# torchio

> Break your containers on purpose: burn CPU, hold memory, fill disk and more, on demand over HTTP.

torchio 4.6.2 is a single static binary written in Rust that burns CPU, holds memory, fills disk, answers slowly or returns any HTTP status, on demand over HTTP. Use it to test autoscaling, resource limits, OOM kills, probes, timeouts and retries. A running server describes its own endpoints, parameters and limits as JSON at `GET /help`, and what it holds right now at `GET /info`; prefer those to guessing. Every endpoint that changes state is also a command: `torchio cpu 80 --hold 60`. The whole documentation in one file is https://torchio.altr.dev/llms-full.txt.

## About

- [Introduction](https://torchio.altr.dev/introduction.html): What torchio is and what it is for

## Use it

- [Install](https://torchio.altr.dev/install.html): Docker, Homebrew, the install script, release binaries, building from source
- [Try it in 30 seconds](https://torchio.altr.dev/getting-started.html): Start a server and watch it in 30 seconds
- [What you can test](https://torchio.altr.dev/what-you-can-test.html): Recipes: OOM kills, autoscaling, probes, timeouts, disk, leaks, readiness, restarts, descriptors
- [Endpoints](https://torchio.altr.dev/endpoints.html): What each endpoint does, with examples and answers
- [Reference](https://torchio.altr.dev/reference.html): Every endpoint with its parameters, defaults, limits and command, written from the same catalog as GET /help
- [Starting a test](https://torchio.altr.dev/starting-a-test.html): The torchio commands that call the endpoints of a running server, including scenario and payload
- [Scenarios](https://torchio.altr.dev/scenarios.html): POST /scenario: a list of steps (cpu, memory, disk, fd, connections, ready, block) run one after the other
- [Shell completion](https://torchio.altr.dev/shell-completion.html): Completion scripts for bash, zsh and fish
- [For AI agents](https://torchio.altr.dev/agents.html): How an agent can use torchio: /help, llms.txt and the skill

## Watch it

- [Watching a server](https://torchio.altr.dev/watch.html): torchio watch: a terminal view of a running server that can also start tests
- [Control panel](https://torchio.altr.dev/control-panel.html): The page at /ui to watch a server and start tests with sliders

## Configure it

- [Configuration](https://torchio.altr.dev/configuration.html): Every environment variable, its default and its limit
- [Token](https://torchio.altr.dev/token.html): TOKEN: which endpoints need a bearer token
- [Logs](https://torchio.altr.dev/logs.html): Plain text and JSON logs (LOG_FORMAT), the access log
- [Shutdown](https://torchio.altr.dev/shutdown.html): What happens on SIGTERM and SIGINT, and the drain

## Run it

- [Docker](https://torchio.altr.dev/docker.html): The image, its tags, platforms, limits and health check
- [Examples](https://torchio.altr.dev/examples.html): Kubernetes manifests, the Helm chart and Compose files

## The project

- [How it works](https://torchio.altr.dev/how-it-works.html): Worker threads and the async runtime
- [Load testing](https://torchio.altr.dev/load-testing.html): Using torchio with hey and k6 to load several cores
- [Development](https://torchio.altr.dev/development.html): Source layout, tests, documentation site and releases
- [Changelog](https://torchio.altr.dev/changelog.html)
- [Security and license](https://torchio.altr.dev/security.html): Where it is safe to run it, and the license

## Optional

- [Skill for agents](https://torchio.altr.dev/SKILL.md): how to run a test with torchio, as a skill file