rcarmo/gotel

記録された最新のソース活動: 2026年2月7日 13:05

SkillsMP カタログ更新: 2026年11月9日 14:22

収集済み skills: 5

GitHub スター: 6
GitHub フォーク: 0

AI アシスタントでインストール

このプロンプトをコピーして、使用中の AI アシスタントに送ってください。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/rcarmo/gotel.

リポジトリの README

README の元ファイル：README.md

README の元ファイルを読む

README 全文を表示

# gotel ![Gotel Logo](docs/icon-256.png) A small OpenTelemetry collector with SQLite-backed trace storage and a separate Bun-powered web UI for errors, slow operations, trace search, metrics and agent/LLM usage. The native UI links metric buckets to complete traces and exports redacted investigation bundles for local, read-only inspection. It uses the existing Preact frontend, without Jaeger. See [native insights](docs/native-insights.md) for limits and telemetry semantics. ## Quick Start ### Docker Compose ```bash docker compose up -d --build ``` This starts two services: - `gotel` — collector + query API - `web` — web UI proxy/static server Persistent data is stored in the named Docker volume `gotel-data`. Published ports bind to loopback by default; set `GOTEL_BIND_HOST` to opt into remote access. The API has no built-in authentication—see [network access](docs/configuration.md#network-access) before exposing it. Open: - Web UI: http://localhost:3000 - Query API: http://localhost:3200 ### From source ```bash go build -o gotel . ./gotel ``` In a second shell, start the web UI: ```bash make deps-frontend make build-frontend cd web && bun run serve ``` > The collector ships with an embedded default config (OTLP gRPC/HTTP → memory_limiter + batch → SQLite). Add `config.yaml` or set `GOTEL_CONFIG`/`OTEL_CONFIG_FILE` to override it. ## Ports ### Collector | Service | Port | Description | | --- | ---: | --- | | OTLP gRPC | 4317 | Trace ingestion over gRPC | | OTLP HTTP | 4318 | Trace ingestion over HTTP | | Query API | 3200 | Trace and metrics query API | ### Web UI | Service | Port | Description | | --- | ---: | --- | | Web UI | 3000 | Bun web server serving the frontend and proxying API requests | ## Query API Useful endpoints exposed by the collector: ```bash # List traces curl http://localhost:3200/api/traces # Fetch all stored spans for one trace curl http://localhost:3200/api/traces/{traceId}/spans # List recent spans curl http://localhost:3200/api/spans # List exceptions curl http://localhost:3200/api/exceptions # List services curl http://localhost:3200/api/services # Storage statistics curl http://localhost:3200/api/status # Readiness check curl http://localhost:3200/ready ``` ## Generated Metrics Traces are converted to metrics stored in SQLite: ```text otel.<service>.<operation>.span_count otel.<service>.<operation>.duration_ms otel.<service>.<operation>.error_count ``` `duration_ms` is a per-batch average, not a histogram or a percentile. Metric names sanitise service/operation names; original names remain in tags. The native UI computes its own distributions from bounded raw-span queries through `/api/insights`; it does not average these batch averages. ## Validation ```bash make deps-frontend check test-race test-frontend smoke ``` The smoke test starts isolated collector/web processes, ingests a 125-span trace, checks exact OTLP timestamps and resource identity, and verifies query ranges, native aggregates, agent usage, export redaction, frontend bundle validation, browser-origin policy and built assets. ## Documentation - [Configuration](docs/configuration.md) - [Web UI](docs/web-ui.md) - [Sending Traces](docs/sending-traces.md) - [Development](docs/development.md) - [Troubleshooting](docs/troubleshooting.md) ## License MIT

リポジトリの README を読み込んでいます…