Resizing a Lite Cluster Resource Pool

Scenarios

The demand for resources in a dedicated resource pool may change due to the changes of AI development services. In this case, you can resize your dedicated resource pool in ModelArts.

Before scaling in a resource pool, ensure that there are no services running in the pool. Alternatively, go to the resource pool details page, delete the nodes where no services are running to scale in the pool.

Constraints

• Only dedicated resource pools in the Running status can be resized.
• When scaling in a dedicated resource pool, the number of flavors or nodes of a flavor cannot be decreased to 0.

Resizing a Dedicated Resource Pool

You can resize a resource pool in any of the following ways:

• Adjusting the number of nodes of existing specifications
• Resizing the container engine space

1. Log in to the ModelArts management console. In the navigation pane, choose Dedicated Resource Pools > Elastic Cluster.
2. Add or delete nodes.

   To resize a resource pool, locate the resource pool on the Elastic Clusters page, and click Adjust Capacity. For Yearly/Month resource pools, only Expand Capacity is displayed on the page. To scale in such a resource pool, click the resource pool name to access the details page, and unsubscribe some nodes.

   In the Resource Configurations area, set AZ to Automatically allocated or Specifies AZ. Click Submit and then OK to save the changes.

   • If you select Automatically allocated, you can increase or decrease the number of target nodes to implement scaling. This allows you to adjust the number of nodes based on service requirements. Scale-out is the process of increasing the number of target nodes, while scale-in is the process of decreasing the number of target nodes. After the scaling, nodes are automatically allocated to AZs.
   • If you select Specifies AZ, you can allocate nodes to different AZs.

     When you purchase a resource pool, the nodes for certain specifications can be purchased by rack. When you resize the resource pool, the nodes are also added or deleted by rack. You can choose to purchase nodes by rack when creating a resource pool, which cannot be modified when resizing a resource pool. You can configure the rack quantity to change the number of Target Nodes.

     

     When you add nodes, you can specify the billing mode that is not the mode for charging resource pools. For example, you can create pay-per-use nodes in a yearly/monthly resource pool. If the billing mode is not specified, it will be the mode for charging resource pools.

3. Add a node pool.

   If you need more node pools, perform the following operations to create a node pool:

   Method 1: On the node pool details page, click the Node Pools tab, and click Create Node Pool.

   Method 2: In the resource pool list, click More > Create Node Pool in the Operation column of a resource pool to go to the Node Pools page and modify the container engine space size.

   For details about the parameters for creating a node pool, see Enabling Lite Cluster Resources.

   When creating a node pool, certain specifications support purchase by rack. If you choose these specifications, you can select either full rack or single node from the drop-down list to create the node pool. In full rack mode, the total number of nodes is calculated as the product of the number of nodes per rack and the number of racks.

   

   When creating a node pool, you can specify the billing mode for the nodes within the pool. For instance, you can create pay-per-use nodes within a yearly/monthly resource pool. If the billing mode is not specified, it will be the mode for charging resource pools.

4. Resize the container engine space.

   If you need larger container engine size, perform any of the following operations:

   • You can specify the container engine space size when creating a resource.
     • Method 1: You can specify the container engine space size when creating a resource pool. For details, see Advanced Setting under Specifications in Enabling Lite Cluster Resources.
     • Method 2: Click the name of a resource pool. On the displayed resource pool details page, click the Node Pools tab, click Create Node Pool, set Engine Space (GB) and click OK.
     • Method 3: In the resource pool list, click More > Create Node Pool in the Operation column of a resource pool to go to the Node Pools page and modify the container engine space size. (Only yearly/monthly node pools can be created.)
   • For existing resources, the container engine space can be modified.
     • Method 1: Click the name of a resource pool. On the displayed resource pool details page, click the Node Pools tab, locate the row that contains the target node pool, click Update in the Operation column, set Engine Space (GB) and click OK. (The container engine space of a new node is automatically set to the configured value by default.)
     • Method 2: Locate the target resource pool and click Adjust Capacity in the Operation column (only for new nodes).
       

       Resizing the container engine space is only applicable to new nodes. Furthermore, dockerBaseSize may vary across nodes of this flavor within the resource pool. Consequently, this can lead to discrepancies in the status of tasks distributed among different nodes.

   You can also change the container engine type on the preceding pages. Container engine, one of the most important components of Kubernetes, manages the lifecycle of images and containers. The kubelet interacts with a container runtime through the Container Runtime Interface (CRI). Containerd has a shorter call chain, fewer components, and lower resource requirements, making it more stable. For details about the differences between Containerd and Docker, see Container Engines.

   The CCE cluster version determines the available container engines. If it is earlier than 1.23, only Docker is supported. If it is 1.27 or later, only containerd is supported. For all other versions, both containerd and Docker are options.