secrets/dataops/qdrant_connection.md

# Qdrant Vector Database Connection Guide

## Connection Details
- **Host**: localhost
- **Port**: 17000
- **Protocol**: HTTP
- **Health Check**: `curl http://localhost:17000/collections`

## Authentication
- No authentication required (localhost binding only)
- All operations are local to the server

## Python Client Example
```python
from qdrant_client import QdrantClient

client = QdrantClient(host="localhost", port=17000)

# Check health
collections = client.get_collections()
print(f"Available collections: {collections}")

# Create collection (if needed)
client.create_collection(
    collection_name="nova_memory",
    vectors_config=VectorParams(size=1536, distance=Distance.COSINE)
)
```

## REST API Examples
```bash
# List collections
curl http://localhost:17000/collections

# Get collection info
curl http://localhost:17000/collections/nova_memory

# Search vectors
curl -X POST http://localhost:17000/collections/nova_memory/points/search \
  -H "Content-Type: application/json" \
  -d '{"vector": [0.1, 0.2, ...], "limit": 10}'
```

## Configuration Notes
- Data directory: `/data/qdrant/storage/`
- Max memory: 50GB (configurable)
- No external network exposure
- Backup location: `/data/adaptai/backups/qdrant/`

## Security
- ❗ Localhost binding only
- ❗ No authentication mechanism
- ❗ Regular backups recommended
- ❗ Monitor disk usage on /data partition

---
**Last Updated:** September 4, 2025