40,9 @@ public static IServiceCollection AddAzureOpenAIServices(
 
         var endpoint = new Uri(aiOptions.Endpoint);
 
+        // Configure HTTP resilience for Azure OpenAI requests
+        ConfigureAzureOpenAIResilience(services);
+
         // Register Azure OpenAI services with Managed Identity authentication
 #pragma warning disable SKEXP0010 // Type is for evaluation purposes only and is subject to change or removal in future updates. Suppress this diagnostic to proceed.
         services.AddAzureOpenAIChatClient(
@@ -71,6 +76,65 @@ public static IServiceCollection AddAzureOpenAIServices(
         return services;
     }
 
+    /// <summary>
+    /// Configures HTTP resilience (retry, circuit breaker, timeout) for Azure OpenAI HTTP clients.
+    /// This handles rate limiting (HTTP 429) and transient errors with exponential backoff.
+    /// </summary>
+    /// <param name="services">The service collection to configure</param>
+    /// <remarks>
+    /// This method configures resilience for ALL HTTP clients created via IHttpClientFactory. 
+    /// 
+    /// IMPORTANT: The Semantic Kernel's AddAzureOpenAI* extension methods (used in this class)
+    /// do NOT expose options to configure specific named or typed HttpClients. The internal
+    /// implementation creates HttpClient instances through IHttpClientFactory without
+    /// providing hooks for per-client configuration. Therefore, ConfigureHttpClientDefaults
+    /// is the ONLY way to apply resilience to Azure OpenAI clients when using Semantic Kernel.
+    /// 
+    /// For Azure OpenAI services specifically, the resilience configuration:
+    /// - Retries HTTP 429 (rate limit), 408 (timeout), and 5xx errors
+    /// - Respects Retry-After headers from Azure OpenAI
+    /// - Uses exponential backoff with jitter
+    /// - Implements circuit breaker pattern
+    /// 
+    /// This is appropriate for applications that primarily use Azure OpenAI services.
+    /// The retry policies are reasonable for most HTTP APIs and should not negatively
+    /// impact other HTTP clients like hCaptcha or Mailjet.
+    /// </remarks>
+    private static void ConfigureAzureOpenAIResilience(IServiceCollection services)
+    {
+        // Configure resilience for all HTTP clients created via IHttpClientFactory
+        // The Semantic Kernel's AddAzureOpenAI* methods do not support named/typed
+        // HttpClient configuration, so ConfigureHttpClientDefaults is required.
+        services.ConfigureHttpClientDefaults(httpClientBuilder =>
+        {
+            httpClientBuilder.AddStandardResilienceHandler(options =>
+            {
+                // Configure retry strategy for rate limiting and transient errors
+                options.Retry.MaxRetryAttempts = 5;
+                options.Retry.Delay = TimeSpan.FromSeconds(2);
+                options.Retry.BackoffType = DelayBackoffType.Exponential;
+                options.Retry.UseJitter = true;
+
+                // The standard resilience handler already handles:
+                // - HTTP 429 (Too Many Requests / Rate Limit)
+                // - HTTP 408 (Request Timeout)
+                // - HTTP 5xx (Server Errors)
+                // - Respects Retry-After header automatically
+
+                // Configure circuit breaker to prevent overwhelming the service
+                options.CircuitBreaker.SamplingDuration = TimeSpan.FromSeconds(30);
+                options.CircuitBreaker.BreakDuration = TimeSpan.FromSeconds(15);
+                options.CircuitBreaker.FailureRatio = 0.2; // Break if 20% of requests fail
+
+                // Configure timeout for individual attempts
+                options.AttemptTimeout.Timeout = TimeSpan.FromSeconds(30);
+
+                // Configure total timeout for all retry attempts
+                options.TotalRequestTimeout.Timeout = TimeSpan.FromMinutes(3);
+            });
+        });
+    }
+
     /// <summary>
     /// Adds Azure OpenAI and related AI services to the service collection using configuration
     /// </summary>
@@ -183,6 +247,9 @@ public static IServiceCollection AddAzureOpenAIServicesWithApiKey(
 
         var endpoint = new Uri(aiOptions.Endpoint);
 
+        // Configure HTTP resilience for Azure OpenAI requests
+        ConfigureAzureOpenAIResilience(services);
+
         // Register Azure OpenAI services with API key authentication
 #pragma warning disable SKEXP0010 // Type is for evaluation purposes only and is subject to change or removal in future updates. Suppress this diagnostic to proceed.
         services.AddAzureOpenAIChatClient(

EssentialCSharp.Web/Extensions/LoggerExtensions.cs

@@ -4,7 +4,7 @@ namespace EssentialCSharp.Web.Extensions;
 
 internal static partial class LoggerExtensions
 {
-    [LoggerMessage(Level = LogLevel.Debug, EventId = 1, Message = "Successful captcha with response of: '{JsonResult}'")]
+    [LoggerMessage(Level = LogLevel.Debug, EventId = 1, Message = "Successful captcha with response")]
     public static partial void HomeControllerSuccessfulCaptchaResponse(
-           this ILogger logger, JsonResult jsonResult);
+           this ILogger logger);
 }