Single-binary OTLP generator: traces, logs, metrics, 28-service topology, AI agentic spans
3.6K
One container that emits realistic, topology-rich OpenTelemetry traces, logs and metrics, including AI agentic spans. No microservices to deploy, no Docker Compose with 15 containers. A single 5.7 MB image simulates a full e-commerce platform: up to 28 services, dozens of pods that scale with the topology, 40 scenario flows, and 10 injectable failure modes, with full OTel GenAI semantic conventions for LLM/agent observability.
Built for testing observability platforms, load-testing trace pipelines, and showcasing distributed-system visualizations, for both traditional APM and LLM observability.
Point it at any OTLP/gRPC collector (Jaeger, Tempo, Grafana, an OTel Collector, or a commercial backend):
# Traces to a collector on your host
docker run --rm immersivefusion/tracegen -insecure -endpoint host.docker.internal:4317
# Crank up the volume and inject failures
docker run --rm immersivefusion/tracegen -insecure -endpoint host.docker.internal:4317 -level 8 -errors 5
The image is multi-arch (linux/amd64, linux/arm64), distroless, and runs as non-root.
| Goal | Command |
|---|---|
| AI agentic traces only (LLM/GenAI spans) | -ai-only |
| Traditional APM only (no AI backends) | -no-ai-backends |
| Chaos (failures + unconsumed queues) | -errors 10 -level 8 -no-consumers |
| Flag | Default | Description |
|---|---|---|
-endpoint | localhost:4317 | OTLP gRPC endpoint (host:port) |
-insecure | false | Plaintext gRPC (no TLS) for local/in-cluster backends |
-headers | OTLP headers, key=value comma-separated (e.g. api-key=SECRET) | |
-level | 1 | Aggressiveness 1-10 (1 = whisper, 10 = SCREAM) |
-errors | 0 | Error rate 0-10 (0 = none, 5 = normal, 10 = chaos) |
-complexity | normal | Topology size: light, normal, heavy |
-ai-only | false | Only run AI agentic scenarios |
-no-ai-backends | false | Disable LLM/AI backends (AI spans emit errors) |
-no-consumers | false | Publish to queues but never consume (backlog demo) |
-no-logs | false | Traces only, no OTel log records |
Existing trace generators are either flat span emitters (no service topology) or full demo apps that need Docker Compose and several GB of RAM, and none of them generate AI agentic traces. TraceGen produces topology-rich, failure-injectable traces from a single binary, covering both traditional microservice flows and AI agentic patterns. One image proves a platform can visualize both.
The traces are real, and you can watch them flow. TraceGen feeds live demo grids that stream OpenTelemetry traces into DeepCube (TM)'s 3D player. Watch them live on Twitch: twitch.tv/immersivefusion
latest follows the newest release; each release also publishes a matching semantic-version tag. Pin by digest for reproducible production/cluster runs.github.com/ImmersiveFusion/opentelemetry-tracegen
Apache-2.0. Built by Immersive Fusion.
Content type
Image
Digest
sha256:b46d529f4…
Size
6.4 MB
Last updated
2 months ago
docker pull immersivefusion/tracegen