Oakrey.Applications.Performance 7.0.1

Oakrey.Applications.Performance

A .NET 10 Windows library for observing application CPU and RAM usage through Rx streams.

NuGet package ID: Oakrey.Applications.Performance

Main features

• CpuUsage exposes the latest CPU percentage as IObservable<int>.
• RamUsage exposes the latest RAM usage in MB as IObservable<int>.
• Usage exposes the original combined PerformanceLog stream.
• CPU and RAM streams immediately publish an initial value of zero to new subscribers.
• The polling period is configurable and must be greater than zero.
• PerformanceMonitoringLoggingService logs samples using placeholder-based Oakrey logging.
• Monitoring, subscriptions, subjects, and the underlying Windows monitor have idempotent cleanup.
• Source errors are logged and forwarded to the derived CPU and RAM streams.

Architecture

classDiagram
    class IPerformanceMonitoringService {
        +IObservable~int~ CpuUsage
        +IObservable~int~ RamUsage
        +IObservable~PerformanceLog~ Usage
    }

    class PerformanceMonitoringService {
        +PerformanceMonitoringService()
        +PerformanceMonitoringService(msPeriod)
        +Dispose()
    }

    class PerformanceMonitoringLoggingService {
        +Preload(cancellationToken) Task
        +Subscribe(observer) IDisposable
        +Dispose()
    }

    IPerformanceMonitoringService <|.. PerformanceMonitoringService
    PerformanceMonitoringLoggingService --> IPerformanceMonitoringService

No public interface signatures are changed by the current implementation.

Requirements and direct dependencies

• .NET 10 for Windows
• Oakrey.Debug.Windows 3.0.1
• Oakrey.Log.Abstractions 3.0.0
• System.Reactive 6.1.0
• Project reference to Oakrey.Applications.Abstractions for IPreLoadable and PreLoadingException

All package references are direct, current, and checked for known vulnerabilities.

Installation

dotnet add package Oakrey.Applications.Performance

Package Manager Console:

Install-Package Oakrey.Applications.Performance

Usage

using Oakrey.Applications.Performance;

using PerformanceMonitoringService monitoring =
    new(msPeriod: 1000);

using IDisposable cpuSubscription = monitoring.CpuUsage.Subscribe(
    cpu => Console.WriteLine($"CPU: {cpu}%"));

using IDisposable ramSubscription = monitoring.RamUsage.Subscribe(
    ram => Console.WriteLine($"RAM: {ram} MB"));

The native monitor can require an initialization period before its first sampled value. The initial zero from each BehaviorSubject is available immediately.

Automatic logging

using PerformanceMonitoringService monitoring = new();
using PerformanceMonitoringLoggingService logging = new(monitoring);

await logging.Preload(cancellationToken);

The logging service subscribes during construction and records each combined sample at information level:

CPU: {0}%, RAM: {1} MB

The placeholders are formatted by the Oakrey logger with CPU and RAM supplied as separate arguments.

Dependency injection

services.AddSingleton<IPerformanceMonitoringService, PerformanceMonitoringService>();
services.AddSingleton<PerformanceMonitoringLoggingService>();

Register the concrete monitoring service for disposal when callers also need to resolve it directly. The current IPerformanceMonitoringService contract does not inherit IDisposable.

Lifecycle and safety notes

• A non-positive polling period throws ArgumentOutOfRangeException.
• Observable wrappers are cached instead of allocated on every property access.
• Performance callbacks and disposal are synchronized.
• Subscriber failures are logged without stopping cleanup of the other streams.
• Source errors are propagated through CpuUsage and RamUsage.
• Disposal is safe to call more than once.
• Preload respects an already-cancelled token and rejects use after disposal.
• The preload exception replay buffer is limited to one item and is completed during disposal.
• Information-level sample logging can produce significant log volume at short polling periods. Choose the period according to operational needs.
• CPU and RAM measurements are operational metadata. Consider access controls and retention policy when exporting logs.

License

MIT - Copyright (c) Oakrey 2016-present