What the errors look like
The three you will see most, each pointing at a different cause:
Error: connect ECONNREFUSED 127.0.0.1:6379 # not reachable
ReplyError: NOAUTH Authentication required. # password missing
Error: Connection is closed. # TLS/scheme or dropped linkECONNREFUSED means nothing is listening there; NOAUTH means it is reachable but you did not authenticate; a closed connection to a managed host usually means you used redis:// where rediss:// (TLS) was required.
Why it happens
Redis not running / wrong host
No redis-server at that host and port, or the URL points somewhere wrong.
localhost inside Docker
localhost in a container is the container itself - use the Compose service name as the host.
NOAUTH - password missing
Redis has requirepass set but the client did not send a password.
Managed Redis needs TLS
Upstash and similar require rediss:// (TLS), not redis://.
Diagnose it in three steps
Is Redis reachable?
redis-cli -h 127.0.0.1 -p 6379 ping # expect: PONGDoes it need a password?
redis-cli -h host -a "$REDIS_PASSWORD" ping
# NOAUTH without -a means requirepass is set - supply the password.Managed host? check the scheme and TLS
# Upstash and friends need TLS:
REDIS_URL=rediss://default:PASSWORD@host:6379 # note: rediss (two s)Reach it, authenticate, use the right scheme
Make Redis reachable from the app - the right host (a service name in Docker), the password supplied, and TLS for a managed endpoint. Then add a retry strategy so a brief blip does not crash the app.
# Docker - point at the service, not localhost
REDIS_URL=redis://redis:6379
# Managed (TLS) - rediss:// and the password
REDIS_URL=rediss://default:PASSWORD@host:6379
// ioredis - URL plus a sane retry strategy
const Redis = require('ioredis');
const redis = new Redis(process.env.REDIS_URL, {
retryStrategy: (times) => Math.min(times * 200, 2000),
});If the app starts before Redis is ready, the retry strategy reconnects; for a hard dependency, order startup so Redis is reachable before the app accepts traffic.
Order startup, health-check the cache
An app that connects to Redis before it is reachable is a startup-ordering problem. On your own servers, Infraveil health-checks dependencies and coordinates how services start, so the app does not go live before Redis is reachable - and connection failures surface with context, with configured, bounded recovery automation and selected operational records.
Frequently asked questions
Why connect ECONNREFUSED 127.0.0.1:6379?
Nothing is listening there: Redis is not running locally, or inside Docker you used localhost, which is the app container itself. Use the Compose service name as the host, e.g. redis://redis:6379.
What does NOAUTH Authentication required mean?
Redis has requirepass set but your client did not send a password. Provide it in the URL (redis://default:PASSWORD@host:6379) or client config; include a username if you use Redis 6+ ACL users.
Why does my managed Redis refuse to connect?
Managed Redis usually requires TLS. Use the rediss:// scheme (two s characters), not redis://, and supply the password. Plain redis:// to a TLS-only endpoint fails or hangs.
How do I connect to Redis from Docker?
Use the Compose service name as the host - redis://redis:6379 - not localhost, which inside a container points at the container itself rather than Redis.
Client-side, no signup — they run in your browser.