feat(memory): add two-tier engine cache lifetime management and host RAM telemetry (#97, #98)

App.axaml

@@ -16,7 +16,7 @@
     </Application.Styles>
     <TrayIcon.Icons>
         <TrayIcons>
-            <TrayIcon Icon="avares://LocalLLMServerManager/Assets/app-icon.ico" ToolTipText="Local LLM Server Manager v4.0.0" IsVisible="True">
+            <TrayIcon Icon="avares://LocalLLMServerManager/Assets/app-icon.ico" ToolTipText="Local LLM Server Manager v4.1.0" IsVisible="True">
                 <TrayIcon.Menu>
                     <NativeMenu>
                         <NativeMenuItem Header="Open Native Dashboard" Click="OnOpenDashboardClick"/>

Endpoints/EngineEndpoints.cs

@@ -1,4 +1,5 @@
 using LocalLLMServerManager.Services;
+using LocalLLMServerManager.Shared.Interfaces;
 using Microsoft.AspNetCore.Builder;
 using Microsoft.AspNetCore.Http;
 using Microsoft.Extensions.Logging;
@@ -9,15 +10,24 @@ public static class EngineEndpoints
 {
     public static void MapEngineEndpoints(this WebApplication app)
     {
-        app.MapGet("/api/gpu/vram", (IGpuTelemetryProvider telemetryProvider) =>
+        app.MapGet("/api/gpu/vram", (IGpuTelemetryProvider telemetryProvider, ISettingsService settingsService) =>
         {
-            var (gpuName, vramBytes, usedVramBytes) = telemetryProvider.GetGpuInfo();
+            return Results.Ok(BuildMemoryTelemetryResponse(telemetryProvider, settingsService));
+        });
+
+        app.MapGet("/api/hardware/memory", (IGpuTelemetryProvider telemetryProvider, ISettingsService settingsService) =>
+        {
+            return Results.Ok(BuildMemoryTelemetryResponse(telemetryProvider, settingsService));
+        });
+
+        app.MapPost("/api/memory/purge", async (VramOrchestrator orchestrator, IGpuTelemetryProvider telemetryProvider, ISettingsService settingsService, CancellationToken ct) =>
+        {
+            var eviction = await orchestrator.PurgeAllCachesAsync(stopIdleEngines: true, ct);
             return Results.Ok(new
             {
-                GpuName = gpuName,
-                VramBytes = vramBytes,
-                UsedVramBytes = usedVramBytes,
-                VramGB = Math.Round((double)vramBytes / (1024 * 1024 * 1024), 2)
+                status = "purged",
+                eviction,
+                memory = BuildMemoryTelemetryResponse(telemetryProvider, settingsService)
             });
         });
 
@@ -43,6 +53,7 @@ public static void MapEngineEndpoints(this WebApplication app)
                 return Results.BadRequest(new { message = $"Invalid or missing ComfyUI executable path: {execPath}" });
             }
 
+            orchestrator.RecordActivity("comfy");
             await orchestrator.EnsureVramForComfyUiAsync();
             var success = await engineManager.StartComfyUiAsync(execPath, logger);
 
@@ -60,7 +71,7 @@ public static void MapEngineEndpoints(this WebApplication app)
             return Results.Ok(new { message = "ComfyUI Stopped" });
         });
 
-        app.MapPost("/api/forge/start", async (IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
+        app.MapPost("/api/forge/start", async (VramOrchestrator orchestrator, IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
         {
             var logger = loggerFactory.CreateLogger("EngineEndpoints");
             var settings = settingsService.LoadSettings();
@@ -71,6 +82,7 @@ public static void MapEngineEndpoints(this WebApplication app)
                 return Results.BadRequest(new { message = $"Invalid or missing SD Forge executable path: {execPath}" });
             }
 
+            orchestrator.RecordActivity("forge");
             var success = await engineManager.StartForgeAsync(execPath, logger);
             if (success)
             {
@@ -88,11 +100,12 @@ public static void MapEngineEndpoints(this WebApplication app)
 
         app.MapPost("/api/comfy/free", async (VramOrchestrator orchestrator) =>
         {
+            orchestrator.RecordActivity("comfy");
             await orchestrator.FreeComfyUiVramAsync();
             return Results.Ok(new { message = "ComfyUI VRAM freed" });
         });
 
-        app.MapPost("/api/audio/start", async (IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
+        app.MapPost("/api/audio/start", async (VramOrchestrator orchestrator, IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
         {
             var logger = loggerFactory.CreateLogger("EngineEndpoints");
             var settings = settingsService.LoadSettings();
@@ -106,6 +119,7 @@ public static void MapEngineEndpoints(this WebApplication app)
                 }
             }
 
+            orchestrator.RecordActivity("audio");
             var success = await engineManager.StartAudioEngineAsync(execPath, logger);
             if (success)
             {
@@ -123,12 +137,13 @@ public static void MapEngineEndpoints(this WebApplication app)
 
         app.MapGet("/api/audio/voices", HandleGetVoicesAsync);
 
-        app.MapPost("/api/ollama/start", async (IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
+        app.MapPost("/api/ollama/start", async (VramOrchestrator orchestrator, IAiEngineManager engineManager, ISettingsService settingsService, ILoggerFactory loggerFactory) =>
         {
             var logger = loggerFactory.CreateLogger("EngineEndpoints");
             var settings = settingsService.LoadSettings();
             var execPath = AiEngineManager.ResolveOllamaExecutable(settings.OllamaExecutablePath);
 
+            orchestrator.RecordActivity("ollama");
             var success = await engineManager.StartOllamaAsync(execPath, logger);
             if (success)
             {
@@ -144,8 +159,9 @@ public static void MapEngineEndpoints(this WebApplication app)
             return Results.Ok(new { message = "Ollama Stopped" });
         });
 
-        app.MapPost("/api/engine/start", async (EngineToggleRequest req, IAiEngineManager engineManager) =>
+        app.MapPost("/api/engine/start", async (EngineToggleRequest req, VramOrchestrator orchestrator, IAiEngineManager engineManager) =>
         {
+            orchestrator.RecordActivity(req.Engine ?? "");
             var result = await engineManager.StartEngineAsync(req.Engine ?? "");
             if (result.Success)
             {
@@ -165,6 +181,44 @@ public static void MapEngineEndpoints(this WebApplication app)
         });
     }
 
+    public static object BuildMemoryTelemetryResponse(IGpuTelemetryProvider telemetryProvider, ISettingsService settingsService)
+    {
+        var (gpuName, vramBytes, usedVramBytes) = telemetryProvider.GetGpuInfo();
+        var sysMem = telemetryProvider.GetSystemMemoryInfo();
+        var settings = settingsService.LoadSettings();
+        double totalVramGb = Math.Round((double)vramBytes / (1024.0 * 1024.0 * 1024.0), 2);
+        double usedVramGb = Math.Round((double)usedVramBytes / (1024.0 * 1024.0 * 1024.0), 2);
+        double freeVramGb = Math.Max(0.0, Math.Round(totalVramGb - usedVramGb, 2));
+        int usedVramPct = vramBytes > 0 ? (int)Math.Clamp(Math.Round((double)usedVramBytes / vramBytes * 100.0), 0, 100) : 0;
+        double totalRamGb = Math.Round((double)sysMem.TotalRamBytes / (1024.0 * 1024.0 * 1024.0), 2);
+        double usedRamGb = Math.Round((double)sysMem.UsedRamBytes / (1024.0 * 1024.0 * 1024.0), 2);
+        double processWsMb = Math.Round((double)sysMem.ProcessWorkingSetBytes / (1024.0 * 1024.0), 1);
+        double nonPagedMb = Math.Round((double)sysMem.NonPagedPoolBytes / (1024.0 * 1024.0), 1);
+
+        return new
+        {
+            GpuName = gpuName,
+            VramBytes = vramBytes,
+            UsedVramBytes = usedVramBytes,
+            VramGB = totalVramGb,
+            TotalVramGb = totalVramGb,
+            UsedVramGb = usedVramGb,
+            FreeVramGb = freeVramGb,
+            UsedPercent = usedVramPct,
+            TotalRamBytes = sysMem.TotalRamBytes,
+            UsedRamBytes = sysMem.UsedRamBytes,
+            AvailableRamBytes = sysMem.AvailableRamBytes,
+            TotalRamGb = totalRamGb,
+            UsedRamGb = usedRamGb,
+            RamUsedPercent = sysMem.RamUsedPercent,
+            ProcessWorkingSetMb = processWsMb,
+            NonPagedPoolMb = nonPagedMb,
+            ModelCacheIdleTimeoutMinutes = settings.ModelCacheIdleTimeoutMinutes,
+            EngineHibernationTimeoutMinutes = settings.EngineHibernationTimeoutMinutes,
+            RamPressureThresholdPercent = settings.RamPressureThresholdPercent
+        };
+    }
+
     public static async Task<IResult> HandleGetVoicesAsync(ISettingsService settingsService, System.Net.Http.IHttpClientFactory clientFactory)
     {
         var settings = settingsService.LoadSettings();
@@ -197,5 +251,3 @@ public static async Task<IResult> HandleGetVoicesAsync(ISettingsService settings
         return Results.Ok(new { voices = defaultVoices, preferred = settings.PreferredAudioVoice });
     }
 }
-
-

Endpoints/HealthEndpoints.cs

@@ -29,7 +29,7 @@ public static void MapHealthEndpoints(this WebApplication app)
                 StableDiffusion = forgeHealthy ? "Online" : "Offline",
                 ComfyUI = comfyHealthy ? "Online" : "Offline",
                 PreferredImageEngine = settings.PreferredImageEngine,
-                Version = "4.0.0",
+                Version = "4.1.0",
             });
         });
     }

LocalLLMServerManager.Shared/Interfaces/ITelemetryService.cs

@@ -7,7 +7,15 @@ public record GpuTelemetryInfo(
     double UsedVramGb,
     double TotalVramGb,
     double FreeVramGb,
-    int Percent
+    int Percent,
+    double UsedRamGb = 0.0,
+    double TotalRamGb = 32.0,
+    int RamPercent = 0,
+    double ProcessWorkingSetMb = 0.0,
+    double NonPagedPoolMb = 0.0,
+    int ModelCacheIdleTimeoutMinutes = 15,
+    int EngineHibernationTimeoutMinutes = 30,
+    int RamPressureThresholdPercent = 75
 );
 
 public interface ITelemetryService

LocalLLMServerManager.Shared/LocalLLMServerManager.Shared.csproj

@@ -4,9 +4,9 @@
     <TargetFramework>net10.0</TargetFramework>
     <Nullable>enable</Nullable>
     <ImplicitUsings>enable</ImplicitUsings>
-    <Version>4.0.0</Version>
-    <AssemblyVersion>4.0.0.0</AssemblyVersion>
-    <FileVersion>4.0.0.0</FileVersion>
+    <Version>4.1.0</Version>
+    <AssemblyVersion>4.1.0.0</AssemblyVersion>
+    <FileVersion>4.1.0.0</FileVersion>
     <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
   </PropertyGroup>
 

LocalLLMServerManager.Shared/Models/AppSettings.cs

@@ -32,7 +32,8 @@ public record AppSettings(
     string AiAssistantApiKey = "",
     string AiAssistantModel = "google/gemini-2.5-flash",
     string AiAssistantCustomSystemPrompt = "",
-    string AiAssistantPromptsDirectory = ""
+    string AiAssistantPromptsDirectory = "",
+    int ModelCacheIdleTimeoutMinutes = 15,
+    int EngineHibernationTimeoutMinutes = 30,
+    int RamPressureThresholdPercent = 75
 );
-
-

LocalLLMServerManager.Shared/Models/EngineStatusCardModel.cs

@@ -84,6 +84,12 @@ public class TelemetryData
     public double TotalVramGb { get; set; } = 16.0;
     public int VramPercent { get; set; }
 
+    public double UsedRamGb { get; set; } = 0.0;
+    public double TotalRamGb { get; set; } = 32.0;
+    public int RamPercent { get; set; } = 0;
+    public double ProcessWorkingSetMb { get; set; } = 0.0;
+    public double NonPagedPoolMb { get; set; } = 0.0;
+
     public TelemetryData()
     {
     }

LocalLLMServerManager.Shared/Services/TelemetryService.cs

@@ -16,18 +16,68 @@ public async Task<GpuTelemetryInfo> QueryGpuVramAsync(string apiBase, HttpClient
             if (resp.IsSuccessStatusCode)
             {
                 var doc = JsonNode.Parse(await resp.Content.ReadAsStringAsync());
-                string gpuName = doc?["gpuName"]?.ToString() ?? "GPU Telemetry Active";
-                double used = doc?["usedVramGb"]?.GetValue<double>() ?? 0.0;
-                double total = doc?["totalVramGb"]?.GetValue<double>() ?? 16.0;
-                double free = doc?["freeVramGb"]?.GetValue<double>() ?? (total - used);
-                int pct = doc?["usedPercent"]?.GetValue<int>() ?? (int)Math.Round((used / (total > 0 ? total : 1.0)) * 100.0);
+                string gpuName = doc?["gpuName"]?.ToString() 
+                    ?? doc?["GpuName"]?.ToString() 
+                    ?? "GPU Telemetry Active";
 
-                return new GpuTelemetryInfo(gpuName, used, total, free, pct);
+                double totalVram = TryGetDouble(doc, "totalVramGb", "TotalVramGb") 
+                    ?? TryGetDouble(doc, "vramGB", "VramGB") 
+                    ?? (TryGetDouble(doc, "vramBytes", "VramBytes") is double vb ? Math.Round(vb / (1024.0 * 1024.0 * 1024.0), 2) : 16.0);
+
+                double usedVram = TryGetDouble(doc, "usedVramGb", "UsedVramGb") 
+                    ?? (TryGetDouble(doc, "usedVramBytes", "UsedVramBytes") is double uvb ? Math.Round(uvb / (1024.0 * 1024.0 * 1024.0), 2) : 0.0);
+
+                double freeVram = TryGetDouble(doc, "freeVramGb", "FreeVramGb") 
+                    ?? Math.Max(0.0, Math.Round(totalVram - usedVram, 2));
+
+                int pct = TryGetInt(doc, "usedPercent", "UsedPercent") 
+                    ?? (int)Math.Round((usedVram / (totalVram > 0 ? totalVram : 1.0)) * 100.0);
+
+                double totalRam = TryGetDouble(doc, "totalRamGb", "TotalRamGb") ?? 32.0;
+                double usedRam = TryGetDouble(doc, "usedRamGb", "UsedRamGb") ?? 0.0;
+                int ramPct = TryGetInt(doc, "ramUsedPercent", "RamUsedPercent") 
+                    ?? (int)Math.Round((usedRam / (totalRam > 0 ? totalRam : 1.0)) * 100.0);
+
+                double processWs = TryGetDouble(doc, "processWorkingSetMb", "ProcessWorkingSetMb") ?? 0.0;
+                double nonPaged = TryGetDouble(doc, "nonPagedPoolMb", "NonPagedPoolMb") ?? 0.0;
+                int modelCacheTimeout = TryGetInt(doc, "modelCacheIdleTimeoutMinutes", "ModelCacheIdleTimeoutMinutes") ?? 15;
+                int hibernationTimeout = TryGetInt(doc, "engineHibernationTimeoutMinutes", "EngineHibernationTimeoutMinutes") ?? 30;
+                int ramPressureThreshold = TryGetInt(doc, "ramPressureThresholdPercent", "RamPressureThresholdPercent") ?? 75;
+
+                return new GpuTelemetryInfo(
+                    gpuName,
+                    usedVram,
+                    totalVram,
+                    freeVram,
+                    pct,
+                    usedRam,
+                    totalRam,
+                    ramPct,
+                    processWs,
+                    nonPaged,
+                    modelCacheTimeout,
+                    hibernationTimeout,
+                    ramPressureThreshold
+                );
             }
         }
         catch { }
 
-        return new GpuTelemetryInfo("GPU Telemetry Active", 0.0, 16.0, 16.0, 0);
+        return new GpuTelemetryInfo("GPU Telemetry Active", 0.0, 16.0, 16.0, 0, 0.0, 32.0, 0, 0.0, 0.0, 15, 30, 75);
+    }
+
+    private static double? TryGetDouble(JsonNode? node, string camel, string pascal)
+    {
+        var target = node?[camel] ?? node?[pascal];
+        if (target == null) return null;
+        try { return target.GetValue<double>(); } catch { return null; }
+    }
+
+    private static int? TryGetInt(JsonNode? node, string camel, string pascal)
+    {
+        var target = node?[camel] ?? node?[pascal];
+        if (target == null) return null;
+        try { return target.GetValue<int>(); } catch { return null; }
     }
 
     public async Task<(bool Ollama, bool Forge, bool ComfyUi)> CheckServiceHealthAsync(string apiBase, string comfyUrl, HttpClient http)

LocalLLMServerManager.Shared/Styles/DesignTokens.axaml

@@ -52,6 +52,7 @@
     <StreamGeometry x:Key="IconSticker">M21.41 11.58l-9-9C12.05 2.22 11.55 2 11 2H4c-1.1 0-2 .9-2 2v7c0 .55.22 1.05.59 1.42l9 9c.36.36.86.58 1.41.58.55 0 1.05-.22 1.41-.59l7-7c.37-.36.59-.86.59-1.41 0-.55-.23-1.06-.59-1.42zM5.5 7C4.67 7 4 6.33 4 5.5S4.67 4 5.5 4 7 4.67 7 5.5 6.33 7 5.5 7z</StreamGeometry>
     <StreamGeometry x:Key="IconGpu">M21 6H3c-1.1 0-2 .9-2 2v8c0 1.1.9 2 2 2h18c1.1 0 2-.9 2-2V8c0-1.1-.9-2-2-2zm-10 7H8v3H6v-3H3v-2h3V8h2v3h3v2zm4.5 2c-.83 0-1.5-.67-1.5-1.5s.67-1.5 1.5-1.5 1.5.67 1.5 1.5-.67 1.5-1.5 1.5zm3-3c-.83 0-1.5-.67-1.5-1.5s.67-1.5 1.5-1.5 1.5.67 1.5 1.5-.67 1.5-1.5 1.5z</StreamGeometry>
     <StreamGeometry x:Key="IconRefresh">M17.65 6.35C16.2 4.9 14.21 4 12 4c-4.42 0-7.99 3.58-7.99 8s3.57 8 7.99 8c3.73 0 6.84-2.55 7.73-6h-2.08c-.82 2.33-3.04 4-5.65 4-3.31 0-6-2.69-6-6s2.69-6 6-6c1.66 0 3.14.69 4.22 1.78L13 11h7V4l-2.35 2.35z</StreamGeometry>
+    <StreamGeometry x:Key="IconPurge">M19,4H15.5L14.5,3H9.5L8.5,4H5V6H19M6,19A2,2 0 0,0 8,21H16A2,2 0 0,0 18,19V7H6V19Z</StreamGeometry>
     <StreamGeometry x:Key="IconFolder">M10 4H4c-1.1 0-1.99.9-1.99 2L2 18c0 1.1.9 2 2 2h16c1.1 0 2-.9 2-2V8c0-1.1-.9-2-2-2h-8l-2-2z</StreamGeometry>
     <StreamGeometry x:Key="IconCopy">M16 1H4c-1.1 0-2 .9-2 2v14h2V3h12V1zm3 4H8c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h11c1.1 0 2-.9 2-2V7c0-1.1-.9-2-2-2zm0 16H8V7h11v14z</StreamGeometry>
     <StreamGeometry x:Key="IconSave">M17 3H5c-1.11 0-2 .9-2 2v14c0 1.1.89 2 2 2h14c1.1 0 2-.9 2-2V7l-4-4zm-5 16c-1.66 0-3-1.34-3-3s1.34-3 3-3 3 1.34 3 3-1.34 3-3 3zm3-10H5V5h10v4z</StreamGeometry>