Quickstart
Install one SDK, connect one runtime, validate ingestion, and prepare Obtrace for remediation workflows.
Quickstart
Obtrace is most useful when telemetry, release metadata, and repository context are connected early. This quickstart focuses on the shortest path to a production-ready signal baseline.
Outcome
By the end of this guide, you should have:
- Every detected stack instrumented (backend, frontend, edge functions).
- Stable
service,env, andversiontagging. - One runtime environment connected.
- Enough context for root cause analysis and remediation decisions.
Step 1: Instrument your service
Option A: Obtrace Zero (zero-code)
The fastest path. Obtrace Zero auto-instruments your services without code changes — using eBPF, language agents, or sidecar injection depending on your runtime.
- Obtrace Zero Quickstart
- How it works
- Language Agents — per-language breakdown of what gets captured automatically
- Instrumentation Strategies
Supports Node.js, Python, Java, .NET, PHP, Ruby, and Go out of the box.
Option B: SDK (manual instrumentation)
For full control over what gets captured, use an SDK directly:
Backend:
Frontend:
Fullstack frameworks:
- Next.js — server + client in one setup
Platforms & Edge:
Multi-stack projects
Obtrace auto-detects all stacks in your repository. If your project combines multiple runtimes, you get a setup snippet for each stack. Instrument all of them for full end-to-end visibility.
Common combinations:
- React + Supabase → Browser SDK + Supabase Managed Agent
- Next.js → single Next.js SDK (covers server + client)
- Vite + Express → Browser SDK + Node SDK
- React + Cloudflare Workers → Browser SDK + Cloudflare SDK
Agents
- Database Agent — collect database metrics from PostgreSQL, MySQL, MongoDB, Redis, and more without code changes
Step 2: Set canonical attributes
Every event, span, and error should include:
serviceenvversionregionwhen applicable
appId and serviceName must resolve to the connected app in the project. You can configure explicit aliases such as core -> web, but arbitrary names are rejected because incident scope, replay, and remediation bind to the canonical project app.
Example:
Step 3: Connect the environment
Start with the runtime where the incident actually happens:
Step 4: Validate ingestion
Expected result:
- Telemetry is continuous.
- Tags are stable across deploys.
- Errors contain enough stack and deploy context for diagnosis.
Step 5: Review how remediation works
Before enabling repository actions, read: