[Bug]: Empty Vectors with Qdrant Configuration

[axunrun]

Do you need to file an issue?

• I have searched the existing issues and this bug is not already filed.
• I believe this is a legitimate bug, not just a question or feature request.

Describe the bug

When using Qdrant as vector storage, documents can be inserted but their vector representations are empty (all zeros). This leads to search responses returning "[no-context]". However, when using PostgreSQL vector storage, vectors are properly stored and search functions normally.

Steps to reproduce

1. Start service with above Qdrant configuration
2. Insert test document via API:
3. Check vector values in Qdrant (all zeros)
4. Execute search request (returns no-context)
5. Switch to PostgreSQL configuration
6. Repeat steps 2-4 (works normally)

Error Logs

[Please provide relevant error logs or exception information]

Expected Behavior

• Documents should be properly processed and stored in Qdrant
• Vector values should be correctly calculated and stored
• Search requests should return relevant results

Actual Behavior

• Documents can be stored in Qdrant
• Vector values are all zeros
• Search requests return "[no-context]" error

Possible Causes

1. Issue with embedding service vector generation
2. Problem in vector data transfer between embedding service and Qdrant
3. Mismatch in vector dimensions or format
4. Configuration issue with Qdrant client
5. Connection problems between services

Additional Notes

• All other storage components (KV, Graph, Doc Status) are functioning normally
• The same embedding model and configuration work correctly with PostgreSQL
• Qdrant server is accessible and running

Questions for Investigation

1. Is the embedding service generating vectors correctly?
2. Are the vectors being properly transferred to Qdrant?
3. Is there any data transformation issue between services?
4. Are there any relevant error logs in the embedding service?

Expected Behavior

No response

LightRAG Config Used

Server Configuration

HOST=0.0.0.0
PORT=9621
WORKERS=1
NAMESPACE_PREFIX=lightrag
MAX_GRAPH_NODES=1000
CORS_ORIGINS=http://localhost:3000,http://localhost:8080

SSL Configuration

SSL=false
SSL_CERTFILE=
SSL_KEYFILE=

Security

LIGHTRAG_API_KEY=

Directory Configuration

WORKING_DIR=/app/data/rag_storage
INPUT_DIR=/app/data/inputs

Ollama Configuration

OLLAMA_EMULATING_MODEL_TAG=latest

Logging Configuration

LOG_LEVEL=INFO
VERBOSE=false
LOG_DIR=/app/data/rag_storage/logs
LOG_MAX_BYTES=10485760
LOG_BACKUP_COUNT=5

LLM Configuration

MAX_ASYNC=4
TIMEOUT=150

RAG Query Settings

HISTORY_TURNS=3
COSINE_THRESHOLD=0.2
TOP_K=60
MAX_TOKEN_TEXT_CHUNK=4000
MAX_TOKEN_RELATION_DESC=4000
MAX_TOKEN_ENTITY_DESC=4000

Document Indexing Settings

CHUNK_SIZE=1200
CHUNK_OVERLAP_SIZE=100
MAX_TOKENS=32768
MAX_TOKEN_SUMMARY=500
SUMMARY_LANGUAGE=Chinese
MAX_EMBED_TOKENS=8192

LLM Binding Configuration

LLM_BINDING=openai
LLM_MODEL=
LLM_BINDING_HOST=
LLM_BINDING_API_KEY=

Embedding Configuration

EMBEDDING_MODEL=bge-m3:latest
EMBEDDING_DIM=1024
EMBEDDING_BINDING=ollama
EMBEDDING_BINDING_HOST=http://192.168.1.99:11434
EMBEDDING_BINDING_API_KEY=

Storage Configuration

LIGHTRAG_KV_STORAGE=PGKVStorage
LIGHTRAG_VECTOR_STORAGE=QdrantVectorDBStorage
LIGHTRAG_GRAPH_STORAGE=Neo4JStorage
LIGHTRAG_DOC_STATUS_STORAGE=PGDocStatusStorage

PostgreSQL Configuration

POSTGRES_HOST=192.168.1.99
POSTGRES_PORT=5432
POSTGRES_USER=postgres
POSTGRES_PASSWORD=
POSTGRES_DATABASE=lightrag

Neo4j Configuration

NEO4J_URI=bolt://192.168.1.99:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=

Qdrant Configuration

QDRANT_URL=http://192.168.1.99:6333

Logs and screenshots

Additional Information

• LightRAG Version:
• Operating System:
• Python Version:
• Related Issues: