[AKS] az aks nodepool add/update/upgrade: Add new parameter --undrainable-node-behavior to specify whether nodes can be cordoned during upgrade

src/azure-cli/azure/cli/command_modules/acs/_help.py

@@ -1635,6 +1635,9 @@
   - name: --node-soak-duration
     type: int
     short-summary: The amount of time (in minutes) to wait after draining a node and before reimaging it and moving on to next node.
+  - name: --undrainable-node-behavior
+    type: string
+    short-summary: Define the behavior for undrainable nodes during upgrade. The value should be "Cordon" or "Schedule". The default value is "Schedule".
   - name: --enable-encryption-at-host
     type: bool
     short-summary: Enable EncryptionAtHost, default value is false.
@@ -1793,6 +1796,9 @@
   - name: --node-soak-duration
     type: int
     short-summary: The amount of time (in minutes) to wait after draining a node and before reimaging it and moving on to next node.
+  - name: --undrainable-node-behavior
+    type: string
+    short-summary: Define the behavior for undrainable nodes during upgrade. The value should be "Cordon" or "Schedule". The default value is "Schedule".
   - name: --node-taints
     type: string
     short-summary: The node taints for the node pool. You can update the existing node taint of a nodepool or create a new node taint for a nodepool. Pass the empty string `""` to remove all taints.
@@ -1865,6 +1871,9 @@
   - name: --node-soak-duration
     type: int
     short-summary: The amount of time (in minutes) to wait after draining a node and before reimaging it and moving on to next node.
+  - name: --undrainable-node-behavior
+    type: string
+    short-summary: Define the behavior for undrainable nodes during upgrade. The value should be "Cordon" or "Schedule". The default value is "Schedule".
   - name: --snapshot-id
     type: string
     short-summary: The source snapshot id used to upgrade this nodepool.

src/azure-cli/azure/cli/command_modules/acs/_params.py

@@ -833,6 +833,7 @@ def load_arguments(self, _):
         c.argument('max_surge', validator=validate_max_surge)
         c.argument('drain_timeout', type=int)
         c.argument('node_soak_duration', type=int)
+        c.argument("undrainable_node_behavior", default='Schedule')
         c.argument('mode', get_enum_type(node_mode_types))
         c.argument('scale_down_mode', arg_type=get_enum_type(scale_down_modes))
         c.argument('max_pods', type=int, options_list=['--max-pods', '-m'])
@@ -873,6 +874,7 @@ def load_arguments(self, _):
         c.argument('max_surge', validator=validate_max_surge)
         c.argument('drain_timeout', type=int)
         c.argument('node_soak_duration', type=int)
+        c.argument("undrainable_node_behavior")
         c.argument('mode', get_enum_type(node_mode_types))
         c.argument('scale_down_mode', arg_type=get_enum_type(scale_down_modes))
         c.argument('allowed_host_ports', nargs='+', validator=validate_allowed_host_ports)
@@ -891,6 +893,7 @@ def load_arguments(self, _):
         c.argument('max_surge', validator=validate_max_surge)
         c.argument('drain_timeout', type=int)
         c.argument('node_soak_duration', type=int)
+        c.argument("undrainable_node_behavior")
         c.argument('snapshot_id', validator=validate_snapshot_id)
         c.argument('yes', options_list=['--yes', '-y'], help='Do not prompt for confirmation.', action='store_true')
 

src/azure-cli/azure/cli/command_modules/acs/agentpool_decorator.py

@@ -1143,6 +1143,26 @@ def get_node_soak_duration(self):
         # this parameter does not need validation
         return node_soak_duration
 
+    def get_undrainable_node_behavior(self) -> str:
+        """Obtain the value of undrainable_node_behavior.
+
+        :return: string
+        """
+        # read the original value passed by the command
+        undrainable_node_behavior = self.raw_param.get("undrainable_node_behavior")
+        # In create mode, try to read the property value corresponding to the parameter from the `agentpool` object
+        if self.decorator_mode == DecoratorMode.CREATE:
+            if (
+                self.agentpool and
+                self.agentpool.upgrade_settings and
+                self.agentpool.upgrade_settings.undrainable_node_behavior is not None
+            ):
+                undrainable_node_behavior = self.agentpool.upgrade_settings.undrainable_node_behavior
+
+        # this parameter does not need dynamic completion
+        # this parameter does not need validation
+        return undrainable_node_behavior
+
     def get_vm_set_type(self) -> str:
         """Obtain the value of vm_set_type, default value is CONST_VIRTUAL_MACHINE_SCALE_SETS.
 
@@ -1821,6 +1841,10 @@ def set_up_upgrade_settings(self, agentpool: AgentPool) -> AgentPool:
         if node_soak_duration:
             upgrade_settings.node_soak_duration_in_minutes = node_soak_duration
 
+        undrainable_node_behavior = self.context.get_undrainable_node_behavior()
+        if undrainable_node_behavior:
+            upgrade_settings.undrainable_node_behavior = undrainable_node_behavior
+
         agentpool.upgrade_settings = upgrade_settings
         return agentpool
 
@@ -2210,6 +2234,11 @@ def update_upgrade_settings(self, agentpool: AgentPool) -> AgentPool:
             upgrade_settings.node_soak_duration_in_minutes = node_soak_duration
             agentpool.upgrade_settings = upgrade_settings
 
+        undrainable_node_behavior = self.context.get_undrainable_node_behavior()
+        if undrainable_node_behavior:
+            upgrade_settings.undrainable_node_behavior = undrainable_node_behavior
+            agentpool.upgrade_settings = upgrade_settings
+
         return agentpool
 
     def update_vm_properties(self, agentpool: AgentPool) -> AgentPool:

src/azure-cli/azure/cli/command_modules/acs/custom.py

@@ -2410,6 +2410,7 @@ def aks_agentpool_add(
     max_surge=None,
     drain_timeout=None,
     node_soak_duration=None,
+    undrainable_node_behavior=None,
     mode=CONST_NODEPOOL_MODE_USER,
     scale_down_mode=CONST_SCALE_DOWN_MODE_DELETE,
     max_pods=None,
@@ -2478,6 +2479,7 @@ def aks_agentpool_update(
     max_surge=None,
     drain_timeout=None,
     node_soak_duration=None,
+    undrainable_node_behavior=None,
     mode=None,
     scale_down_mode=None,
     no_wait=False,
@@ -2530,6 +2532,7 @@ def aks_agentpool_upgrade(cmd, client, resource_group_name, cluster_name,
                           max_surge=None,
                           drain_timeout=None,
                           node_soak_duration=None,
+                          undrainable_node_behavior=None,
                           snapshot_id=None,
                           no_wait=False,
                           aks_custom_headers=None,
@@ -2549,7 +2552,7 @@ def aks_agentpool_upgrade(cmd, client, resource_group_name, cluster_name,
         )
 
     # Note: we exclude this option because node image upgrade can't accept nodepool put fields like max surge
-    if (max_surge or drain_timeout or node_soak_duration) and node_image_only:
+    if (max_surge or drain_timeout or node_soak_duration or undrainable_node_behavior) and node_image_only:
         raise MutuallyExclusiveArgumentError(
             'Conflicting flags. Unable to specify max-surge/drain-timeout/node-soak-duration with node-image-only.'
             'If you want to use max-surge/drain-timeout/node-soak-duration with a node image upgrade, please first '
@@ -2604,6 +2607,8 @@ def aks_agentpool_upgrade(cmd, client, resource_group_name, cluster_name,
         instance.upgrade_settings.drain_timeout_in_minutes = drain_timeout
     if isinstance(node_soak_duration, int) and node_soak_duration >= 0:
         instance.upgrade_settings.node_soak_duration_in_minutes = node_soak_duration
+    if undrainable_node_behavior:
+        instance.upgrade_settings.undrainable_node_behavior = undrainable_node_behavior
 
     # custom headers
     aks_custom_headers = extract_comma_separated_string(

src/azure-cli/azure/cli/command_modules/acs/linter_exclusions.yml

@@ -200,4 +200,17 @@ aks nodepool add:
       disable_windows_outbound_nat:
         rule_exclusions:
         - option_length_too_long
+      undrainable_node_behavior:
+        rule_exclusions:
+        - option_length_too_long
+aks nodepool update:
+  parameters:
+      undrainable_node_behavior:
+        rule_exclusions:
+        - option_length_too_long
+aks nodepool upgrade:
+  parameters:
+      undrainable_node_behavior:
+        rule_exclusions:
+        - option_length_too_long
 ...