xpk 0.13.0__py3-none-any.whl → 0.14.1__py3-none-any.whl

xpk/core/network.py

@@ -17,7 +17,7 @@ limitations under the License.
 from ..utils.console import xpk_exit, xpk_print
 from ..utils.file import write_tmp_file
 from .commands import run_command_for_value, run_command_with_updates
-from .gcloud_context import zone_to_region
+from .gcloud_context import zone_to_region, get_cluster_location
 
 # cluster_network_yaml: the config when creating the network for a3 cluster
 CLUSTER_NETWORK_YAML = """
@@ -126,7 +126,7 @@ def create_cluster_network(args, index) -> int:
         ' --subnet-mode=custom --mtu=8244'
     )
     return_code = run_command_with_updates(
-        command, 'Create Cluster Network', args, verbose=False
+        command, 'Create Cluster Network', verbose=False
     )
 
     if return_code != 0:
@@ -152,7 +152,9 @@ def create_cluster_subnet(args, index) -> int:
   if return_code > 0:
     xpk_print('Listing all subnets failed!')
     return return_code
-  subnet_name = f'{args.cluster}-{zone_to_region(args.zone)}-sub-{index}'
+  subnet_name = (
+      f'{args.cluster}-{get_cluster_location(args.project, args.cluster, args.zone)}-sub-{index}'
+  )
   if subnet_name not in existing_subnet_names:
     command = (
         f'gcloud compute --project={args.project}'
@@ -161,7 +163,7 @@ def create_cluster_subnet(args, index) -> int:
         f' --region={zone_to_region(args.zone)} --range=192.168.{index}.0/24'
     )
     return_code = run_command_with_updates(
-        command, 'Create Cluster Subnet', args, verbose=False
+        command, 'Create Cluster Subnet', verbose=False
     )
 
     if return_code != 0:
@@ -197,7 +199,7 @@ def create_cluster_firewall_rule(args, index) -> int:
         ' --rules=tcp:0-65535,udp:0-65535,icmp --source-ranges=192.168.0.0/16'
     )
     return_code = run_command_with_updates(
-        command, 'Create Cluster Firewall Rule', args, verbose=False
+        command, 'Create Cluster Firewall Rule', verbose=False
     )
 
     if return_code != 0:
@@ -224,7 +226,7 @@ def create_cluster_network_config(args) -> int:
   command = f'kubectl apply -f {str(tmp)}'
 
   return_code = run_command_with_updates(
-      command, 'GKE Cluster Create Network Config', args
+      command, 'GKE Cluster Create Network Config'
   )
   if return_code != 0:
     xpk_print(
@@ -235,19 +237,14 @@ def create_cluster_network_config(args) -> int:
   return 0
 
 
-def get_cluster_subnetworks(args) -> list[str]:
+def get_cluster_subnetworks() -> list[str]:
   """Gets the list of cluster networks.
 
-  Args:
-    args: user provided arguments for running the command.
-
   Returns:
     list[str]: list of cluster networks
   """
   command = 'kubectl get GKENetworkParamSet'
-  return_code, stdout = run_command_for_value(
-      command, 'Get Cluster Networks', args
-  )
+  return_code, stdout = run_command_for_value(command, 'Get Cluster Networks')
   if return_code != 0:
     xpk_print('GKE Cluster Get NetworkParamSet failed')
     xpk_exit(return_code)
@@ -302,7 +299,7 @@ def delete_cluster_subnets(args) -> int:
     )
 
     return_code = run_command_with_updates(
-        command, 'Delete Cluster Subnet', args, verbose=False
+        command, 'Delete Cluster Subnet', verbose=False
     )
 
     if return_code != 0:
@@ -328,7 +325,7 @@ def get_all_networks_programmatic(args) -> tuple[list[str], int]:
       f' --project={args.project}'
   )
   return_code, raw_network_output = run_command_for_value(
-      command, 'Get All Networks', args
+      command, 'Get All Networks'
   )
   if return_code != 0:
     xpk_print(f'Get All Networks returned ERROR {return_code}')
@@ -346,14 +343,16 @@ def get_all_subnets_programmatic(args) -> tuple[list[str], int]:
   Returns:
     List of subnets and 0 if successful and 1 otherwise.
   """
-  subnet_name_filter = f'{args.cluster}-{zone_to_region(args.zone)}-sub-*'
+  subnet_name_filter = (
+      f'{args.cluster}-{get_cluster_location(args.project, args.cluster, args.zone)}-sub-*'
+  )
 
   command = (
       'gcloud compute networks subnets list'
       f' --filter=name~"{subnet_name_filter}" --project={args.project}'
   )
   return_code, raw_subnets_output = run_command_for_value(
-      command, 'Get All Subnets', args
+      command, 'Get All Subnets'
   )
   if return_code != 0:
     xpk_print(f'Get All Subnets returned ERROR {return_code}')
@@ -380,7 +379,7 @@ def get_all_firewall_rules_programmatic(args) -> tuple[list[str], int]:
       f' --project={args.project}'
   )
   return_code, raw_subnets_output = run_command_for_value(
-      command, 'Get All Firewall Rules', args
+      command, 'Get All Firewall Rules'
   )
   if return_code != 0:
     xpk_print(f'Get All Firewall Rules returned ERROR {return_code}')

xpk/core/nodepool.py

@@ -16,6 +16,7 @@ limitations under the License.
 
 from typing import List
 from ..utils.console import get_user_input, xpk_print
+from ..utils.topology import get_topology_product, is_topology_valid
 from .capacity import (
     AUTOPROVISIONING_CONFIG_VALUE,
     H100_MEGA_DEVICE_TYPE,
@@ -25,7 +26,7 @@ from .capacity import (
     print_reservations,
 )
 from .commands import run_command_for_value, run_commands
-from .gcloud_context import GkeServerConfig, zone_to_region
+from .gcloud_context import GkeServerConfig, get_cluster_location, zone_to_region
 from .resources import (
     CLUSTER_CONFIGMAP_YAML,
     CLUSTER_RESOURCES_CONFIGMAP,
@@ -33,8 +34,7 @@ from .resources import (
     create_or_update_cluster_configmap,
 )
 from .system_characteristics import AcceleratorType
-from functools import reduce
-from operator import mul
+
 
 CLOUD_PLATFORM_AUTH_SCOPE_URL = (
     '"https://www.googleapis.com/auth/cloud-platform"'
@@ -147,7 +147,7 @@ def run_gke_node_pool_create_command(
         command = (
             'gcloud beta container node-pools delete'
             f' {node_pool_name} --cluster={args.cluster}'
-            f' --zone={zone_to_region(args.zone)}'
+            f' --zone={get_cluster_location(args.project, args.cluster, args.zone)}'
             f' --project={args.project} --quiet'
         )
         task = f'NodepoolDelete-{node_pool_name}'
@@ -173,9 +173,7 @@ def run_gke_node_pool_create_command(
           ):
             command = (
                 'gcloud container node-pools update'
-                f' {node_pool_name} --cluster={args.cluster}'
-                f' --zone={zone_to_region(args.zone)}'
-                f' --project={args.project} --quiet'
+                f' {node_pool_name} --cluster={args.cluster} --location={get_cluster_location(args.project, args.cluster, args.zone)} --project={args.project} --quiet'
                 ' --workload-metadata=GKE_METADATA'
             )
             task = (
@@ -212,7 +210,6 @@ def run_gke_node_pool_create_command(
         delete_commands,
         'Delete Nodepools',
         delete_task_names,
-        dry_run=args.dry_run,
     )
     if max_return_code != 0:
       xpk_print(f'Delete Nodepools returned ERROR {max_return_code}')
@@ -240,7 +237,6 @@ def run_gke_node_pool_create_command(
           update_WI_commands,
           'Enable Workload Identity on existing Nodepools',
           update_WI_task_names,
-          dry_run=args.dry_run,
       )
       if max_return_code != 0:
         xpk_print(
@@ -265,12 +261,16 @@ def run_gke_node_pool_create_command(
       )
       configmap_yml = {}
       configmap_yml[resources_configmap_name] = resources_yml
-      return_code = create_or_update_cluster_configmap(
-          configmap_yml, args.dry_run
-      )
+      return_code = create_or_update_cluster_configmap(configmap_yml)
       if return_code != 0:
         return 1
 
+  placement_args = ''
+  if system.requires_workload_policy and is_topology_valid(system.topology):
+    placement_policy = f'{args.cluster}-placement-policy'
+    ensure_resource_policy_exists(placement_policy, args, system.topology)
+    placement_args = f' --placement-policy={placement_policy}'
+
   create_commands = []
   create_task_names = []
   for node_pool_name in desired_node_pool_names:
@@ -279,19 +279,18 @@ def run_gke_node_pool_create_command(
     command = (
         'gcloud beta container node-pools create'
         f' {node_pool_name}'
-        f' --region={zone_to_region(args.zone)}'
+        f' --location={get_cluster_location(args.project, args.cluster, args.zone)}'
         f' --cluster={args.cluster}'
         f' --project={args.project} --node-locations={args.zone}'
         f' --machine-type={system.gce_machine_type}'
         f' --host-maintenance-interval={args.host_maintenance_interval}'
         f' {capacity_args}'
+        f'{placement_args}'
         ' --enable-gvnic'
     )
     if system.accelerator_type == AcceleratorType['TPU']:
       command += f' --node-version={gke_node_pool_version}'
-      topology_product = reduce(
-          mul, (int(x) for x in system.topology.split('x')), 1
-      )
+      topology_product = get_topology_product(system.topology)
       if capacity_type == CapacityType.FLEX_START:
         command += ' --num-nodes=0'
       elif topology_product > 1:
@@ -301,11 +300,18 @@ def run_gke_node_pool_create_command(
       )
 
       if topology_product > 1:
-        command += ' --placement-type=COMPACT  --max-pods-per-node 15'
-        command += f' --tpu-topology={system.topology}'
+        # --placement-type=COMPACT enables group placement policy which
+        # is mutually exclusive with workload policy, --tpu-topology should
+        # also not be passed when workload policy is used
+        if not system.requires_workload_policy:
+          command += ' --placement-type=COMPACT'
+          command += f' --tpu-topology={system.topology}'
+        command += ' --max-pods-per-node 15'
         command += f' {args.custom_tpu_nodepool_arguments}'
     elif system.accelerator_type == AcceleratorType['GPU']:
-      subnet_prefix = f'{args.cluster}-{zone_to_region(args.zone)}'
+      subnet_prefix = (
+          f'{args.cluster}-{get_cluster_location(args.project, args.cluster, args.zone)}'
+      )
       if capacity_type == CapacityType.FLEX_START:
         command += ' --num-nodes=0'
       else:
@@ -348,7 +354,7 @@ def run_gke_node_pool_create_command(
         continue
       command = (
           'gcloud beta container node-pools create'
-          f' {node_pool_name} --node-version={gke_node_pool_version} --cluster={args.cluster} --project={args.project} --node-locations={args.zone} --region={zone_to_region(args.zone)} --num-nodes=1'
+          f' {node_pool_name} --node-version={gke_node_pool_version} --cluster={args.cluster} --project={args.project} --node-locations={args.zone} --location={get_cluster_location(args.project, args.cluster, args.zone)} --num-nodes=1'
           f' --machine-type={args.pathways_gce_machine_type} --scopes=storage-full,gke-default,{CLOUD_PLATFORM_AUTH_SCOPE_URL} --enable-autoscaling'
           ' --min-nodes=1 --max-nodes=20'
       )
@@ -362,7 +368,6 @@ def run_gke_node_pool_create_command(
       create_commands,
       'Create Nodepools',
       create_task_names,
-      dry_run=args.dry_run,
   )
   if max_return_code != 0:
     xpk_print(f'Create Nodepools returned ERROR {max_return_code}')
@@ -432,11 +437,12 @@ def get_all_nodepools_programmatic(args) -> tuple[list[str], int]:
   command = (
       'gcloud beta container node-pools list'
       ' --cluster'
-      f' {args.cluster} --project={args.project} --region={zone_to_region(args.zone)}'
+      f' {args.cluster} --project={args.project} '
+      f'--location={get_cluster_location(args.project, args.cluster, args.zone)}'
       ' --format="csv[no-heading](name)"'
   )
   return_code, raw_nodepool_output = run_command_for_value(
-      command, 'Get All Node Pools', args
+      command, 'Get All Node Pools'
   )
   if return_code != 0:
     xpk_print(f'Get All Node Pools returned ERROR {return_code}')
@@ -460,10 +466,10 @@ def get_nodepool_zone(args, nodepool_name) -> tuple[int, str | None]:
   command = (
       f'gcloud beta container node-pools describe {nodepool_name}'
       f' --cluster {args.cluster} --project={args.project}'
-      f' --region={zone_to_region(args.zone)} --format="value(locations)"'
+      f' --location={get_cluster_location(args.project, args.cluster, args.zone)} --format="value(locations)"'
   )
   return_code, nodepool_zone = run_command_for_value(
-      command, 'Get Node Pool Zone', args, dry_run_return_val=args.zone
+      command, 'Get Node Pool Zone', dry_run_return_val=args.zone
   )
   if return_code != 0:
     xpk_print(f'Get Node Pool Zone returned ERROR {return_code}')
@@ -490,13 +496,13 @@ def get_gke_node_pool_version(
   # By default use the current gke master version for creating node pools.
   command_description = 'Determine current gke master version'
   command = (
-      f'gcloud beta container clusters describe {args.cluster}'
-      f' --region {zone_to_region(args.zone)} --project {args.project}'
-      ' --format="value(currentMasterVersion)"'
+      f'gcloud beta container clusters describe {args.cluster} --location'
+      f' {get_cluster_location(args.project, args.cluster, args.zone)} --project'
+      f' {args.project} --format="value(currentMasterVersion)"'
   )
 
   return_code, current_gke_master_version = run_command_for_value(
-      command, command_description, args
+      command, command_description
   )
   if return_code != 0:
     xpk_print(
@@ -540,52 +546,6 @@ def get_gke_node_pool_version(
   return 0, node_pool_gke_version
 
 
-def upgrade_gke_nodepools_version(args, default_rapid_gke_version) -> int:
-  """Upgrade nodepools in the cluster to default rapid gke version. Recreates the nodes.
-
-  Args:
-    args: user provided arguments for running the command.
-    default_rapid_gke_version: Rapid default version for the upgrade.
-
-  Returns:
-    0 if successful and 1 otherwise.
-  """
-  existing_node_pool_names, return_code = get_all_nodepools_programmatic(args)
-  if return_code != 0:
-    xpk_print('Listing all node pools failed!')
-    return return_code
-
-  # Batch execution to upgrade node pools simultaneously
-  commands = []
-  task_names = []
-  for node_pool_name in existing_node_pool_names:
-    commands.append(
-        'gcloud container clusters upgrade'
-        f' {args.cluster} --project={args.project}'
-        f' --region={zone_to_region(args.zone)}'
-        f' --cluster-version={default_rapid_gke_version}'
-        f' --node-pool={node_pool_name}'
-        ' --quiet'
-    )
-    task_names.append(f'Upgrading node pool {node_pool_name}.')
-
-  for i, command in enumerate(commands):
-    xpk_print(f'To complete {task_names[i]} we are executing {command}')
-  max_return_code = run_commands(
-      commands,
-      'Update GKE node pools to default RAPID GKE version',
-      task_names,
-      dry_run=args.dry_run,
-  )
-  if max_return_code != 0:
-    xpk_print(
-        'GKE node pools update to default RAPID GKE version returned ERROR:'
-        f' {max_return_code}'
-    )
-    return int(max_return_code)
-  return 0
-
-
 def get_nodepool_workload_metadata_mode(
     args, nodepool_name
 ) -> tuple[int, str | None]:
@@ -601,10 +561,10 @@ def get_nodepool_workload_metadata_mode(
   command = (
       f'gcloud beta container node-pools describe {nodepool_name}'
       f' --cluster {args.cluster} --project={args.project}'
-      f' --region={zone_to_region(args.zone)} --format="value(config.workloadMetadataConfig.mode)"'
+      f' --location={get_cluster_location(args.project, args.cluster, args.zone)} --format="value(config.workloadMetadataConfig.mode)"'
   )
   return_code, nodepool_WI_mode = run_command_for_value(
-      command, 'Get Node Pool Workload Identity Metadata Mode', args
+      command, 'Get Node Pool Workload Identity Metadata Mode'
   )
   if return_code != 0:
     xpk_print(
@@ -632,3 +592,32 @@ def get_desired_node_pool_names(
     result.add(f'{cluster_name}-np-{i}')
     i += 1
   return list(result)
+
+
+def ensure_resource_policy_exists(
+    resource_policy_name: str, args, topology: str
+) -> None:
+  return_code, _ = run_command_for_value(
+      (
+          'gcloud compute resource-policies describe'
+          f' {resource_policy_name} '
+          f'--project={args.project} '
+          f'--region={zone_to_region(args.zone)}'
+      ),
+      'Retrieve resource policy',
+  )
+
+  if return_code == 0:
+    return
+
+  return_code, _ = run_command_for_value(
+      (
+          'gcloud compute resource-policies create workload-policy'
+          f' {resource_policy_name} --project={args.project} --region={zone_to_region(args.zone)} --type=HIGH_THROUGHPUT'
+          f' --accelerator-topology={topology}'
+      ),
+      'Create resource policy',
+  )
+
+  if return_code != 0:
+    raise RuntimeError('Unable to create resource policy')

xpk/core/nodepool_test.py

@@ -14,7 +14,13 @@ See the License for the specific language governing permissions and
 limitations under the License.
 """
 
-from xpk.core.nodepool import get_desired_node_pool_names
+import pytest
+from xpk.core.nodepool import (
+    ensure_resource_policy_exists,
+    get_desired_node_pool_names,
+    run_gke_node_pool_create_command,
+)
+from xpk.core.system_characteristics import AcceleratorType, SystemCharacteristics
 
 CLUSTER_NAME = "running-cucumber"
 
@@ -80,3 +86,194 @@ def test_compute_desired_node_pool_names_with_unknown_node_pools():
 
   expected_result = [node_pool_name(0), node_pool_name(3)]
   assert set(result) == set(expected_result)
+
+
+def test_ensure_resource_policy_exists_with_existing_policy_retrieves_existing_policy(
+    mocker,
+):
+  args = mocker.Mock(project="test-project", zone="us-central1-a")
+  mocker.patch("xpk.core.nodepool.get_cluster_location", return_value=args.zone)
+  mock = mocker.patch(
+      "xpk.core.nodepool.run_command_for_value", return_value=(0, "")
+  )
+  ensure_resource_policy_exists("resource-policy", args, "2x2x1")
+  mock.assert_called_once()
+
+
+def test_ensure_resource_policy_exists_without_existing_policy_creates_policy(
+    mocker,
+):
+  args = mocker.Mock(project="test-project", zone="us-central1-a")
+  mocker.patch("xpk.core.nodepool.get_cluster_location", return_value=args.zone)
+  mock = mocker.patch(
+      "xpk.core.nodepool.run_command_for_value", side_effect=[(1, ""), (0, "")]
+  )
+  ensure_resource_policy_exists("resource-policy", args, "2x2x1")
+  assert mock.call_count == 2
+  assert mock.call_args_list[0].args[1] == "Retrieve resource policy"
+
+
+def test_ensure_resource_policy_exits_without_existing_policy_throws_when_creation_fails(
+    mocker,
+):
+  with pytest.raises(RuntimeError):
+    args = mocker.Mock(project="test-project", zone="us-central1-a")
+    mocker.patch(
+        "xpk.core.nodepool.get_cluster_location", return_value=args.zone
+    )
+    mocker.patch(
+        "xpk.core.nodepool.run_command_for_value",
+        side_effect=[(1, ""), (1, "")],
+    )
+    ensure_resource_policy_exists("resource-policy", args, "2x2x1")
+
+
+@pytest.fixture
+def mock_nodepool_dependencies(mocker):
+  """Mocks dependencies for run_gke_node_pool_create_command."""
+  mocker.patch(
+      "xpk.core.nodepool.get_all_nodepools_programmatic", return_value=([], 0)
+  )
+  mocker.patch(
+      "xpk.core.nodepool.get_capacity_type", return_value=("on-demand", 0)
+  )
+  mocker.patch(
+      "xpk.core.nodepool.get_capacity_arguments_from_capacity_type",
+      return_value=("--on-demand", 0),
+  )
+  mocker.patch(
+      "xpk.core.nodepool.get_cluster_location", return_value="us-central1"
+  )
+  mocker.patch("xpk.core.nodepool.run_commands", return_value=0)
+  mocker.patch("xpk.core.nodepool.get_user_input", return_value=True)
+  mock_is_topology_valid = mocker.patch("xpk.core.nodepool.is_topology_valid")
+  mock_ensure_resource_policy = mocker.patch(
+      "xpk.core.nodepool.ensure_resource_policy_exists"
+  )
+  return mock_is_topology_valid, mock_ensure_resource_policy
+
+
+def test_placement_policy_created_for_gpu_with_valid_topology(
+    mocker, mock_nodepool_dependencies
+):
+  """Tests that placement policy is created for GPUs with a valid topology."""
+  mock_is_topology_valid, mock_ensure_resource_policy = (
+      mock_nodepool_dependencies
+  )
+  mock_is_topology_valid.return_value = True
+  args = mocker.Mock(
+      tpu_type=None,
+      device_type="h100-80gb-8",
+      cluster="test-cluster",
+      project="test-project",
+      zone="us-central1-a",
+  )
+  system = SystemCharacteristics(
+      topology="N/A",
+      vms_per_slice=1,
+      gke_accelerator="nvidia-h100-80gb",
+      gce_machine_type="a3-highgpu-8g",
+      chips_per_vm=8,
+      accelerator_type=AcceleratorType["GPU"],
+      device_type="h100-80gb-8",
+      supports_sub_slicing=False,
+  )
+
+  run_gke_node_pool_create_command(args, system, "1.2.3")
+
+  mock_ensure_resource_policy.assert_called_once()
+
+
+def test_placement_policy_not_created_for_gpu_with_invalid_topology(
+    mocker, mock_nodepool_dependencies
+):
+  """Tests that placement policy is not created for GPUs with an invalid topology."""
+  mock_is_topology_valid, mock_ensure_resource_policy = (
+      mock_nodepool_dependencies
+  )
+  mock_is_topology_valid.return_value = False
+  args = mocker.Mock(
+      tpu_type=None,
+      device_type="h100-80gb-8",
+      cluster="test-cluster",
+      zone="us-central1-a",
+  )
+  system = SystemCharacteristics(
+      topology="N/A",
+      vms_per_slice=1,
+      gke_accelerator="nvidia-h100-80gb",
+      gce_machine_type="a3-highgpu-8g",
+      chips_per_vm=8,
+      accelerator_type=AcceleratorType["GPU"],
+      device_type="h100-80gb-8",
+      supports_sub_slicing=False,
+  )
+
+  run_gke_node_pool_create_command(args, system, "1.2.3")
+
+  mock_ensure_resource_policy.assert_not_called()
+
+
+def test_placement_policy_created_for_tpu7x_with_valid_topology(
+    mocker, mock_nodepool_dependencies
+):
+  """Tests that placement policy is created for tpu7x with a valid topology."""
+  mock_is_topology_valid, mock_ensure_resource_policy = (
+      mock_nodepool_dependencies
+  )
+  mock_is_topology_valid.return_value = True
+  args = mocker.Mock(
+      tpu_type="tpu7x-8",
+      device_type=None,
+      num_slices=1,
+      cluster="test-cluster",
+      project="test-project",
+      zone="us-central1-a",
+  )
+  system = SystemCharacteristics(
+      topology="2x2x1",
+      vms_per_slice=1,
+      gke_accelerator="tpu7x",
+      gce_machine_type="tpu7x-standard-4t",
+      chips_per_vm=4,
+      accelerator_type=AcceleratorType["TPU"],
+      device_type="tpu7x-8",
+      requires_workload_policy=True,
+      supports_sub_slicing=False,
+  )
+
+  run_gke_node_pool_create_command(args, system, "1.2.3")
+
+  mock_ensure_resource_policy.assert_called_once()
+
+
+def test_placement_policy_not_created_for_non7x_tpu(
+    mocker, mock_nodepool_dependencies
+):
+  """Tests that placement policy is not created for non-tpu7x TPUs."""
+  mock_is_topology_valid, mock_ensure_resource_policy = (
+      mock_nodepool_dependencies
+  )
+  mock_is_topology_valid.return_value = True
+  args = mocker.Mock(
+      tpu_type="v6e",
+      device_type=None,
+      num_slices=1,
+      cluster="test-cluster",
+      project="test-project",
+      zone="us-central1-a",
+  )
+  system = SystemCharacteristics(
+      topology="2x2",
+      vms_per_slice=1,
+      gke_accelerator="v6e",
+      gce_machine_type="tpu-v6e-slice",
+      chips_per_vm=4,
+      accelerator_type=AcceleratorType["TPU"],
+      device_type="v6e-4",
+      supports_sub_slicing=True,
+  )
+
+  run_gke_node_pool_create_command(args, system, "1.2.3")
+
+  mock_ensure_resource_policy.assert_not_called()

xpk/core/pathways.py

@@ -16,7 +16,7 @@ limitations under the License.
 
 from ..core.commands import run_command_for_value, run_command_with_updates, run_commands
 from ..core.docker_container import get_user_workload_container
-from ..core.gcloud_context import zone_to_region
+from ..core.gcloud_context import get_cluster_location
 from ..core.nodepool import get_all_nodepools_programmatic
 from ..utils.console import xpk_exit, xpk_print
 from ..utils.execution_context import is_dry_run
@@ -116,7 +116,7 @@ def check_if_pathways_job_is_installed(args) -> bool:
       ' custom-columns=NAME:.metadata.name'
   )
   task = f'Check if PathwaysJob is installed on {args.cluster}'
-  return_code, return_msg = run_command_for_value(command, task, args)
+  return_code, return_msg = run_command_for_value(command, task)
   # return_msg contains the name of the controller pod, if found.
   xpk_print('check_if_pathways_job_is_installed', return_code, return_msg)
 
@@ -138,7 +138,7 @@ def get_pathways_unified_query_link(args) -> str:
   query_params = (
       'resource.type%3D"k8s_container"%0A'
       f'resource.labels.project_id%3D"{args.project}"%0A'
-      f'resource.labels.location%3D"{zone_to_region(args.zone)}"%0A'
+      f'resource.labels.location%3D"{get_cluster_location(args.project, args.cluster, args.zone)}"%0A'
       f'resource.labels.cluster_name%3D"{args.cluster}"%0A'
       f'resource.labels.pod_name:"{args.workload}-"%0A'
       'severity>%3DDEFAULT'
@@ -323,10 +323,10 @@ def try_to_delete_pathwaysjob_first(args, workloads) -> bool:
 
   # Not batching deletion for single workload
   if len(workloads) == 1:
-    return_code = run_command_with_updates(commands[0], 'Delete Workload', args)
+    return_code = run_command_with_updates(commands[0], 'Delete Workload')
   else:
     return_code = run_commands(
-        commands, 'Delete Workload', task_names, batch=100, dry_run=args.dry_run
+        commands, 'Delete Workload', task_names, batch=100
     )
 
   if return_code != 0: