traceway: OpenTelemetry-Native Observability in One Platform

Overview

Traceway is an OpenTelemetry-native observability platform built on OpenTelemetry. It brings traces, logs, metrics, exceptions, session replay, AI tracing, and on-call paging together so teams can connect production signals and investigate problems without assembling a separate toolchain. Backend services send telemetry over OTLP/HTTP; browser and mobile apps use Traceway integrations.

It is aimed at teams that want a unified observability service they can self-host or use through Traceway Cloud. The repository includes a Go backend, SvelteKit dashboard, CLI, agent skills, and examples. Its MIT license applies to the project, and the README describes its features as available without an open-core split.

Key Features

• Ingest OpenTelemetry traces, logs, and metrics over OTLP/HTTP, with trace-linked log search and span waterfalls.
• Group exceptions by fingerprint and rank issues, with source maps and mobile symbolication support.
• Track endpoint latency, throughput, and error rates, and define SLOs with error budgets and burn-rate alerts.
• Capture session replays for web and mobile projects to provide context around errors.
• Inspect AI usage, including token and cost data, latency, and conversations.
• Configure on-call rotations, escalation policies, and notifications through supported channels.
• Deploy with Docker Compose, use embedded Go mode, or choose the documented SQLite and DuckDB deployment options.
• Use the read-only traceway CLI and agent skills to query telemetry; write access is a separate opt-in feature.

Use Cases

• An application team wants traces, logs, and metrics connected while diagnosing a production error.
• A team shipping web or mobile apps needs to inspect the user session leading up to a crash or exception.
• Platform engineers want to monitor endpoint performance and alert on service-level objectives.
• Teams operating LLM-backed applications want visibility into AI request costs, tokens, latency, and conversations.
• A Go developer wants to embed a lightweight observability backend in an application rather than deploy the standalone stack.

Getting Started

git clone https://github.com/tracewayapp/traceway
cd traceway && docker compose up -d

Open http://localhost and point an OpenTelemetry exporter at the local OTLP endpoint. See the README and self-hosting documentation for configuration and production deployment details.

Alternatives

• hertzbeat: HertzBeat focuses on agentless infrastructure and service monitoring, metrics, logs, and alerts rather than application traces, errors, and session replay.

Considerations

• Project health is marked Active, with regular commits and releases and ongoing issue and pull-request handling. However, most reported code contributions come from one person, which is a contributor-concentration risk.
• The recommended standalone setup uses Docker Compose and the documented stack includes ClickHouse and PostgreSQL. Embedded SQLite mode is available, with DuckDB as another documented option.
• Profiling is explicitly described as experimental, so teams should assess its suitability before relying on it.
• The README provides test commands, but the supplied requirements data does not indicate tests as a project requirement. Teams adopting it should assess the test coverage relevant to their deployment.
• The repository is MIT licensed. Review the license and deployment documentation for your own distribution and operational requirements.