Results DB Service

Results database service

Model documents

Model documents define the schema for a model's stored results

Files

Results Database Entries

Results objects provide an easy interface with ...

PrefectResults stored in Prefect Core's server

Generic

Impact

Custom

Subclasses of the generic Result must define the model_type as a pydantic field with the alias collection.

class CustomResult(Result):
    """Extends Result base and implements model specific attributes"""

    model_type: str = Field("MyCustomModel", alias="collection")

LUME-services is packaged with seveal pre-configured model results: Impact, Generic

The model categories

Development

In the event that a different result storage scheme would like to be used the steps are as followed: 1. Subclass result - 2. Creation of database service - Implementation of ResultsDB class in lume_services.services.results.db. - Methods should manage connections (multiprocessing and thread-safe) & translate above custom document representations to database

from lume_services.services.results.db import ResultsDB, ResultsDBConfig
from lume_services.services.results.results_service import ResultsDBService

class MyCustomDB(ResultsDB):
    ...

class CustomDBServiceConfig(ResultsDBConfig):
    url: str

custom_db_service_config = CustomDBServiceConfig(
    url="blah://my-connection-url",
)

my_db_service = MyCustomDBService(
    custom_db_service_config
)
results_db_service = ResultsDBService(my_db_service)

Result documents

Results are organized into artifacts called documents

Custom indices

It may be useful to overwrite the indices given in the base class...

User roles

https://www.mongodb.com/docs/manual/core/collection-level-access-control/