Resources

Material that sits beside the main docs. Not a tutorial, not a reference page, not a how-to — the pages here are what you reach for when you already know roughly what you want and need confirmation, comparison, or context.

Best practices collects the patterns that fall out of running Flux in production. Workflow shape, task shape, agent defaults, operations posture, what to keep out of a workflow body. Read it once before going live, scan it again when something feels off. Start at Best practices.

Troubleshooting is symptom-first. Find the error message or the misbehavior, get the root cause and the fix. The list is short because most Flux failures share root causes — bootstrap token unset, encryption key missing, a workflow doing non-deterministic work in its body. Start at Troubleshooting.

FAQ answers the questions that come up before installation. Windows support, the smallest deployment, comparison to other engines, Python version requirements. Short answers with links to the page that has the full story. Start at FAQ.

Glossary is the vocabulary index. Workflow, task, agent, replay, idempotency, namespace, principal — fifty short entries, each pointing at the page where the term gets its full treatment. Use it when a doc page assumes a term you haven’t met yet. Start at Glossary.

Performance and benchmarking is honest about what’s measured and what isn’t. Flux 0.56.0 ships measured numbers for runner overhead and dispatch stress tests, but no regression-tested public benchmark suite. The page gives tuning starting points — worker capacity slots, dispatch mode, runner choice, storage pool sizing, trace sample rate — and is candid about where real numbers still need to land. Start at Performance and benchmarking.

Migrating from other engines (Airflow, Prefect, Temporal, Dagster, plain Celery) lives under Migrating from. Contributing to Flux itself lives under Contributing. AI-friendliness — the llms.txt index, how to feed these docs to a model, and how the docs are structured for retrieval — lives under AI-friendliness.

Release notes and the Roadmap track what shipped and what’s queued, and are linked from each topic page where the version matters.