Distributed n8n deployments can now track instances via Redis

Distributed n8n deployments previously lacked a shared storage mechanism for instance tracking. The existing memory-based storage worked only for single-instance setups—each node maintained its own separate registry, invisible to other instances. This made coordinated multi-instance deployments impossible.

The instance registry now supports a Redis storage backend alongside the existing memory option. Distributed deployments can store instance registrations, heartbeats, and state information in Redis, where all cluster members can read and write the same data. The implementation uses Lua scripts for atomic multi-key operations—registering an instance, reading all active registrations, cleaning up stale entries, and unregistering instances all happen atomically to prevent race conditions across nodes.

This matters for anyone running n8n across multiple servers or in a containerized environment. Load balancers and other coordination tools can now discover active n8n instances by querying the shared registry. The five-minute state TTL enables leadership handoff—if a primary instance fails, others can see which instances were last active before the failure.

The changes live in the @n8n CLI package, specifically within the instance-registry module and its storage layer.