fix: stop running OPTIMIZE for auto liquid clustering

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ### Fixes
 
+- Let Predictive Optimization manage models with `auto_liquid_cluster` instead of running synchronous `OPTIMIZE` after materialization ([#1658](https://github.com/databricks/dbt-databricks/pull/1658) resolves [#1655](https://github.com/databricks/dbt-databricks/issues/1655))
 - Skip unnecessary Unity Catalog constraint metadata queries for incremental models without enforced contracts ([#1643](https://github.com/databricks/dbt-databricks/pull/1643) resolves [#1641](https://github.com/databricks/dbt-databricks/issues/1641))
 - Recreate materialized views when query schema drifts, honoring `on_configuration_change` ([#1621](https://github.com/databricks/dbt-databricks/pull/1621) resolves [#1359](https://github.com/databricks/dbt-databricks/issues/1359))
 - Warn when documented columns are missing from V1 and V2 table and incremental models, and honor `persist_docs.columns` for V2 column comments ([#1563](https://github.com/databricks/dbt-databricks/pull/1563) ports [dbt-adapters#1684](https://github.com/dbt-labs/dbt-adapters/pull/1684), resolving [dbt-adapters#1690](https://github.com/dbt-labs/dbt-adapters/issues/1690)).

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

@@ -7,7 +7,7 @@
   {%- elif var('DATABRICKS_SKIP_OPTIMIZE', 'false')|lower != 'true' and
         var('databricks_skip_optimize', 'false')|lower != 'true' and
         adapter.resolve_file_format(config) == 'delta' -%}
-    {%- if (config.get('zorder', False) or config.get('liquid_clustered_by', False)) or config.get('auto_liquid_cluster', False) -%}
+    {%- if config.get('zorder', False) or config.get('liquid_clustered_by', False) -%}
       {%- call statement('run_optimize_stmt') -%}
         {{ get_optimize_sql(relation) }}
       {%- endcall -%}

docs/flow/incremental_flow.md

@@ -1,6 +1,6 @@
 # Incremental Flow
 
-_Last updated: 2026-08-20_
+_Last updated: 2026-08-31_
 
 > Two diagrams follow: **Existing** is the default path, **New** is used when the
 > `use_materialization_v2` behavior flag is enabled. See [flow/README.md](README.md) for what the
@@ -40,7 +40,7 @@ flowchart LR
     NEWCFG --> GRANTS[Apply grants]
     REPLACEDOCS --> GRANTS
     DOCS --> GRANTS
-    GRANTS --> OPT[Run optimize]
+    GRANTS --> OPT[Run optimize for zorder / liquid_clustered_by]
     OPT --> POST[Run post-hooks]
     POST --> STATIC[Restore static overwrite mode for non-full-refresh insert_overwrite]
 ```
@@ -78,7 +78,7 @@ flowchart LR
     CREATE --> GRANTS[Apply grants]
     SAFE --> GRANTS
     MERGE --> GRANTS
-    GRANTS --> OPT[Run optimize]
+    GRANTS --> OPT[Run optimize for zorder / liquid_clustered_by]
     OPT --> PYCLEAN{Python model?}
     PYCLEAN -- yes --> CLEAN[Drop intermediate relation]
     PYCLEAN -- no --> POST[Run post-hooks]

docs/flow/snapshot_flow.md

@@ -1,6 +1,6 @@
 # Snapshot Flow
 
-_Last updated: 2026-08-10_
+_Last updated: 2026-08-31_
 
 > Snapshots do **not** use the `use_materialization_v2` flag — there is a single path. Source:
 > `dbt/include/databricks/macros/materializations/snapshot.sql`. Strategy dispatch and the
@@ -46,7 +46,7 @@ flowchart TD
     CLEAN -- yes --> POSTSNAP[post_snapshot cleanup]
     CLEAN -- no --> CONST
     POSTSNAP --> CONST[Persist constraints]
-    CONST --> OPT[Run optimize]
+    CONST --> OPT[Run optimize for zorder / liquid_clustered_by]
     OPT --> POSTOUT["Run post-hooks (outside transaction)"]
 ```
 

docs/flow/table_flow.md

@@ -1,6 +1,6 @@
 # Table Flow
 
-_Last updated: 2026-08-10_
+_Last updated: 2026-08-31_
 
 > Two diagrams follow: **V1** is the default path, **V2** is used when the `use_materialization_v2`
 > behavior flag is enabled. See [flow/README.md](README.md) for what the flag is and how the
@@ -27,7 +27,7 @@ flowchart LR
     TAGS --> COLTAGS[Apply column tags]
     COLTAGS --> DOCS[Persist docs]
     DOCS --> CONSTRAINTS[Persist constraints]
-    CONSTRAINTS --> OPT[Run optimize]
+    CONSTRAINTS --> OPT[Run optimize for zorder / liquid_clustered_by]
     OPT --> POST[Run post-hooks]
 ```
 
@@ -53,7 +53,7 @@ flowchart LR
 
     CREATE --> GRANTS[Apply grants]
     SAFE --> GRANTS
-    GRANTS --> OPT[Run optimize]
+    GRANTS --> OPT[Run optimize for zorder / liquid_clustered_by]
     OPT --> PYCLEAN{Python model?}
     PYCLEAN -- yes --> CLEAN[Drop intermediate relation]
     PYCLEAN -- no --> POST[Run post-hooks]

tests/unit/macros/relations/test_optimize_macros.py

@@ -35,8 +35,16 @@ def test_macros_optimize_with_extraneous_info(self, config, var, template_bundle
 
         assert result == "run_optimize_stmt"
 
+    def test_macros_optimize_skips_auto_liquid_clustering(self, config, template_bundle):
+        config["auto_liquid_cluster"] = True
+
+        result = self.render_bundle(template_bundle, "optimize")
+
+        assert result == ""
+
     @pytest.mark.parametrize("key_val", ["DATABRICKS_SKIP_OPTIMIZE", "databricks_skip_optimize"])
-    def test_macros_optimize_with_skip(self, key_val, var, template_bundle):
+    def test_macros_optimize_with_skip(self, key_val, config, var, template_bundle):
+        config["liquid_clustered_by"] = ["foo"]
         var[key_val] = True
         r = self.render_bundle(template_bundle, "optimize")