Merge pull request #4 from cernuto/feature/performance-and-reliability-optimizations

Optimizations with reliability changes

Author: KSemenenko

.claude/settings.local.json

@@ -0,0 +1,12 @@
+{
+  "permissions": {
+    "allow": [
+      "Bash(tree:*)",
+      "Bash(dotnet build:*)",
+      "Bash(dotnet test:*)",
+      "Bash(git checkout:*)",
+      "Bash(git push:*)",
+      "Bash(git remote add:*)"
+    ]
+  }
+}

.editorconfig

@@ -453,6 +453,20 @@ dotnet_diagnostic.IDE0200.severity = warning
 dotnet_style_allow_multiple_blank_lines_experimental = false
 dotnet_diagnostic.IDE2000.severity = warning
 
+# SignalR Hub methods are called by string name from clients - suppress async naming rule
+[**/Hubs/*.cs]
+dotnet_naming_rule.async_methods_should_end_with_async.severity = none
+
+# ASP.NET Controller actions don't typically follow async naming conventions
+[**/Controllers/*.cs]
+dotnet_naming_rule.async_methods_should_end_with_async.severity = none
+
+# Orleans grains with [PersistentState] attributes cannot use primary constructors
+# because attributes on constructor parameters cannot be applied to primary constructor parameters
+[**/SignalRConnectionHeartbeatGrain.cs]
+csharp_style_prefer_primary_constructors = false:none
+dotnet_diagnostic.IDE0290.severity = none
+
 # Verify settings for test files
 [*.{received,verified}.{txt,xml,json}]
 charset = utf-8-bom

.gitignore

@@ -646,4 +646,7 @@ MigrationBackup/
 # Ionide (cross platform F# VS Code tools) working folder
 .ionide/
 
-# End of https://www.toptal.com/developers/gitignore/api/intellij,intellij+all,macos,linux,windows,visualstudio,visualstudiocode,rider
+# End of https://www.toptal.com/developers/gitignore/api/intellij,intellij+all,macos,linux,windows,visualstudio,visualstudiocode,rider
+
+# Claude Code temporary directories
+tmpclaude-*/

ManagedCode.Orleans.SignalR.Client/Properties/launchSettings.json

@@ -0,0 +1,12 @@
+{
+  "profiles": {
+    "ManagedCode.Orleans.SignalR.Client": {
+      "commandName": "Project",
+      "launchBrowser": true,
+      "environmentVariables": {
+        "ASPNETCORE_ENVIRONMENT": "Development"
+      },
+      "applicationUrl": "https://localhost:56460;http://localhost:56463"
+    }
+  }
+}

ManagedCode.Orleans.SignalR.Core/Config/OrleansSignalROptions.cs

@@ -54,4 +54,102 @@ public class OrleansSignalROptions
     ///     Used as a hint when determining how many partitions to allocate dynamically.
     /// </summary>
     public int GroupsPerPartitionHint { get; set; } = 1_000;
+
+    /// <summary>
+    ///     Maximum number of messages to queue per user when they are disconnected.
+    ///     Oldest messages are dropped when the limit is exceeded.
+    ///     The default value is 100.
+    /// </summary>
+    public int MaxQueuedMessagesPerUser { get; set; } = 100;
+
+    /// <summary>
+    ///     Number of consecutive failures before an observer is considered dead and removed.
+    ///     Set to 0 to disable failure tracking.
+    ///     The default value is 3.
+    /// </summary>
+    public int ObserverFailureThreshold { get; set; } = 3;
+
+    /// <summary>
+    ///     Time window for counting observer failures. Failures older than this are forgotten.
+    ///     The default value is 30 seconds.
+    /// </summary>
+    public TimeSpan ObserverFailureWindow { get; set; } = TimeSpan.FromSeconds(30);
+
+    /// <summary>
+    ///     Enables circuit breaker pattern for observers to prevent cascade failures.
+    ///     When enabled, failing observers are temporarily blocked from receiving messages.
+    ///     The default value is true.
+    /// </summary>
+    public bool EnableCircuitBreaker { get; set; } = true;
+
+    /// <summary>
+    ///     Duration to keep the circuit open (blocking requests) after failure threshold is reached.
+    ///     After this duration, the circuit transitions to half-open state for testing.
+    ///     The default value is 30 seconds.
+    /// </summary>
+    public TimeSpan CircuitBreakerOpenDuration { get; set; } = TimeSpan.FromSeconds(30);
+
+    /// <summary>
+    ///     Interval between test requests when circuit is in half-open state.
+    ///     The default value is 5 seconds.
+    /// </summary>
+    public TimeSpan CircuitBreakerHalfOpenTestInterval { get; set; } = TimeSpan.FromSeconds(5);
+
+    /// <summary>
+    ///     Grace period before an observer is hard-removed after a failure.
+    ///     During this period, messages are buffered and replayed if the observer recovers.
+    ///     This handles timing edge cases like GC pauses, network latency, or silo overload.
+    ///     Set to TimeSpan.Zero to disable grace period buffering.
+    ///     The default value is 10 seconds.
+    /// </summary>
+    public TimeSpan ObserverGracePeriod { get; set; } = TimeSpan.FromSeconds(10);
+
+    /// <summary>
+    ///     Maximum number of messages to buffer per observer during the grace period.
+    ///     Oldest messages are dropped when the limit is exceeded.
+    ///     The default value is 50.
+    /// </summary>
+    public int MaxBufferedMessagesPerObserver { get; set; } = 50;
+
+    /// <summary>
+    ///     Maximum number of connections allowed per partition grain.
+    ///     New connections are rejected when the limit is exceeded.
+    ///     Set to 0 to disable connection limits (not recommended for production).
+    ///     The default value is 100,000.
+    /// </summary>
+    public int MaxConnectionsPerPartition { get; set; } = 100_000;
+
+    /// <summary>
+    ///     Maximum number of groups per partition grain.
+    ///     New groups are rejected when the limit is exceeded.
+    ///     Set to 0 to disable group limits.
+    ///     The default value is 50,000.
+    /// </summary>
+    public int MaxGroupsPerPartition { get; set; } = 50_000;
+
+    /// <summary>
+    ///     Timeout for slow client message delivery.
+    ///     Connections that cannot receive messages within this time may be terminated.
+    ///     The default value is 10 seconds.
+    /// </summary>
+    public TimeSpan SlowClientTimeout { get; set; } = TimeSpan.FromSeconds(10);
+
+    /// <summary>
+    ///     Enables backpressure handling for slow clients.
+    ///     When enabled, messages to slow clients are dropped or the connection is terminated.
+    ///     The default value is true.
+    /// </summary>
+    public bool EnableSlowClientHandling { get; set; } = true;
+
+    /// <summary>
+    ///     Maximum number of pending messages allowed per connection before backpressure is applied.
+    ///     The default value is 1000.
+    /// </summary>
+    public int MaxPendingMessagesPerConnection { get; set; } = 1000;
+
+    /// <summary>
+    ///     Enables metrics collection for monitoring and diagnostics.
+    ///     The default value is true.
+    /// </summary>
+    public bool EnableMetrics { get; set; } = true;
 }