Use LTS Databricks runtime version (#3459)

## Changes
Use LTS Databricks runtime version because the convert to external table
migration strategy fails on the latest runtime

> Note: This postpones the problem! Which is preferred for the next
release as users can experience this problem with the latest UCX
release. However, it does not solve the problem.

### Linked issues

Resolves #3453 (for now)
Reslves #3461

### Functionality

- [x] modified existing workflow: `migrate-tables`

### Tests

- [x] reused integration tests

Author: JCZuurmond

src/databricks/labs/ucx/hive_metastore/table_migrate.py

@@ -293,7 +293,12 @@ def _catalog_type(self):
     def _catalog_table(self):
         return self._spark._jvm.org.apache.spark.sql.catalyst.catalog.CatalogTable  # pylint: disable=protected-access
 
-    def _convert_hms_table_to_external(self, src_table: Table):
+    def _convert_hms_table_to_external(self, src_table: Table) -> bool:
+        """Converts a Hive metastore table to external using Spark JVM methods.
+
+        TODO:
+            This method fails for Databricks runtime 16.0, probably due to the JDK update (https://docs.databricks.com/en/release-notes/runtime/16.0.html#breaking-change-jdk-17-is-now-the-default).
+        """
         logger.info(f"Changing HMS managed table {src_table.name} to External Table type.")
         inventory_table = self._tables_crawler.full_name
         try:

src/databricks/labs/ucx/installer/policy.py

@@ -106,7 +106,7 @@ def _get_instance_pool_id(self) -> str | None:
             return None
 
     def _definition(self, conf: dict, instance_profile: str | None, instance_pool_id: str | None) -> str:
-        latest_lts_dbr = self._ws.clusters.select_spark_version(latest=True)
+        latest_lts_dbr = self._ws.clusters.select_spark_version(latest=True, long_term_support=True)
         node_type_id = self._ws.clusters.select_node_type(local_disk=True, min_memory_gb=32, min_cores=4)
         policy_definition = {
             "spark_version": self._policy_config(latest_lts_dbr),

tests/integration/hive_metastore/test_workflows.py

@@ -71,6 +71,12 @@ def test_table_migration_job_refreshes_migration_status(
 
 
 def test_table_migration_convert_manged_to_external(installation_ctx, make_table_migration_context) -> None:
+    """Convert managed tables to external before migrating.
+
+    Note:
+        This test fails from Databricks runtime 16.0 (https://docs.databricks.com/en/release-notes/runtime/16.0.html),
+        probably due to the JDK update (https://docs.databricks.com/en/release-notes/runtime/16.0.html#breaking-change-jdk-17-is-now-the-default).
+    """
     tables, dst_schema = make_table_migration_context("managed", installation_ctx)
     ctx = installation_ctx.replace(
         config_transform=lambda wc: dataclasses.replace(

tests/integration/install/test_installation.py

@@ -108,15 +108,15 @@ def test_job_failure_propagates_correct_error_message_and_logs(ws, sql_backend,
 
 
 @retried(on=[NotFound, InvalidParameterValue], timeout=timedelta(minutes=3))
-def test_job_cluster_policy(ws, installation_ctx):
+def test_job_cluster_policy(ws, installation_ctx) -> None:
     installation_ctx.workspace_installation.run()
     user_name = ws.current_user.me().user_name
     cluster_policy = ws.cluster_policies.get(policy_id=installation_ctx.config.policy_id)
     policy_definition = json.loads(cluster_policy.definition)
 
     assert cluster_policy.name == f"Unity Catalog Migration ({installation_ctx.inventory_database}) ({user_name})"
 
-    spark_version = ws.clusters.select_spark_version(latest=True)
+    spark_version = ws.clusters.select_spark_version(latest=True, long_term_support=True)
     assert policy_definition["spark_version"]["value"] == spark_version
     assert policy_definition["node_type_id"]["value"] == ws.clusters.select_node_type(local_disk=True, min_memory_gb=32)
     if ws.config.is_azure: