usespitfire/spitfire repository overview

⁠Spitfire — self-hosted distributed load testing

Spitfire is a self-hosted, distributed load testing, stress testing and performance testing platform with a web UI. Build a test step by step in the browser (or write it as JSON), let runners in one or more locations generate the load, and watch requests per second, p95/p99 latency and error rate live — with thresholds that turn every run into a pass/fail result.

🌐 Website: https://usespitfire.com⁠ · 📘 Install guide: https://usespitfire.com/install⁠ · 📚 Guides: https://usespitfire.com/guides⁠ · 🧪 Examples: https://github.com/usespitfire/spitfire-examples⁠

⁠Quick start

Requirements: Docker (plus kubectl for Kubernetes).

# single machine: Postgres + controller + web UI + 2 runners
curl -fsSL https://spitfire.sh | bash -s -- docker

# into the current kubectl context
curl -fsSL https://spitfire.sh | bash -s -- kubernetes

Windows (Docker Desktop, PowerShell): irm https://spitfire.sh/ps | iex

The web UI opens on port 8470.

⁠What it does

• Load test more than HTTP — HTTP/HTTPS (REST, JSON, XML), WebSocket, Server-Sent Events, gRPC, Kafka, MQTT, AMQP / RabbitMQ, Redis, SQL (PostgreSQL, MySQL, SQL Server, Oracle) and MongoDB, all in the same test.
• No scripting required — user journeys with values extracted from responses (JSONPath, XPath, regex, headers), checks on every response, CSV test data, think time.
• Realistic traffic models — virtual users or a fixed arrival rate, stages, several scenarios side by side.
• Distributed and multi-location — split one run across runners and locations (e.g. 60% Istanbul, 40% Frankfurt). Remote runners connect outbound over TLS with key pinning; no inbound ports.
• Thresholds and CI — p(95)<500, rate<0.01; a crossed threshold fails the run and the CI job.
• Findings and the breaking point — the run page lists what its numbers show, and Find the breaking point climbs the load in steps to state your system's capacity.
• Backend view, no agent — OpenTelemetry (OTLP), Prometheus, Tempo, Jaeger and Loki data aligned with the load stages.
• Import — OpenAPI 3 / Swagger 2.0, Postman collections, HAR recordings, k6 scripts and JMeter plans.
• Your data stays in your network — tests, results and encrypted connection credentials live in your own Postgres.

⁠The image

One multi-arch image (linux/amd64, linux/arm64) holds every component:

The image runs as a non-root user (uid 10001). The install script above sets up Postgres, the secrets and the network for you; it is the supported way to run the controller.

Run a test file locally with the CLI, no controller needed (ready-made tests for HTTP, gRPC, Kafka, MQTT, RabbitMQ, Redis and SQL: spitfire-examples⁠):

docker run --rm -v "$PWD:/work" -w /work usespitfire/spitfire spitfire validate test.json
docker run --rm -v "$PWD:/work" -w /work usespitfire/spitfire spitfire run test.json

⁠Tags

• latest — the newest release
• X.Y.Z (e.g. 0.17.0) — a specific release; pin this in production

Release notes: https://usespitfire.com/release-notes⁠

⁠License

Spitfire is commercial software with a free edition (100 virtual users, 15-minute runs, one runner per run; every protocol and CI pass/fail included). Paid plans raise or remove the limits with a license key verified offline — no phone-home. Pricing: https://usespitfire.com/buy⁠