rust-bench documentation

中文版:README_CN.md

This is the unified index for the rust-bench documentation set. It covers both reference docs (factual/structural — what things are) and guides (procedural — how to do things). The two layers cross-link: guides link to reference for background, reference links to guides for runbooks.

For the project overview and quick start, see the root README.md.

Reference docs

Factual/structural descriptions of the system.

  • Glossary — domain terminology (runtime, testing, analysis, job queue, platforms/repos).
  • Architecture — components (Site/Collector/Database/benchlib), multi-platform/multi-repo model, request lifecycle, data model.
  • Configurationsite-config.toml field reference (keys, repos, collectors, per-collector benchmark overrides, env fallback).
  • Job queue — distributed queue mechanics: request states, queue ordering, job enqueue, backfilling, retries.
  • Deployment — deployment model (Docker image, Postgres, collect-job-queue.sh, env vars, machine preparation).
  • Comparison analysis — significance (IQR fence), magnitude buckets, relevance; per-test-result only.
  • Database schema — runtime + job-queue + artifact + error tables (compile-time tables out of scope).
  • Manual modifications — exceptional DB operations (purge_artifact, re-queue, deactivate collector).

Guides

Step-by-step runbooks. Each has prerequisites → steps → verification → troubleshooting.

Scope note

This documentation set covers runtime benchmarks and the distributed benchmarking system. Compile-time benchmark machinery inherited from upstream rustc-perf exists in the codebase but is deliberately out of scope here.