fix: Add option for disabling ownership extraction

datahub-web-react/src/app/ingest/source/builder/sources.json

@@ -309,7 +309,7 @@
         "displayName": "Dremio",
         "description": "Import Spaces, Sources, Tables and statistics from Dremio.",
         "docsUrl": "https://datahubproject.io/docs/metadata-ingestion/",
-        "recipe": "source:\n    type: dremio\n    config:\n        # Coordinates\n        hostname: null\n        port: null\n        #true if https, otherwise false\n        tls: true\n\n        #For cloud instance\n        #is_dremio_cloud: True\n        #dremio_cloud_project_id: <project_id>\n\n        #Credentials with personal access token\n        authentication_method: PAT\n        password: pass\n\n        #Or Credentials with basic auth\n        #authentication_method: password\n        #username: null\n        #password: null\n\n        stateful_ingestion:\n            enabled: true"
+        "recipe": "source:\n    type: dremio\n    config:\n        # Coordinates\n        hostname: null\n        port: null\n        #true if https, otherwise false\n        tls: true\n\n        #For cloud instance\n        #is_dremio_cloud: True\n        #dremio_cloud_project_id: <project_id>\n\n        #Credentials with personal access token\n        authentication_method: PAT\n        password: pass\n\n        #Or Credentials with basic auth\n        #authentication_method: password\n        #username: null\n        #password: null\n\n       ingest_owner: true\n\n        stateful_ingestion:\n            enabled: true"
     },
     {
         "urn": "urn:li:dataPlatform:cassandra",

metadata-ingestion/docs/sources/dremio/dremio_recipe.yml

@@ -20,6 +20,8 @@ source:
 
     include_query_lineage: True
 
+    ingest_owner: true
+
     #Optional
     source_mappings:
       - platform: s3

metadata-ingestion/src/datahub/ingestion/source/dremio/dremio_aspects.py

@@ -142,6 +142,7 @@ def __init__(
         platform: str,
         ui_url: str,
         env: str,
+        ingest_owner: bool,
         domain: Optional[str] = None,
         platform_instance: Optional[str] = None,
     ):
@@ -150,6 +151,7 @@ def __init__(
         self.env = env
         self.domain = domain
         self.ui_url = ui_url
+        self.ingest_owner = ingest_owner
 
     def get_container_key(
         self, name: Optional[str], path: Optional[List[str]]
@@ -426,21 +428,23 @@ def _create_external_url(self, dataset: DremioDataset) -> str:
         return f'{self.ui_url}/{container_type}/{dataset_url_path}"{dataset.resource_name}"'
 
     def _create_ownership(self, dataset: DremioDataset) -> Optional[OwnershipClass]:
-        if not dataset.owner:
-            return None
-        owner = (
-            make_user_urn(dataset.owner)
-            if dataset.owner_type == "USER"
-            else make_group_urn(dataset.owner)
-        )
-        return OwnershipClass(
-            owners=[
-                OwnerClass(
-                    owner=owner,
-                    type=OwnershipTypeClass.TECHNICAL_OWNER,
-                )
-            ]
-        )
+        if self.ingest_owner and dataset.owner:
+            owner_urn = (
+                make_user_urn(dataset.owner)
+                if dataset.owner_type == "USER"
+                else make_group_urn(dataset.owner)
+            )
+            ownership: OwnershipClass = OwnershipClass(
+                owners=[
+                    OwnerClass(
+                        owner=owner_urn,
+                        type=OwnershipTypeClass.TECHNICAL_OWNER,
+                    )
+                ]
+            )
+            return ownership
+
+        return None
 
     def _create_glossary_terms(self, entity: DremioDataset) -> GlossaryTermsClass:
         return GlossaryTermsClass(

metadata-ingestion/src/datahub/ingestion/source/dremio/dremio_config.py

@@ -174,3 +174,8 @@ def is_profiling_enabled(self) -> bool:
         default=False,
         description="Whether to include query-based lineage information.",
     )
+
+    ingest_owner: bool = Field(
+        default=True,
+        description="Ingest Owner from source. This will override Owner info entered from UI",
+    )

metadata-ingestion/src/datahub/ingestion/source/dremio/dremio_source.py

@@ -123,6 +123,7 @@ def __init__(self, config: DremioSourceConfig, ctx: PipelineContext):
         self.dremio_aspects = DremioAspects(
             platform=self.get_platform(),
             domain=self.config.domain,
+            ingest_owner=self.config.ingest_owner,
             platform_instance=self.config.platform_instance,
             env=self.config.env,
             ui_url=dremio_api.ui_url,
@@ -569,11 +570,14 @@ def process_query(self, query: DremioQuery) -> None:
             )
 
         # Add observed query
+
         self.sql_parsing_aggregator.add_observed_query(
             ObservedQuery(
                 query=query.query,
                 timestamp=query.submitted_ts,
-                user=CorpUserUrn(username=query.username),
+                user=CorpUserUrn(username=query.username)
+                if self.config.ingest_owner
+                else None,
                 default_db=self.default_db,
             )
         )