All 16 Diagnostics & Performance questions
More in Diagnostics & Performance
What is the Diagnostics IPC channel, and how do diagnostic tools like dotnet-trace connect to a running .NET process without a debugger? How does dotnet-counters provide live performance monitoring, and what key metrics does it expose by default? What is Application Insights (or OpenTelemetry) integration, and what value does it add over basic logging? What is BenchmarkDotNet, and why is it preferred over manual Stopwatch-based timing for measuring code performance? What is a memory dump, and how do you capture and analyze one for a .NET application experiencing high memory usage or a crash? What is the purpose of health check response caching and how does frequent health check polling itself impact application performance? What is the Large Object Heap (LOH), and why do objects larger than 85,000 bytes behave differently in the garbage collector? How does the dotnet-gcdump tool help diagnose managed memory leaks, and how do you analyze the resulting heap snapshot? How do EventCounters and the newer System.Diagnostics.Metrics API (Meter) let you expose custom application-specific performance metrics? How would you diagnose a thread pool starvation issue in a production ASP.NET Core application?
Advertisement
.NET › Diagnostics & Performance

What is the purpose of the dotnet-monitor tool, and how does it enable diagnostics in containerized/production environments?

Intermediate

Answer

dotnet-monitor runs as a sidecar or standalone process alongside your application, exposing an HTTP API to collect diagnostics artifacts (traces, dumps, logs, metrics) on demand or automatically when trigger rules match (like high CPU or an exception occurring), without needing direct shell access to the container or manually installing diagnostic tools inside a production container -- particularly valuable in Kubernetes environments where interactive debugging access is often restricted.
# Running as a sidecar in Kubernetes, collecting a trace via its HTTP API:
curl -X POST "http://localhost:52323/trace?pid=1&durationSeconds=30" -o trace.nettrace

# Or configured to auto-collect a dump when CPU exceeds a threshold for 30 seconds
{ "CollectionRuleDefaults": { "Triggers": { "CPUUsage": { "GreaterThan": 90, "SlidingWindowDuration": "00:00:30" } } } }
Real-world example A Kubernetes-hosted microservices platform deploys dotnet-monitor as a sidecar in every pod, allowing on-call engineers to remotely capture a memory dump from a misbehaving production pod via a simple HTTP call, without needing kubectl exec shell access.

Common follow-ups: How do collection rules automate diagnostics without human intervention?;What security considerations apply to exposing dotnet-monitor's API?

Docker & Containerization;Health Checks & Readiness/Liveness Probes
All 16 Diagnostics & Performance questions