Python · Docs · KVNest

Each example reads the connection URL from VALKEY_URL; copy it from your instance page in the dashboard. See connection details and limits for what applies to every client.

On this page: valkey-py · FastAPI (asyncio) · Django cache · Celery

valkey-py

The official Valkey client, a fork of redis-py with the same API. redis-py works the same way, with Redis and Valkey instances alike: import redis instead.

Install

pip install valkey

main.py

import os import valkey client = valkey.from_url(os.environ["VALKEY_URL"], decode_responses=True) client.set("greeting", "hello", ex=60)print(client.get("greeting"))

FastAPI (asyncio)

Open one async client for the app's lifetime and share it between requests.

Install

pip install fastapi uvicorn redis

main.py

import osfrom contextlib import asynccontextmanager from fastapi import FastAPIfrom redis.asyncio import Redis  @asynccontextmanagerasync def lifespan(app: FastAPI):    app.state.valkey = Redis.from_url(os.environ["VALKEY_URL"], decode_responses=True)    yield    await app.state.valkey.aclose()  app = FastAPI(lifespan=lifespan)  @app.get("/hits")async def hits():    return {"hits": await app.state.valkey.incr("hits")}

Django cache

Django 4.0+ ships a Redis cache backend; it only needs redis-py.

Install

pip install redis

settings.py

import os CACHES = {    "default": {        "BACKEND": "django.core.cache.backends.redis.RedisCache",        "LOCATION": os.environ["VALKEY_URL"],        "KEY_PREFIX": "myapp",    }} # Optional: keep sessions in the cache too.SESSION_ENGINE = "django.contrib.sessions.backends.cache"

Celery

Celery refuses a rediss:// URL unless it states how to check the certificate, so add ssl_cert_reqs=required.

Install

pip install "celery[redis]"

tasks.py

import os from celery import Celery url = os.environ["VALKEY_URL"] + "?ssl_cert_reqs=required" app = Celery("tasks", broker=url, backend=url)app.conf.broker_transport_options = {"health_check_interval": 30}  @app.taskdef add(x, y):    return x + y

Starter evicts keys when memory is full (allkeys-lru), which can silently drop jobs. Use it for experiments; production queues need a persistent plan with noeviction (Casual or Pro, coming soon).