Issue Summary
The Redis Vector Store node (version 1.3) fails when connecting to AWS ElastiCache with Valkey 8.2 engine due to an automatic TEXT field creation that Valkey doesn’t support.
Error Message
Error: Invalid field type for field `content`: Unknown argument `TEXT`
Environment Details
- n8n Version: 1.117.3 (self-hosted on AWS)
- Redis Vector Store Node: Version 1.3 (Latest)
- Redis Service: AWS ElastiCache with Valkey 8.2 engine
- Operation Mode: Insert Documents
- Index Type: JSON with HNSW vector search
What’s Happening
The Redis Vector Store node automatically attempts to create a TEXT field for document content, regardless of node configuration. This happens even when:
- Content Key option is NOT added to the node
- Only Key Prefix, Metadata Key, and Embedding Key are configured
- Redis index is created with only vector fields
Root Cause
The node appears to hardcode TEXT field creation for `pageContent` or `content` fields from Document Loader input, which is incompatible with Valkey’s field type system.
Expected:
- Node should respect the Redis index schema
- Should only index configured fields (embedding + metadata)
- Content should be stored in JSON without TEXT indexing
Actual:
- Node automatically tries to create TEXT field
- Fails with “Unknown argument TEXT” error
- Cannot store any documents
Redis Index Schema (Working)
FT.CREATE my-semantic-cache
ON JSON
PREFIX 1 vector:
SCHEMA
$.embedding AS embedding VECTOR HNSW 6
TYPE FLOAT32
DIM 1024
DISTANCE_METRIC COSINE
n8n Node Configuration
Redis Vector Store Node:
- Operation Mode: Insert Documents
- Redis Index: `my-semantic-cache
- Key Prefix: `vector`
- Metadata Key: `metadata`
- Embedding Key: `embedding`
- Content Key: **NOT CONFIGURED** (intentionally left empty)
Document Loader Input:
{
"pageContent": "[TENANT:SOMETHING][USER:SOMETHING] something",
"metadata": {
"cache_key": "sem:SOMETHING:SOMETHING:something",
"intent_key": "something",
"tenant": "SOMETHING",
"userType": "SOMETHING"
}
}
Reproduction Steps
1. Create AWS ElastiCache cluster with Valkey 8.2 engine
2. Create Redis Vector index with only vector field (no TEXT field)
3. Configure n8n Redis Vector Store node with Key Prefix, Metadata Key, Embedding Key only
4. Connect Document Loader with pageContent + metadata structure
5. Execute workflow
6. Observe TEXT field error
Workaround Attempted
Tried recreating Redis index with TEXT field:
FT.CREATE index ON JSON SCHEMA
$.content AS content TEXT
$.embedding AS embedding VECTOR ...
Result: Same error - Valkey doesn’t support TEXT field type in the same way as Redis Stack.
Impact
This makes the Redis Vector Store node completely unusable with:
- AWS ElastiCache Valkey 8.2
- Any Redis-compatible service that doesn’t support Redis Stack’s TEXT field syntax
Additional Context
- ElastiCache Valkey supports vector search via RediSearch module
- The vector indexing itself works perfectly (VECTOR HNSW fields)
- Issue is specifically with automatic TEXT field creation for content
- Same workflow works with Redis Stack in other environments
workflow