Stampede documentation

Stampede describes how your users behave, checks each journey with a single user, then runs many of them at once and tells you where your product breaks. Everything here describes what the repository does today; features that are not built yet are marked planned.

Start

Quick start: ShopLab, a first run, a report, a breakpoint, in about five minutes
Installation: binary, Docker Compose, Kubernetes
Your first real target
Using Stampede from the CLI: setup, projects, targets, runs, reports, schedules, AI and notifications from the terminal

Everything is done with the stampede CLI. A server’s web UI is read only, for analysis and reporting, apart from the Stop, Kill and Kill all safety controls.

Concepts

Scenarios and journeys
Open and closed load models
How Stampede measures: coordinated omission, histograms, saturation

Guides

Test types: every traffic shape and the targeted stresses
Reading a report
Comparing releases
CI integration: exit codes, JUnit, the GitHub Action
Scheduled runs: cron schedules, time zones, skips and who can change them
Integrations: Prometheus target metrics, trace links, notifications
Fault injection: break dependencies on purpose during a run
Coverage and drift: endpoints no journey tests, journeys an API change broke
AI journey generation
Product packs: using and writing them
Protocols: HTTP/1.1, HTTP/2, GraphQL, WebSocket, SSE, gRPC
Plugins: MQTT, Kafka, Redis, SQL and UDP steps; installing plugins and writing your own

Operate

Docker Compose, Helm, operator, Terraform workers
Air-gapped install
Upgrades, backups and restore
Safety: ownership checks, caps, kill switch
Configuration
Threat model

Reference

Scenario format and its JSON Schema
CLI (generated from the binary)
REST API (generated from api/openapi.yaml, which the server also serves at /api/v1/openapi.yaml)