TaskHub.Shared

Observability

Observability is a first-class concern in TaskHub.Shared. Every service is instrumented for logs, metrics, and traces out of the box, with a single shared backend (Loki + Prometheus + Tempo, all behind Grafana).

🎯 Design Goals

1. No service ships without telemetry. The FullHostBuilder wires the full observability stack — opting in is the default, opting out is explicit.
2. One set of conventions across services. Same span names, same metric labels, same log levels — so cross-service dashboards work without per-service tweaks.
3. Three signals, one correlation key. A request’s TraceId lives in every related log line, every span, and every metric exemplar. Click in Grafana, see the full picture.

📦 Modules

Logging

• Logger — Serilog-based structured logging, request logging middleware, sensitive-data redaction.
• Logging overview — what to log, levels, sinks.

Metrics

• Metrics Abstractions — provider-neutral interfaces for counters, gauges, histograms.
• Metrics Implementation — prometheus-net backed implementation with /metrics endpoint.

Tracing

• OpenTelemetry — OTLP exporter, automatic ASP.NET / HttpClient / EF Core / Redis instrumentation, custom sources.
• OpenTelemetry overview — the three signals and how they fit together.

🔗 How They Fit Together

   ┌───────────────────────────────────────────────────────────────┐
   │                        Your Service                            │
   │                                                                │
   │   Serilog ───► Loki         (logs, enriched with TraceId)     │
   │   Activity ──► Tempo        (spans, the request shape)        │
   │   Prometheus ─► Prometheus  (counters/gauges/histograms)      │
   │                                                                │
   └───────────────────────────────────────────────────────────────┘
                                │
                                ▼
                          Grafana (single pane)

The shared TraceId makes the three signals click together in Grafana — go from a slow histogram bucket to the actual log lines of the offending request in two clicks.

✅ When to Reach for Each

Question	Use
“What happened in this one request?”	Logs (with TraceId filter)
“How’s the system doing right now?”	Metrics (rates, latencies, error rates)
“Where is this request spending its time?”	Traces (Tempo span waterfall)
“Did the rate of failures change?”	Metrics + alerts
“Why did this request fail?”	Trace → linked logs

🚀 Bootstrap in One Line

The FullHostBuilder registers the whole stack:

new FullHostBuilder(args).Start<MyDbContext>();
// Includes: AddAppSerilog() + AddAppMetrics() + AddAppOpenTelemetry()

For services using BasicHostBuilder, opt in module by module — see each module’s usage doc.

This site is open source. Improve this page.