Container Service for Kubernetes:Best practices for associating deployment sets with node pools

Cluster feature usage guidelines

• Deployment sets are supported by ACK dedicated clusters and ACK managed clusters.

• You can associate a deployment set with a node pool only when you create the node pool. Existing node pools cannot have deployment sets enabled. You can associate only one deployment set with each node pool.

• You cannot manually add ECS instances to or remove ECS instances from deployment sets. If you want to change the number of ECS instances in a deployment set, you can scale the node pool with which the deployment set is associated. For more information, see Create and manage a node pool.

• After you associate a deployment set with a node pool, the node pool does not support preemptible instances.

Deployment set quotas and specifications limits

• Deployment set quota limits

  • By default, node pool deployment sets are implemented based on a high-availability strategy. In a deployment set that adopts the high availability strategy, you can create up to 20 ECS instances per zone. You can use the following formula to calculate the maximum number of ECS instances that you can create in a deployment set within an Alibaba Cloud region: 20 × Number of zones within the region. Ensure that there are sufficient ECS instances within the deployment set and adequate stock of the required instance types. For more information, see Deployment set.

    You cannot increase the number of ECS instances in a deployment set. However, if you want to increase the maximum number of deployment sets that your Alibaba Cloud account can have, request a quota increase in the Quota Center console. For more information about the limits and quotas of deployment sets, see Deployment sets.

  • Insufficient instance resources within the region may result in a failure to create ECS instances or start pay-as-you-go instances that were stopped in economical mode in a deployment set. Wait for a while and then try to create or start the instances again.

• Instance family limits

  Deployment strategies that can be used may vary based on the instance family. The following table describes the deployment strategies that are supported by different instance families.

  Note

  To query the instance families that support a specific deployment strategy, call the DescribeDeploymentSetSupportedInstanceTypeFamily operation.

  Deployment strategy	Instance families that support the deployment strategy
  High availability strategy or high availability group strategy	g8a, g8i, g8y, g7se, g7a, g7, g7h, g7t, g7ne, g7nex, g6, g6e, g6a, g5, g5ne, sn2ne, sn2, and sn1 c8a, c8i, c8y, c7se, c7, c7t, c7nex, c7a, c6, c6a, c6e, c5, ic5, and sn1ne r8a, r8i, r8y, r7, r7se, r7t, r7a, r6, r6e, r6a, re6, re6p, r5, re4, se1ne, and se1 hfc8i, hfg8i, hfr8i, hfc7, hfg7, hfr7, hfc6, hfg6, hfr6, hfc5, and hfg5 d3c, d2s, d2c, d1, d1ne, d1-c14d3, and d1-c8d3 i3g, i3, i2, i2g, i2ne, i2gne, and i1 ebmg5, ebmc7, ebmg7, ebmr7, sccgn6, scch5, scch5s, sccg5, and sccg5s e, t6, xn4, mn4, n4, e4, n2, and n1 gn6i
  Low latency strategy	g8a, g8i, g8ae, and g8y c8a, c8i, c8ae, and c8y ebmc8i, ebmg8i, and ebmr8i r8a, r8i, r8ae, and r8y ebmc7, ebmg7, and ebmr7

Pod anti-affinity is configured for pods that belong to the same workload in the same node pool and each node can host only one pod

This example shows how to schedule three replicated pods in a Deployment to three different nodes.

1. Create a node pool that contains three nodes and specify the ID of the deployment set that you want to associate with the node pool. For more information, see Associate a deployment set with a node pool in the ACK console.

   After you create the node pool, click the name of the node pool on the Node Pools page. The Nodes tab shows the nodes that are contained in the node pool.

   Log on to the ECS console. In the left-side navigation pane, choose Deployment & Elasticity > Deployment Sets. The Deployment Sets page shows that the three nodes belong to the specified deployment set.

2. Create a YAML file based on the following content.

   Deploy three pods and configure anti-affinity to schedule the pods to three different nodes.

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: nginx
     labels:
       app: nginx
   spec:
     replicas: 3
     selector:
       matchLabels:
         app: nginx
     template:
       metadata:
         name: nginx
         labels:
           app: nginx
       spec:
         affinity:
           podAntiAffinity:
             requiredDuringSchedulingIgnoredDuringExecution:  # Configure the scheduling policy. 
             - labelSelector:
                 matchExpressions:
                 - key: app
                   operator: In
                   values:
                   - nginx
               topologyKey: kubernetes.io/hostname
         nodeSelector:
             alibabacloud.com/nodepool-id: <nodepool-id>  # Specify the ID of the node pool. 
         containers:
         - name: nginx
           image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/openanolis/nginx:1.14.1-8.6
           resources:
             limits:
               cpu: 1
             requests:
               cpu: 1

   Results

   On the Deployments page, click the Deployment that you want to manage. The Pods tab shows that the pods are scheduled to three different nodes.

Pods that belong to the same workload in the same node pool can be scheduled to different nodes across multiple zones

1. Create a node pool that contains four nodes, specify the ID of the deployment set that you want to associate with the node pool, and select vSwitches that are deployed in multiple zones. For more information, see Associate a deployment set with a node pool in the ACK console.

   After you create the node pool, click the name of the node pool on the Node Pools page. Click the Nodes tab, four nodes and four zones are displayed. ECS instances are deployed in four different zones and are added to the same deployment set based on the balanced distribution policy of Auto Scaling.

   Log on to the ECS console. In the left-side navigation pane, choose Deployment & Elasticity > deployment set. The Deployment Sets page shows that the four nodes in the node pool belong to the specified deployment set.

2. Create a YAML file based on the following content.

   You can use pod topology spread constraints to schedule pods that belong to the same application to different nodes across multiple zones. For more information, see Pod topology spread constraints.

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: nginx
     labels:
       app: nginx
   spec:
     replicas: 4
     selector:
       matchLabels:
         app: nginx
     template:
       metadata:
         name: nginx
         labels:
           app: nginx
       spec:
         topologySpreadConstraints:
           - maxSkew: 1
             topologyKey: kubernetes.io/hostname
             whenUnsatisfiable: DoNotSchedule
             labelSelector:
                 matchLabels:
                   app: nginx
           - maxSkew: 1
             topologyKey: topology.kubernetes.io/zone
             whenUnsatisfiable: DoNotSchedule
             labelSelector:
                 matchLabels:
                   app: nginx
         nodeSelector:
             alibabacloud.com/nodepool-id: <nodepool-id>  #Specify the ID of the node pool. 
         containers:
         - name: nginx
           image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/openanolis/nginx:1.14.1-8.6
           resources:
             limits:
               cpu: 1
             requests:
               cpu: 1

   Results

   On the Deployments page, click the Deployment that you want to manage. The Pods tab shows that pods belonging to the same application are scheduled to different nodes across multiple zones.