Adding Servers
Managing Server Pools
Server pools are groups of worker nodes within your Cloudspace that share the same configuration. You can add new node pools to an existing Cloudspace at any time to scale your capacity, add different instance types (like GPUs), or introduce a mix of Spot and On-Demand nodes for hybrid configurations.
Why Add More Server Pools?
- Scaling: Increase the overall compute, memory, or GPU capacity of your cluster.
- Hybrid Clusters: Add an On-Demand pool to a Spot-only cluster (or vice-versa) for better cost/stability balance. See Spot vs On-Demand.
- Specialized Hardware: Add a pool with GPU instances for machine learning or other intensive workloads.
- Targeted Workloads: Use node labels and annotation selectors to schedule specific applications onto specific node pools (e.g., running databases on On-Demand nodes).
Adding a Server Pool to an existing Cloudspace
- Log in to the Rackspace Spot Dashboard.
- Select Cloudspace Actions: Navigate to the Cloudspace Actions drop-down.
- Select "Edit Cloudspace"
- Click "Add Server Pool"
- Configure the New Pool:
- Server Class: Choose Spot or On-Demand.
- Instance Type/Flavor: Select the desired General Purpose, Compute Heavy, Memory Heavy, or GPU servers for the nodes in this pool. Each Server pool contains a single type/flavor. See Cloud Server Classes
- Number of Nodes: Specify the initial number of nodes for this pool.
- Alternatively, if Autoscaling a Spot Node Pool is enabled, set the maximum server count to scale to.
Spot Only Max Bid Price: If creating a Spot pool, enter the maximum price per hour you're willing to pay. See Bidding Best Practices.Optional Labels & Annotations: Apply custom Kubernetes labels or annotations to nodes in this pool for scheduling or identification purposes.
- Save and Confirm: Review the configuration and confirm the addition of the server pool. The new billing amount will be reflected before confirmation
Please allow a few minutes for provisioning of the new servers.
Verifying New Nodes
Once the new servers are provisioned and have joined the cluster, you can verify them using kubectl
:
# Ensure your KUBECONFIG is set for the correct Cloudspace
kubectl get nodes -A -o wide
By adding node pools strategically, you can effectively scale and tailor your Cloudspace to meet the evolving demands of your applications.
Was this page helpful?