perf: skip unnecessary metadata fetch calls for tags when not configured

dbt/adapters/databricks/impl.py

@@ -81,6 +81,9 @@
     StreamingTableConfig,
 )
 from dbt.adapters.databricks.relation_configs.table_format import TableFormat
+from dbt.adapters.databricks.relation_configs.tags import (
+    TagsProcessor,
+)
 from dbt.adapters.databricks.relation_configs.tblproperties import TblPropertiesConfig
 from dbt.adapters.databricks.relation_configs.view import ViewConfig
 from dbt.adapters.databricks.utils import (
@@ -967,17 +970,21 @@ def parse_columns_and_constraints(
         return enriched_columns, parsed_constraints
 
     @available.parse(lambda *a, **k: {})
-    def get_relation_config(self, relation: DatabricksRelation) -> DatabricksRelationConfigBase:
+    def get_relation_config(
+        self,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
+    ) -> DatabricksRelationConfigBase:
         if relation.type == DatabricksRelationType.MaterializedView:
-            return MaterializedViewAPI.get_from_relation(self, relation)
+            return MaterializedViewAPI.get_from_relation(self, relation, model_config)
         elif relation.type == DatabricksRelationType.StreamingTable:
-            return StreamingTableAPI.get_from_relation(self, relation)
+            return StreamingTableAPI.get_from_relation(self, relation, model_config)
         elif relation.type == DatabricksRelationType.Table:
-            return IncrementalTableAPI.get_from_relation(self, relation)
+            return IncrementalTableAPI.get_from_relation(self, relation, model_config)
         elif relation.type == DatabricksRelationType.View:
-            return ViewAPI.get_from_relation(self, relation)
+            return ViewAPI.get_from_relation(self, relation, model_config)
         elif relation.type == DatabricksRelationType.MetricView:
-            return MetricViewAPI.get_from_relation(self, relation)
+            return MetricViewAPI.get_from_relation(self, relation, model_config)
         else:
             raise NotImplementedError(f"Relation type {relation.type} is not supported.")
 
@@ -1060,12 +1067,15 @@ def config_type(cls) -> type[DatabricksRelationConfig]:
 
     @classmethod
     def get_from_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> DatabricksRelationConfig:
         """Get the relation config from the relation."""
 
         assert relation.type == cls.relation_type
-        results = cls._describe_relation(adapter, relation)
+        results = cls._describe_relation(adapter, relation, model_config)
         return cls.config_type().from_results(results)
 
     @classmethod
@@ -1077,7 +1087,10 @@ def get_from_relation_config(cls, relation_config: RelationConfig) -> Databricks
     @classmethod
     @abstractmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         """Describe the relation and return the results."""
 
@@ -1087,11 +1100,14 @@ def _describe_relation(
 class DeltaLiveTableAPIBase(RelationAPIBase[DatabricksRelationConfig]):
     @classmethod
     def get_from_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> DatabricksRelationConfig:
         """Get the relation config from the relation."""
 
-        relation_config = super().get_from_relation(adapter, relation)
+        relation_config = super().get_from_relation(adapter, relation, model_config)
 
         # Ensure any current refreshes are completed before returning the relation config
         tblproperties = cast(TblPropertiesConfig, relation_config.config["tblproperties"])
@@ -1111,7 +1127,10 @@ def config_type(cls) -> type[MaterializedViewConfig]:
 
     @classmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         kwargs = {"table_name": relation}
         results: RelationResults = dict()
@@ -1123,9 +1142,18 @@ def _describe_relation(
         results["information_schema.views"] = get_first_row(
             adapter.execute_macro("get_view_description", kwargs=kwargs)
         )
-        results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
         results["show_tblproperties"] = adapter.execute_macro("fetch_tbl_properties", kwargs=kwargs)
+
+        # To be backward compatible model_config can be None. In that case, tags should be fetched
+        # to maintain backward compatibility.
+        table_tag_config = model_config.config.get(TagsProcessor.name) if model_config else None
+        if table_tag_config is None or table_tag_config.requires_server_metadata_for_diff():
+            results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
+        else:
+            results["information_schema.tags"] = None
+
         results["row_filters"] = adapter.execute_macro("fetch_row_filters", kwargs=kwargs)
+
         return results
 
 
@@ -1138,7 +1166,10 @@ def config_type(cls) -> type[StreamingTableConfig]:
 
     @classmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         kwargs = {"table_name": relation}
         results: RelationResults = dict()
@@ -1162,16 +1193,37 @@ def config_type(cls) -> type[IncrementalTableConfig]:
 
     @classmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         results = {}
         kwargs = {"relation": relation}
 
         if not relation.is_hive_metastore():
-            results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
-            results["information_schema.column_tags"] = adapter.execute_macro(
-                "fetch_column_tags", kwargs=kwargs
+            # To be backward compatible model_config can be None. In that case, tags should be
+            # fetched to maintain backward compatibility.
+            table_tag_config = model_config.config.get(TagsProcessor.name) if model_config else None
+            if table_tag_config is None or table_tag_config.requires_server_metadata_for_diff():
+                results["information_schema.tags"] = adapter.execute_macro(
+                    "fetch_tags", kwargs=kwargs
+                )
+            else:
+                results["information_schema.tags"] = None
+
+            # To be backward compatible model_config can be None. In that case, tags should be
+            # fetched to maintain backward compatibility.
+            column_tag_config = (
+                model_config.config.get(ColumnTagsProcessor.name) if model_config else None
             )
+            if column_tag_config is None or column_tag_config.requires_server_metadata_for_diff():
+                results["information_schema.column_tags"] = adapter.execute_macro(
+                    "fetch_column_tags", kwargs=kwargs
+                )
+            else:
+                results["information_schema.column_tags"] = None
+
             results["non_null_constraint_columns"] = adapter.execute_macro(
                 "fetch_non_null_constraint_columns", kwargs=kwargs
             )
@@ -1201,15 +1253,26 @@ def config_type(cls) -> type[ViewConfig]:
 
     @classmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         results = {}
         kwargs = {"relation": relation}
 
         results["information_schema.views"] = get_first_row(
             adapter.execute_macro("get_view_description", kwargs=kwargs)
         )
-        results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
+
+        # To be backward compatible model_config can be None. In that case, tags should be fetched
+        # to maintain backward compatibility.
+        table_tag_config = model_config.config.get(TagsProcessor.name) if model_config else None
+        if table_tag_config is None or table_tag_config.requires_server_metadata_for_diff():
+            results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
+        else:
+            results["information_schema.tags"] = None
+
         results["show_tblproperties"] = adapter.execute_macro("fetch_tbl_properties", kwargs=kwargs)
 
         kwargs = {"table_name": relation}
@@ -1228,14 +1291,23 @@ def config_type(cls) -> type[MetricViewConfig]:
 
     @classmethod
     def _describe_relation(
-        cls, adapter: DatabricksAdapter, relation: DatabricksRelation
+        cls,
+        adapter: DatabricksAdapter,
+        relation: DatabricksRelation,
+        model_config: Optional[DatabricksRelationConfigBase] = None,
     ) -> RelationResults:
         results = {}
         kwargs = {"relation": relation}
-        results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
         results["show_tblproperties"] = adapter.execute_macro("fetch_tbl_properties", kwargs=kwargs)
         kwargs = {"table_name": relation}
         results["describe_extended"] = adapter.execute_macro(
             DESCRIBE_TABLE_EXTENDED_MACRO_NAME, kwargs=kwargs
         )
+
+        table_tag_config = model_config.config.get(TagsProcessor.name) if model_config else None
+        if table_tag_config is None or table_tag_config.requires_server_metadata_for_diff():
+            results["information_schema.tags"] = adapter.execute_macro("fetch_tags", kwargs=kwargs)
+        else:
+            results["information_schema.tags"] = None
+
         return results

dbt/adapters/databricks/relation_configs/base.py

@@ -34,6 +34,12 @@ def get_diff(self, other: Self) -> Optional[Self]:
             return self
         return None
 
+    def requires_server_metadata_for_diff(self) -> bool:
+        """
+        Indicates whether server metadata is required to compute the diff for this component.
+        """
+        return True
+
 
 class DatabricksRelationChangeSet(BaseModel):
     """Class for encapsulating the changes that need to be applied to a Databricks relation."""

dbt/adapters/databricks/relation_configs/column_tags.py

@@ -41,6 +41,12 @@ def get_diff(self, other: "ColumnTagsConfig") -> Optional["ColumnTagsConfig"]:
             return ColumnTagsConfig(set_column_tags=set_column_tags)
         return None
 
+    def requires_server_metadata_for_diff(self) -> bool:
+        """
+        Indicates whether server metadata is required to compute the diff for this component.
+        """
+        return self.set_column_tags is not None and len(self.set_column_tags) > 0
+
 
 class ColumnTagsProcessor(DatabricksComponentProcessor[ColumnTagsConfig]):
     name: ClassVar[str] = "column_tags"

dbt/adapters/databricks/relation_configs/tags.py

@@ -22,6 +22,12 @@ def get_diff(self, other: "TagsConfig") -> Optional["TagsConfig"]:
             return TagsConfig(set_tags=self.set_tags)
         return None
 
+    def requires_server_metadata_for_diff(self) -> bool:
+        """
+        Indicates whether server metadata is required to compute the diff for this component.
+        """
+        return self.set_tags is not None and len(self.set_tags) > 0
+
 
 class TagsProcessor(DatabricksComponentProcessor[TagsConfig]):
     name: ClassVar[str] = "tags"

dbt/include/databricks/macros/materializations/incremental/incremental.sql

@@ -133,8 +133,8 @@
         {{ set_overwrite_mode('DYNAMIC') }}
       {%- endif -%}
       {#-- Relation must be merged --#}
-      {%- set _existing_config = adapter.get_relation_config(existing_relation) -%}
       {%- set model_config = adapter.get_config_from_model(config.model) -%}
+      {%- set _existing_config = adapter.get_relation_config(existing_relation, model_config) -%}
       {%- set _configuration_changes = model_config.get_changeset(_existing_config) -%}
       {%- call statement('create_temp_relation', language=language) -%}
         {{ create_table_as(True, temp_relation, compiled_code, language) }}
@@ -241,8 +241,8 @@
 {% macro process_config_changes(target_relation) %}
   {% set apply_config_changes = config.get('incremental_apply_config_changes', True) | as_bool %}
   {% if apply_config_changes %}
-    {%- set existing_config = adapter.get_relation_config(target_relation) -%}
     {%- set model_config = adapter.get_config_from_model(config.model) -%}
+    {%- set existing_config = adapter.get_relation_config(target_relation, model_config) -%}
     {%- set configuration_changes = model_config.get_changeset(existing_config) -%}
     {{ apply_config_changeset(target_relation, model, configuration_changes) }}
   {% endif %}

dbt/include/databricks/macros/relations/config.sql

@@ -1,6 +1,6 @@
 {%- macro get_configuration_changes(existing_relation) -%}
-    {%- set existing_config = adapter.get_relation_config(existing_relation) -%}
     {%- set model_config = adapter.get_config_from_model(config.model) -%}
+    {%- set existing_config = adapter.get_relation_config(existing_relation, model_config) -%}
     {%- set configuration_changes = model_config.get_changeset(existing_config) -%}
     {% do return(configuration_changes) %}
 {%- endmacro -%}

tests/functional/adapter/fixtures.py

@@ -1,5 +1,21 @@
 import pytest
 
+fail_if_tag_fetch_called_macros = """
+{% macro fetch_tags(relation) %}
+  {{ exceptions.raise_compiler_error("fetch_tags should not be called") }}
+{% endmacro %}
+"""
+
+fail_if_tag_and_column_tag_fetch_called_macros = """
+{% macro fetch_tags(relation) %}
+  {{ exceptions.raise_compiler_error("fetch_tags should not be called") }}
+{% endmacro %}
+
+{% macro fetch_column_tags(relation) %}
+  {{ exceptions.raise_compiler_error("fetch_column_tags should not be called") }}
+{% endmacro %}
+"""
+
 
 class MaterializationV1Mixin:
     @pytest.fixture(scope="class")

tests/functional/adapter/incremental/fixtures.py

@@ -76,6 +76,47 @@
         - name: color
 """
 
+metadata_fetch_incremental_sql = """
+{{ config(
+    materialized = 'incremental',
+    unique_key = 'id',
+) }}
+
+select cast(1 as bigint) as id
+"""
+
+metadata_fetch_no_tags_schema = """
+version: 2
+
+models:
+  - name: metadata_fetch_incremental
+    columns:
+      - name: id
+"""
+
+metadata_fetch_table_tags_schema = """
+version: 2
+
+models:
+  - name: metadata_fetch_incremental
+    config:
+      databricks_tags:
+        classification: internal
+    columns:
+      - name: id
+"""
+
+metadata_fetch_column_tags_schema = """
+version: 2
+
+models:
+  - name: metadata_fetch_incremental
+    columns:
+      - name: id
+        databricks_tags:
+          classification: internal
+"""
+
 tblproperties_a = """
 version: 2