Crawlers: Use TRUNCATE TABLE instead of DELETE FROM when resettin…

…g crawler tables. (#2392)

## Changes

This PR updates the `.reset()` implementation for crawlers to use
`TRUNCATE TABLE` instead of `DELETE FROM`. This is slightly more
efficient and idiomatic (but not an issue of correctness).

### Functionality

- [X] modified existing workflow: `migrate-data-reconciliation`

### Tests

- [X] updated unit tests
- [X] covered by integration tests

src/databricks/labs/ucx/framework/crawlers.py

@@ -5,6 +5,8 @@
 from databricks.labs.lsql.backends import SqlBackend
 from databricks.sdk.errors import NotFound
 
+from databricks.labs.ucx.framework.utils import escape_sql_identifier
+
 logger = logging.getLogger(__name__)
 
 
@@ -52,7 +54,7 @@ def reset(self):
         Delete the content of the inventory table.
         The next call to `snapshot` will re-populate the table.
         """
-        self._exec(f"DELETE FROM {self.full_name}")
+        self._exec(f"TRUNCATE TABLE {escape_sql_identifier(self.full_name)}")
 
     @staticmethod
     def _valid(name: str) -> str:

tests/unit/hive_metastore/test_table_migrate.py

@@ -921,7 +921,7 @@ def test_table_status_reset():
     table_status_crawler = MigrationStatusRefresher(client, backend, "ucx", table_crawler)
     table_status_crawler.reset()
     assert backend.queries == [
-        "DELETE FROM hive_metastore.ucx.migration_status",
+        "TRUNCATE TABLE hive_metastore.ucx.migration_status",
     ]
     table_crawler.snapshot.assert_not_called()
     client.catalogs.list.assert_not_called()

tests/unit/hive_metastore/test_workflows.py

@@ -67,6 +67,6 @@ def test_migrate_ctas_views(run_workflow):
 def test_refresh_migration_status_is_refreshed(run_workflow, workflow):
     """Migration status is refreshed by deleting and showing new tables"""
     ctx = run_workflow(getattr(workflow, "refresh_migration_status"))
-    assert "DELETE FROM hive_metastore.ucx.migration_status" in ctx.sql_backend.queries
+    assert "TRUNCATE TABLE hive_metastore.ucx.migration_status" in ctx.sql_backend.queries
     assert "SHOW DATABASES" in ctx.sql_backend.queries
     # No "SHOW TABLE FROM" query as table are not mocked