FAQ¶
Is Palfrey a drop-in replacement for Uvicorn?¶
Palfrey is built for Uvicorn-compatible operational behavior and CLI shape in covered paths. Use staging validation for your exact workload and options.
Should I use reload and workers together?¶
Treat them as separate modes:
- reload for development
- workers for production scale
Which loop should I use?¶
Default to --loop auto unless you need strict platform-specific reproducibility.
Which websocket mode should I use?¶
Start with default mode unless you have a measured reason to pin a specific backend.
How do I run behind a reverse proxy?¶
Enable --proxy-headers and set --forwarded-allow-ips to trusted proxy sources.
How do I improve performance safely?¶
- benchmark your own workload
- change one runtime variable at a time
- keep benchmark commands and environment details versioned
How do I run Palfrey in Python code?¶
Use programmatic startup patterns from quickstart/reference pages.
Where do I report bugs?¶
Use the project issue tracker with reproducible command, environment, and traceback details.
Non-technical summary¶
FAQ answers are decision shortcuts. Use them to pick a safe default, then validate in your own environment.