Skip to content

README.md

Latest commit

 

History

History
77 lines (59 loc) · 2.89 KB

README.md

File metadata and controls

77 lines (59 loc) · 2.89 KB
Raw

Praxis Benchmarks

praxis-bench is the shared benchmarking and performance-testing tool for Praxis. It load-tests a Praxis (or compatible) proxy image and produces comparison reports against Envoy, NGINX, and HAProxy baselines, so any repository with a standard proxy build (praxis, ai, experimental) can measure it the same way.

The proxy under test is treated as an opaque container image: plug in your image with --image and get results. The tool does not depend on Praxis source.

Usage

Run the CLI directly:

cargo run -p praxis-bench -- --image ghcr.io/praxis-proxy/praxis:latest

Or use the container runner, which bundles the docker CLI and the vegeta/fortio load generators. Share the host Docker socket and network so it can start proxy containers and reach them:

docker run --rm --network host \
  -v /var/run/docker.sock:/var/run/docker.sock \
  ghcr.io/praxis-proxy/benchmarks --image ghcr.io/praxis-proxy/praxis:latest

Subcommands:

praxis-bench                 # run benchmarks, write a report
praxis-bench visualize ...   # render an SVG chart from a report
praxis-bench compare A B     # compare two reports for regressions

A full run needs a Docker (or Podman) daemon and the vegeta and fortio load generators on PATH; both are bundled in the container image.

What Is Enforced

This repository follows the praxis-proxy conventions: the same machine-enforced quality standards used across the organization.

  • Lints: ~200 rustc/clippy/rustdoc lints at deny, including no unchecked arithmetic, no as casts, no unwrap/panic, exhaustive enum matching, and documentation on every item (public and private)
  • Testing: unit + integration tests required; 90% line / 80% region coverage floor; mutation testing; property-based testing conventions
  • Supply chain: cargo audit + cargo deny with pinned registries and a license allowlist
  • Reviewability: PRs capped at 750 added production lines, with required descriptions, conventional commits, DCO sign-off, and signed commits
  • Everything else: markdown, TOML, shell, spelling, and workflow files are linted too

Development

Install the requirements, then:

make all            # build + fmt + lint + test + audit
make help           # every available target

Documentation

Document Contents
conventions.md Coding style, testing, type design, lint policy
development.md Requirements, build/test/coverage commands
enhancements Proposal lifecycle for larger changes
release.md Versioning, tagging, release pipeline
CONTRIBUTING.md Contributor entry point and PR gates