三层架构项目日志监控最优方案选型：StackExchange.Exceptional相关咨询

Hey there! Let me break this down based on my hands-on experience with these tools, especially in high-throughput .NET environments like yours.

StackExchange.Exceptional is the lightweight, high-performance exception logging library built by the Stack Exchange team for .NET (including .NET Core). It’s designed specifically to solve the pain points of tracking exceptions in production:

• It automatically captures unhandled exceptions across your Web API and WinForms client, and supports manual logging of custom exceptions too.
• It aggregates duplicate exceptions (grouped by stack trace, message, etc.) which is a huge win for your high-log-frequency scenario—no more drowning in thousands of identical error entries.
• It stores rich context with each exception: request details, user info, environment data, and even custom properties you add, making debugging way faster.

The catch? It’s primarily focused on exception logging. Your needs include database query logs and audit logs, which aren’t natively supported by Exceptional. But that’s not a dealbreaker—you can pair it with other tools to cover those gaps.

Also, Exceptional isn’t tied to a specific storage backend. While it defaults to SQL Server, there are community-supported extensions for MongoDB, PostgreSQL, and Elasticsearch. So you can use it with the NoSQL store you were already considering.

OPServer is Stack Exchange’s open-source monitoring dashboard—it’s not a log storage system, but a visualization layer. It pulls data from Exceptional, server metrics, database stats, and more to give you a single pane of glass for your infrastructure.

You don’t need it to use Exceptional, but it’s extremely useful if your team wants:

• A unified view of exception trends, server load, and PostgreSQL performance.
• Alerting for critical errors or resource spikes.
• Quick access to detailed exception reports without digging through raw logs.

Think of it as a nice-to-have add-on, not a requirement. Exceptional works perfectly well on its own with your chosen storage.

First, let’s clarify: these aren’t mutually exclusive options. MongoDB is a storage medium, while Exceptional is a tool to collect, process, and organize exception logs.

• If you go with just MongoDB: You’ll have to build all the logging logic from scratch—catching exceptions, intercepting DB queries, writing audit logs, handling duplicate entries, and optimizing writes for high concurrency. That’s a lot of custom code, and easy to mess up in high-throughput scenarios.
• If you go with Exceptional + MongoDB: Exceptional takes care of all the exception logging heavy lifting. You just need to configure it to use MongoDB as the backend, then pair it with a general-purpose logging library like Serilog or NLog to handle DB query and audit logs (both can write directly to MongoDB). This way, all your logs live in one place, and you get the benefit of Exceptional’s aggregation and context-rich exception tracking.

For your 200+ concurrent users and high log frequency, MongoDB is a great choice—it’s built for high-write throughput and flexible schemas, which fits the varied structure of exception, query, and audit logs perfectly.

Here’s a step-by-step plan that fits your exact scenario:

• Exception Logging: Use StackExchange.Exceptional in both your .NET Core Web API and WinForms client. Configure it to store exceptions in MongoDB (use the community StackExchange.Exceptional.MongoDB package to add this support). This will automatically capture and aggregate all unhandled exceptions, plus any you log manually.
• Database Query Logging: Add an EF Core DbCommandInterceptor to your Web API to intercept all PostgreSQL queries. Use Serilog to log query text, execution time, parameters, and associated user info directly to MongoDB.
• Audit Logging: Build a simple audit logging service in your business logic layer. Whenever a user performs a critical action (like updating data), call this service to log details (user ID, timestamp, action type, affected data) and write it to MongoDB via Serilog/NLog.
• Optional Monitoring: If your team wants centralized monitoring, deploy OPServer. Configure it to read exception data from your MongoDB instance, and add monitors for your Web API servers and PostgreSQL database. This gives you a single dashboard to track all critical metrics.

If down the line you need more advanced log search and analysis (like full-text search across all logs), you could swap MongoDB for Elasticsearch—Exceptional supports Elasticsearch too, and Serilog/NLog can write to it seamlessly. The tradeoff is slightly higher deployment and maintenance complexity, but better query capabilities.

内容的提问来源于stack exchange，提问作者Jvardas