Retry potentially transient HTTP errors in Python.
See documentation.
Several HTTP errors are often transient, and might succeed if retried:
- HTTP status codes
429 Too Many Requests
(rate limited)500 Internal Server Error
502 Bad Gateway
503 Service Unavailable
504 Gateway Timeout
- Network errors
- Timeouts
This project aims to simplify retrying these, by extending tenacity
with custom retry and wait strategies, as well as a custom decorator. Defaults are sensible for most use cases, but are fully customizable.
Supports both requests
and httpx
natively, but could be customized to use with any library that raises exceptions for the conditions listed above.
Install from PyPI:
pip install retryhttp # Supports both HTTPX and requests
You can also install support for only HTTPX or requests, if you would rather not install unnecessary dependencies:
pip install retryhttp[httpx] # Supports only HTTPX
pip install retryhttp[requests] # Supports only requests
Or, install the latest development snapshot from git:
pip install git+https://github.com/austind/retryhttp.git@develop
import httpx
from retryhttp import retry
# Retries safely retryable status codes (429, 500, 502, 503, 504), network errors,
# and timeouts, up to a total of 3 times, with appropriate wait strategies for each
# type of error. All of these behaviors are customizable.
@retry
def example():
response = httpx.get("https://example.com/")
response.raise_for_status()
return response.text
Contributions welcome! Bug fixes and minor tweaks can jump straight to a pull request. For more involved changes, open an issue and let's chat about your idea. Thanks for your contribution!