How can n8n automatically reconnect to Redis after a container restart?

Below is some context-based guidance:

n8n’s built-in Redis credential support assumes that once the connection is established, it will remain available. If Redis restarts unexpectedly (for example, due to pod restarts), n8n (or more precisely, its underlying ioredis client) may not automatically re-establish the connection. This can result in errors like:

JAVASCRIPTCopied
[ioredis] Unhandled error event: Error: connect ECONNREFUSED 10.4.34.46:6379

At this time there isn’t a built-in auto-reconnection mechanism in n8n that can automatically recover from a lost Redis connection. As you’ve discovered, a common workaround is manually triggering a restart (or “rollout”) of the n8n deployment so that it reconnects once Redis is available.

Here are a couple of strategies you could consider:

1. External Monitoring & Probes:
   Use Kubernetes readiness or liveness probes that can detect when n8n is unable to connect to Redis. With a properly tuned probe, Kubernetes could automatically restart the pod if it detects that n8n is in a failed state due to a lost connection. This proactive restart mimics the manual “rollout restart” workaround.

2. Sidecar or Wrapper Approach:
   Consider running a sidecar or wrapping the Redis client with additional logic (if you have a custom deployment) to detect and handle reconnections more gracefully. Although this isn’t supported out-of-the-box by n8n, it’s a technique used in other environments to reduce downtime.

Until n8n includes more robust reconnection logic, these external operational measures are the recommended approaches. Additionally, ensure that your Redis credentials (host, port, password, etc.) are configured correctly per the Redis credentials documentation so that any issues aren’t compounded by misconfiguration.

Hope this helps clarify the current behavior and potential workarounds!