[AKS] az aks nodepool add/update/upgrade: Add --max-blocked-nodes to specify maximum number or percentage of extra nodes that are allowed to be blocked in the agent pool during an upgrade (Azure#8787)

Author: zjpjack-github

src/aks-preview/HISTORY.rst

@@ -12,6 +12,10 @@ To release a new version, please select a new version number (usually plus 1 to
 Pending
 +++++++
 
+18.0.0b9
++++++++
+* Add `--max-blocked-nodes` to the `az aks nodepool add/update/upgrade` commands.
+
 18.0.0b8
 +++++++
 * Populate location of managed namespaces using managed cluster's location

src/aks-preview/azext_aks_preview/_help.py

@@ -1934,6 +1934,9 @@
         - name: --max-unavailable
           type: string
           short-summary: The maximum number or percentage of nodes that can be simultaneously unavailable during upgrade. When specified, it represents the number or percent used, eg. 1 or 5%
+        - name: --max-blocked-nodes
+          type: string
+          short-summary: The maximum number or percentage of extra nodes that are allowed to be blocked in the agent pool during an upgrade when undrainable node behavior is Cordon. When specified, it represents the number or percent used, eg. 1 or 5%.
         - name: --kubelet-config
           type: string
           short-summary: Kubelet configurations for agent nodes.
@@ -2076,6 +2079,9 @@
         - name: --max-unavailable
           type: string
           short-summary: The maximum number or percentage of nodes that can be simultaneously unavailable during upgrade. When specified, it represents the number or percent used, eg. 1 or 5%
+        - name: --max-blocked-nodes
+          type: string
+          short-summary: The maximum number or percentage of extra nodes that are allowed to be blocked in the agent pool during an upgrade when undrainable node behavior is Cordon. When specified, it represents the number or percent used, eg. 1 or 5%.
         - name: --aks-custom-headers
           type: string
           short-summary: Send custom headers. When specified, format should be Key1=Value1,Key2=Value2
@@ -2128,6 +2134,9 @@
         - name: --max-unavailable
           type: string
           short-summary: The maximum number or percentage of nodes that can be simultaneously unavailable during upgrade. When specified, it represents the number or percent used, eg. 1 or 5%
+        - name: --max-blocked-nodes
+          type: string
+          short-summary: The maximum number or percentage of extra nodes that are allowed to be blocked in the agent pool during an upgrade when undrainable node behavior is Cordon. When specified, it represents the number or percent used, eg. 1 or 5%.
         - name: --mode
           type: string
           short-summary: The mode for a node pool which defines a node pool's primary function. If set as "System", AKS prefers system pods scheduling to node pools with mode `System`. Learn more at https://aka.ms/aks/nodepool/mode.

src/aks-preview/azext_aks_preview/_params.py

@@ -215,6 +215,7 @@
     validate_bootstrap_container_registry_resource_id,
     validate_gateway_prefix_size,
     validate_max_unavailable,
+    validate_max_blocked_nodes,
     validate_resource_group_parameter,
     validate_location_resource_group_cluster_parameters,
 )
@@ -1594,6 +1595,7 @@ def load_arguments(self, _):
         c.argument("node_soak_duration", type=int)
         c.argument("undrainable_node_behavior")
         c.argument("max_unavailable", validator=validate_max_unavailable)
+        c.argument("max_blocked_nodes", validator=validate_max_blocked_nodes)
         c.argument("mode", arg_type=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"])
@@ -1726,6 +1728,7 @@ def load_arguments(self, _):
         c.argument("node_soak_duration", type=int)
         c.argument("undrainable_node_behavior")
         c.argument("max_unavailable", validator=validate_max_unavailable)
+        c.argument("max_blocked_nodes", validator=validate_max_blocked_nodes)
         c.argument("mode", arg_type=get_enum_type(node_mode_types))
         c.argument("scale_down_mode", arg_type=get_enum_type(scale_down_modes))
         # extensions
@@ -1797,6 +1800,7 @@ def load_arguments(self, _):
         c.argument("node_soak_duration", type=int)
         c.argument("undrainable_node_behavior")
         c.argument("max_unavailable", validator=validate_max_unavailable)
+        c.argument("max_blocked_nodes", validator=validate_max_blocked_nodes)
         c.argument("snapshot_id", validator=validate_snapshot_id)
         c.argument(
             "yes",

src/aks-preview/azext_aks_preview/_validators.py

@@ -519,6 +519,22 @@ def validate_max_unavailable(namespace):
         raise CLIError("--max-unavailable should be an int or percentage")
 
 
+def validate_max_blocked_nodes(namespace):
+    """validates parameters max blocked nodes is positive integers or percents."""
+    if namespace.max_blocked_nodes is None:
+        return
+    int_or_percent = namespace.max_blocked_nodes
+    if int_or_percent.endswith('%'):
+        int_or_percent = int_or_percent.rstrip('%s')
+
+    try:
+        if int(int_or_percent) < 0:
+            raise InvalidArgumentValueError('--max-blocked-nodes must be be positive')
+    except ValueError:
+        # pylint: disable=raise-missing-from
+        raise InvalidArgumentValueError('--max-blocked-nodes should be an int or percentage')
+
+
 def validate_assign_identity(namespace):
     if namespace.assign_identity is not None:
         if namespace.assign_identity == '':

src/aks-preview/azext_aks_preview/agentpool_decorator.py

@@ -503,6 +503,26 @@ def get_max_unavailable(self) -> str:
         # this parameter does not need validation
         return max_unavailable
 
+    def get_max_blocked_nodes(self) -> str:
+        """Obtain the value of max_blocked_nodes.
+
+        :return: string
+        """
+        # read the original value passed by the command
+        max_blocked_nodes = self.raw_param.get("max_blocked_nodes")
+        # 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.max_blocked_nodes is not None
+            ):
+                max_blocked_nodes = self.agentpool.upgrade_settings.max_blocked_nodes
+
+        # this parameter does not need dynamic completion
+        # this parameter does not need validation
+        return max_blocked_nodes
+
     def get_enable_artifact_streaming(self) -> bool:
         """Obtain the value of enable_artifact_streaming.
         :return: bool
@@ -1127,6 +1147,10 @@ def set_up_upgrade_settings(self, agentpool: AgentPool) -> AgentPool:
         if max_unavailable:
             upgrade_settings.max_unavailable = max_unavailable
 
+        max_blocked_nodes = self.context.get_max_blocked_nodes()
+        if max_blocked_nodes:
+            upgrade_settings.max_blocked_nodes = max_blocked_nodes
+
         agentpool.upgrade_settings = upgrade_settings
         return agentpool
 
@@ -1352,6 +1376,11 @@ def update_upgrade_settings(self, agentpool: AgentPool) -> AgentPool:
             upgrade_settings.undrainable_node_behavior = undrainable_node_behavior
             agentpool.upgrade_settings = upgrade_settings
 
+        max_blocked_nodes = self.context.get_max_blocked_nodes()
+        if max_blocked_nodes:
+            upgrade_settings.max_blocked_nodes = max_blocked_nodes
+            agentpool.upgrade_settings = upgrade_settings
+
         max_unavailable = self.context.get_max_unavailable()
         if max_unavailable:
             upgrade_settings.max_unavailable = max_unavailable

src/aks-preview/azext_aks_preview/custom.py

@@ -1359,6 +1359,7 @@ def aks_agentpool_add(
     node_soak_duration=None,
     undrainable_node_behavior=None,
     max_unavailable=None,
+    max_blocked_nodes=None,
     mode=CONST_NODEPOOL_MODE_USER,
     scale_down_mode=CONST_SCALE_DOWN_MODE_DELETE,
     max_pods=0,
@@ -1442,6 +1443,7 @@ def aks_agentpool_update(
     node_soak_duration=None,
     undrainable_node_behavior=None,
     max_unavailable=None,
+    max_blocked_nodes=None,
     mode=None,
     scale_down_mode=None,
     no_wait=False,
@@ -1534,6 +1536,7 @@ def aks_agentpool_upgrade(cmd,
                           node_soak_duration=None,
                           undrainable_node_behavior=None,
                           max_unavailable=None,
+                          max_blocked_nodes=None,
                           snapshot_id=None,
                           no_wait=False,
                           aks_custom_headers=None,
@@ -1553,13 +1556,24 @@ def aks_agentpool_upgrade(cmd,
         )
 
     # Note: we exclude this option because node image upgrade can't accept nodepool put fields like max surge
-    hasUpgradeSetting = max_surge or drain_timeout or node_soak_duration or undrainable_node_behavior or max_unavailable
+    hasUpgradeSetting = (
+        max_surge or
+        drain_timeout or
+        node_soak_duration or
+        undrainable_node_behavior or
+        max_unavailable or
+        max_blocked_nodes)
     if hasUpgradeSetting 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 "
-            "update max-surge/drain-timeout/node-soak-duration using "
-            '"az aks nodepool update --max-surge/--drain-timeout/--node-soak-duration".'
+            "Conflicting flags. Unable to specify "
+            "max-surge/drain-timeout/node-soak-duration/undrainable-node-behavior/max-unavailable/max-blocked-nodes"
+            " with node-image-only.If you want to use "
+            "max-surge/drain-timeout/node-soak-duration/undrainable-node-behavior/max-unavailable/max-blocked-nodes"
+            " with a node image upgrade, please first update "
+            "max-surge/drain-timeout/node-soak-duration/undrainable-node-behavior/max-unavailable/max-blocked-nodes"
+            " using 'az aks nodepool update "
+            "--max-surge/--drain-timeout/--node-soak-duration/"
+            "--undrainable-node-behavior/--max-unavailable/--max-blocked-nodes'."
         )
 
     if node_image_only:
@@ -1621,6 +1635,8 @@ def aks_agentpool_upgrade(cmd,
         instance.upgrade_settings.undrainable_node_behavior = undrainable_node_behavior
     if max_unavailable:
         instance.upgrade_settings.max_unavailable = max_unavailable
+    if max_blocked_nodes:
+        instance.upgrade_settings.max_blocked_nodes = max_blocked_nodes
 
     # custom headers
     aks_custom_headers = extract_comma_separated_string(