Merge pull request #3741 from Skarlso/suspend_asg_processes

Suspend and Resume ASG Processes

Author: k8s-ci-robot

config/crd/bases/infrastructure.cluster.x-k8s.io_awsmachinepools.yaml

@@ -846,6 +846,37 @@ spec:
                       type: string
                   type: object
                 type: array
+              suspendProcesses:
+                description: SuspendProcesses defines a list of processes to suspend
+                  for the given ASG. This is constantly reconciled. If a process is
+                  removed from this list it will automatically be resumed.
+                properties:
+                  all:
+                    type: boolean
+                  processes:
+                    description: Processes defines the processes which can be enabled
+                      or disabled individually.
+                    properties:
+                      addToLoadBalancer:
+                        type: boolean
+                      alarmNotification:
+                        type: boolean
+                      azRebalance:
+                        type: boolean
+                      healthCheck:
+                        type: boolean
+                      instanceRefresh:
+                        type: boolean
+                      launch:
+                        type: boolean
+                      replaceUnhealthy:
+                        type: boolean
+                      scheduledActions:
+                        type: boolean
+                      terminate:
+                        type: boolean
+                    type: object
+                type: object
             required:
             - awsLaunchTemplate
             - maxSize

docs/book/src/topics/suspend-asg-processes.md

@@ -0,0 +1,105 @@
+# Suspend ASG Processes
+
+- **Feature status:** Experimental
+- **Feature gate:** MachinePool=true
+
+MachinePool allows users to manage many machines as a single entity. Infrastructure providers implement a separate CRD that handles infrastructure side of the feature.
+
+## Suspend Processes
+
+It's possible to suspend certain processes for ASG. The list of processes can be found [here](https://docs.aws.amazon.com/autoscaling/ec2/APIReference/API_SuspendProcesses.html).
+
+To utilize this feature, simply denote the list of processes that are desired to be suspended.
+
+```yaml
+---
+apiVersion: infrastructure.cluster.x-k8s.io/v1beta2
+kind: AWSMachinePool
+metadata:
+  name: capa-mp-0
+spec:
+  minSize: 1
+  maxSize: 10
+  availabilityZones:
+    - "${AWS_AVAILABILITY_ZONE}"
+  awsLaunchTemplate:
+    instanceType: "${AWS_CONTROL_PLANE_MACHINE_TYPE}"
+    sshKeyName: "${AWS_SSH_KEY_NAME}"  
+  suspendProcesses:
+    processes:
+        launch: true
+        alarmNotification: true
+        azRebalance: true
+---
+```
+
+## Resume Processes
+
+If a service is desired to be resumed, simply remove it from the list of suspended processes. The reconciler will then
+resume any process that is not part of the desired suspended processes list.
+
+```yaml
+---
+apiVersion: infrastructure.cluster.x-k8s.io/v1beta2
+kind: AWSMachinePool
+metadata:
+  name: capa-mp-0
+spec:
+  minSize: 1
+  maxSize: 10
+  availabilityZones:
+    - "${AWS_AVAILABILITY_ZONE}"
+  awsLaunchTemplate:
+    instanceType: "${AWS_CONTROL_PLANE_MACHINE_TYPE}"
+    sshKeyName: "${AWS_SSH_KEY_NAME}"  
+  suspendProcesses:
+    processes:
+      launch: true
+---
+```
+
+_Note_ that now `AlarmNotification` and `AZRebalance` will be resumed, but the reconciler will not try to suspend
+`Launch` again. So it doesn't incur additional expensive, redundant API calls.
+
+## Optional `All`
+
+An option is also provided to suspend all processes without having to set each of them to `true`. Simply use `all` like
+this:
+
+```yaml
+apiVersion: infrastructure.cluster.x-k8s.io/v1beta2
+kind: AWSMachinePool
+metadata:
+  name: capa-mp-0
+spec:
+  minSize: 1
+  maxSize: 10
+  availabilityZones:
+    - "${AWS_AVAILABILITY_ZONE}"
+  awsLaunchTemplate:
+    instanceType: "${AWS_CONTROL_PLANE_MACHINE_TYPE}"
+    sshKeyName: "${AWS_SSH_KEY_NAME}"  
+  suspendProcesses:
+    all: true
+```
+
+To exclude individual processes from `all` simply add them with value `false`:
+
+```yaml
+apiVersion: infrastructure.cluster.x-k8s.io/v1beta2
+kind: AWSMachinePool
+metadata:
+  name: capa-mp-0
+spec:
+  minSize: 1
+  maxSize: 10
+  availabilityZones:
+    - "${AWS_AVAILABILITY_ZONE}"
+  awsLaunchTemplate:
+    instanceType: "${AWS_CONTROL_PLANE_MACHINE_TYPE}"
+    sshKeyName: "${AWS_SSH_KEY_NAME}"  
+  suspendProcesses:
+    all: true
+    processes:
+      launch: false
+```

exp/api/v1beta1/conversion.go

@@ -21,6 +21,7 @@ import (
 	infrav1beta1 "sigs.k8s.io/cluster-api-provider-aws/api/v1beta1"
 	infrav1 "sigs.k8s.io/cluster-api-provider-aws/api/v1beta2"
 	infrav1exp "sigs.k8s.io/cluster-api-provider-aws/exp/api/v1beta2"
+	utilconversion "sigs.k8s.io/cluster-api/util/conversion"
 	"sigs.k8s.io/controller-runtime/pkg/conversion"
 )
 
@@ -30,7 +31,17 @@ func (src *AWSMachinePool) ConvertTo(dstRaw conversion.Hub) error {
 	if err := Convert_v1beta1_AWSMachinePool_To_v1beta2_AWSMachinePool(src, dst, nil); err != nil {
 		return err
 	}
-	
+
+	// Manually restore data.
+	restored := &infrav1exp.AWSMachinePool{}
+	if ok, err := utilconversion.UnmarshalData(src, restored); err != nil || !ok {
+		return err
+	}
+
+	if restored.Spec.SuspendProcesses != nil {
+		dst.Spec.SuspendProcesses = restored.Spec.SuspendProcesses
+	}
+
 	return nil
 }
 
@@ -41,8 +52,8 @@ func (r *AWSMachinePool) ConvertFrom(srcRaw conversion.Hub) error {
 	if err := Convert_v1beta2_AWSMachinePool_To_v1beta1_AWSMachinePool(src, r, nil); err != nil {
 		return err
 	}
-	
-	return nil
+
+	return utilconversion.MarshalData(src, r)
 }
 
 // ConvertTo converts the v1beta1 AWSMachinePoolList receiver to a v1beta2 AWSMachinePoolList.
@@ -75,7 +86,7 @@ func (r *AWSManagedMachinePool) ConvertFrom(srcRaw conversion.Hub) error {
 	if err := Convert_v1beta2_AWSManagedMachinePool_To_v1beta1_AWSManagedMachinePool(src, r, nil); err != nil {
 		return err
 	}
-	
+
 	return nil
 }
 
@@ -147,3 +158,20 @@ func Convert_v1beta1_Instance_To_v1beta2_Instance(in *infrav1beta1.Instance, out
 func Convert_v1beta2_AWSLaunchTemplate_To_v1beta1_AWSLaunchTemplate(in *infrav1exp.AWSLaunchTemplate, out *AWSLaunchTemplate, s apiconversion.Scope) error {
 	return autoConvert_v1beta2_AWSLaunchTemplate_To_v1beta1_AWSLaunchTemplate(in, out, s)
 }
+
+func Convert_v1beta1_AWSMachinePoolSpec_To_v1beta2_AWSMachinePoolSpec(in *AWSMachinePoolSpec, out *infrav1exp.AWSMachinePoolSpec, s apiconversion.Scope) error {
+	return autoConvert_v1beta1_AWSMachinePoolSpec_To_v1beta2_AWSMachinePoolSpec(in, out, s)
+}
+
+func Convert_v1beta2_AWSMachinePoolSpec_To_v1beta1_AWSMachinePoolSpec(in *infrav1exp.AWSMachinePoolSpec, out *AWSMachinePoolSpec, s apiconversion.Scope) error {
+	return autoConvert_v1beta2_AWSMachinePoolSpec_To_v1beta1_AWSMachinePoolSpec(in, out, s)
+}
+
+func Convert_v1beta1_AutoScalingGroup_To_v1beta2_AutoScalingGroup(in *AutoScalingGroup, out *infrav1exp.AutoScalingGroup, s apiconversion.Scope) error {
+	return autoConvert_v1beta1_AutoScalingGroup_To_v1beta2_AutoScalingGroup(in, out, s)
+}
+
+func Convert_v1beta2_AutoScalingGroup_To_v1beta1_AutoScalingGroup(in *infrav1exp.AutoScalingGroup, out *AutoScalingGroup, s apiconversion.Scope) error {
+	// explicitly ignore CurrentlySuspended.
+	return autoConvert_v1beta2_AutoScalingGroup_To_v1beta1_AutoScalingGroup(in, out, s)
+}