pragmatiks-qdrant-provider 0.16.0

Qdrant provider for Pragmatiks

Qdrant Provider

Qdrant vector database provider for Pragmatiks.

Resources

database

Deploys a Qdrant database to a GKE cluster with external LoadBalancer access.

provider: qdrant
resource: database
name: my-qdrant
config:
  cluster:
    provider: gcp
    resource: gke
    name: my-cluster
  replicas: 1
  generate_api_key: true
  storage:
    size: 20Gi
    class: premium-rwo
  resources:
    memory: 4Gi
    cpu: "2"

Configuration

Field	Type	Required	Description
cluster	Dependency[GKE]	Yes	GKE cluster to deploy to
replicas	int	No	Number of replicas (default: 1)
image	str	No	Docker image (default: qdrant/qdrant:latest)
api_key	str	No	API key for authentication (mutually exclusive with generate_api_key)
generate_api_key	bool	No	Generate a secure 32-char API key (default: false)
storage	StorageConfig	No	Persistent storage configuration
resources	ResourceConfig	No	CPU and memory limits

StorageConfig

Field	Type	Required	Description
size	str	No	Volume size (default: 10Gi)
class	str	No	Storage class (default: standard-rwo)

ResourceConfig

Field	Type	Required	Description
memory	str	No	Memory limit (default: 2Gi)
cpu	str	No	CPU limit (default: 1)

Outputs

Field	Type	Description
url	str	HTTP endpoint (e.g., http://34.x.x.x:6333)
grpc_url	str	gRPC endpoint (e.g., http://34.x.x.x:6334)
api_key	str | None	API key if configured
ready	bool	Whether database is ready

---

collection

Manages Qdrant vector collections.

apiVersion: qdrant/v1
kind: collection
metadata:
  name: company-docs
  namespace: demo
spec:
  api_key: $ref{qdrant-secret.data.api_key}  # Optional for local
  url: https://xyz.qdrant.io:6333  # Or http://localhost:6333
  name: company-docs
  vectors:
    size: 1536
    distance: Cosine  # Cosine, Euclid, Dot
  on_disk: true  # Optional

Configuration

Field	Type	Required	Description
api_key	Field[str]	No	API key for Qdrant Cloud (optional for local)
url	str	No	Qdrant server URL (default: http://localhost:6333)
name	str	Yes	Collection name
vectors	VectorConfig	Yes	Vector configuration
on_disk	bool	No	Store vectors on disk (default: false)

VectorConfig

Field	Type	Required	Description
size	int	Yes	Vector dimension
distance	str	No	Distance metric: Cosine, Euclid, Dot (default: Cosine)

Outputs

Field	Type	Description
name	str	Collection name
indexed_vectors_count	int	Number of indexed vectors
points_count	int	Number of points in collection
status	str	Collection status (green, yellow, red)

Lifecycle

• on_create: Creates collection if it doesn't exist
• on_update: Recreates collection if vector config changes (destructive)
• on_delete: Deletes collection

Installation

pip install pragmatiks-qdrant-provider

Development

# Run tests
task qdrant:test

# Format code
task qdrant:format

# Lint and type check
task qdrant:check