Signoz

SigNoz is the open-source Datadog alternative for full-stack observability — instrument your application with OpenTelemetry (one SDK for traces, metrics, and logs), send telemetry to a self-hosted SigNoz instance, and get distributed tracing, service maps, RED metrics, and log correlation in one dashboard without per-host pricing. For agents: instrument LLM applications with OTel auto-instrumentation, correlate slow LLM calls with infrastructure metrics, and alert on error rate or latency thresholds. SigNoz Cloud available for managed hosting. 20k+ GitHub stars. Confidence is docs-derived.

Instrumentation: OpenTelemetry SDK (any language). Python example: pip install opentelemetry-distro opentelemetry-exporter-otlp; opentelemetry-bootstrap -a install; OTEL_EXPORTER_OTLP_ENDPOINT="http://signoz-host:4317" OTEL_SERVICE_NAME="my-service" opentelemetry-instrument python app.py — all Flask/Django/FastAPI/SQLAlchemy/HTTPX calls auto-traced. Manual spans: from opentelemetry import trace; tracer = trace.get_tracer("my-tracer"); with tracer.start_as_current_span("my-op") as span: span.set_attribute("user.id", "123"); do_work(). Logs: OTEL log export to SigNoz via OTLP exporter. Metrics: OTEL metrics API for custom metrics; auto-collected for instrumented frameworks. SigNoz REST API: GET /api/v1/services lists instrumented services. GET /api/v1/traces retrieves traces with filters. POST /api/v1/channels creates alert notification channel. Alert rules: configured via SigNoz UI or REST API.

Self-hosted: no external auth required (deploy behind your own auth proxy). SigNoz Cloud: login via email/Google SSO; API tokens for programmatic access. OTEL endpoint: configurable auth header (OTEL_EXPORTER_OTLP_HEADERS="signoz-ingestion-key=..."). RBAC: SigNoz v0.23+ adds team roles (admin, editor, viewer). HTTPS: configure at reverse proxy (nginx/Caddy) for self-hosted; enforced on SigNoz Cloud. Data: all telemetry stored in ClickHouse (self-hosted: on your infrastructure; Cloud: SigNoz-managed).

ClickHouse: SigNoz uses ClickHouse for high-throughput trace/metric/log storage — scales to millions of spans/day on modest hardware. Retention: configurable retention policies per telemetry type (default 15 days for traces). Self-hosted resource: ClickHouse is memory-hungry — allocate ≥4GB RAM for production. Docker Compose: suitable for dev; Kubernetes for production. Trace sampling: configurable head/tail sampling to reduce volume. Alert latency: SigNoz alerts run on periodic queries (1-minute granularity). Upgrade path: SigNoz has active release cadence — test upgrades in staging before production.

signoz.io/docs covers OTel instrumentation guides (Python, Go, Java, Node.js, etc.), self-hosting with Docker/K8s, alert configuration, and dashboard guide. Getting started: docker compose up — SigNoz running in 5 minutes; instrument first app in 10 minutes. SigNoz Cloud: free tier (25GB data/month). Self-hosted: fully free, ClickHouse storage. GitHub: SigNoz/signoz (20k+ stars). Community via Slack (very active). Excellent documentation; one of the best-documented open-source observability platforms.