analytics-assets

Analytics Asset Facades

OpenMetadata 2.0 RC1 only. This live example uses the exact dependency in requirements-sdk.txt; automated validation is offline, not server-certified.

This scenario directly demonstrates DashboardServices, Charts, DashboardDataModels, Dashboards, MLModels, Pipelines, Queries, and SearchIndexes. It creates a custom dashboard service, linked chart/data model/dashboard, representative ML model, three-task pipeline with one status, SQL query, and typed search index.

OpenMetadata 2.0 RC1 does not expose public service facades for the ML model, pipeline, or search services. Create those services first and supply their fully qualified names. Supply an existing database service for the query.

Pipeline Service Versus Pipeline Asset

A Custom Pipeline service is the parent connection; it is not itself a Pipeline asset. For example:

Service FQN:  custom-pipeline-m1
Pipeline FQN: custom-pipeline-m1.my_daily_etl

Creating only custom-pipeline-m1 does not add that service—or a Pipeline beneath it—to Pipelines.list() or Pipelines.list_all(). A connector or Pipelines.create() call must create my_daily_etl beneath that service. This scenario demonstrates that second step using the service named by OPENMETADATA_PIPELINE_SERVICE_FQN.

Pipelines.list() returns one page, whose assets are in .entities; Pipelines.list_all() follows pagination and returns every Pipeline asset. Neither method lists Pipeline services. See the pinned RC1 Pipelines facade and list()/list_all() implementation.

Run

python -m pip install -r requirements-sdk.txt
export OPENMETADATA_HOST=http://localhost:8585/api
export OPENMETADATA_JWT_TOKEN='<token>'
export OPENMETADATA_DEMO_NAMESPACE=demo
export OPENMETADATA_DATABASE_SERVICE_FQN=sample_database_service
export OPENMETADATA_ML_MODEL_SERVICE_FQN=sample_ml_service
export OPENMETADATA_PIPELINE_SERVICE_FQN=sample_pipeline_service
export OPENMETADATA_SEARCH_SERVICE_FQN=sample_search_service
python analytics-assets/analytics_assets.py

The namespace must match ^[a-z][a-z0-9_]{2,31}$. The script reads credentials only from the environment. It prints safe entity FQNs and IDs, not connections or tokens.

Facade.create() performs a full PUT upsert. Every entity created by this scenario carries [openmetadata-demo:<namespace>:analytics-assets]; reruns are allowed only for that exact marker. Foreign collisions abort. No cleanup, restore, search reindex, or global settings mutation is included.