Node Pools

Node pools let you group worker nodes with the same configuration (flavor, volume size, GPU type) within a Kubernetes cluster. You can add, remove, and scale node pools independently.

---

List

List all node pools for a cluster.

ace k8s cluster nodepool list <cluster-name-or-id>

You can also use the --cluster flag:

ace k8s cluster nodepool list --cluster my-cluster

Aliases: ls

Example output:

NAME                      ID                   NODES    FLAVOR       CLUSTER
pool1                     np-abc123            3        ACE.CPU.4C16R c-m-xyz789
gpu-pool                  np-def456            1        ACE.GPU.A100  c-m-xyz789

Note: The node count is obtained by querying the nodes endpoint for each pool. If a cluster is still provisioning, node pools may not be available yet.

---

Get / Show

Get detailed information about a specific node pool. Requires --cluster.

ace k8s cluster nodepool get <nodepool-name-or-id> --cluster <cluster-name-or-id>

Aliases: show

Examples:

ace k8s cluster nodepool get pool1 --cluster my-cluster
ace k8s cluster np show pool1 --cluster my-cluster

Example output:

  Name:                pool1
  ID:                  np-abc123
  Cluster:             c-m-xyz789
  Flavor:              ACE.CPU.4C16R
  Nodes:               3
  State:               active
  Volume:              50
  Autoscale:           1 - 5

---

Create

Add a new node pool to an existing cluster.

ace k8s cluster nodepool create <cluster-name-or-id> \
  --name <pool-name> \
  --flavor <flavor-name-or-id> \
  [options]

You can also use --cluster instead of a positional argument:

ace k8s cluster nodepool create --cluster my-cluster --name pool2 --flavor ACE.CPU.4C16R

Required Flags

Flag	Description
--name	Node pool name (lowercase alphanumeric with hyphens, max 63 chars)
--flavor	Flavor name or UUID for worker nodes

Optional Flags

Flag	Default	Description
--count	1	Number of nodes (1-8)
--volume-size	50	Volume size in GB per node (20-256)
--security-group		Security group UUID to attach
--gpu	false	Mark as GPU flavor

Examples

Basic node pool:

ace k8s cluster nodepool create my-cluster \
  --name pool2 \
  --flavor ACE.CPU.4C16R

GPU node pool with custom volume:

ace k8s cluster nodepool create my-cluster \
  --name gpu-pool \
  --flavor ACE.GPU.A100 \
  --count 2 \
  --volume-size 100 \
  --gpu

With security group:

ace k8s cluster nodepool create my-cluster \
  --name secure-pool \
  --flavor ACE.CPU.8C32R \
  --count 3 \
  --security-group <sg-uuid>

---

Scale

Set a node pool's desired node count, kubectl-style.

ace k8s cluster nodepool scale <nodepool-id> \
  --cluster <cluster-name-or-id> \
  --count <target-node-count>

Required Flags

Flag	Description
--cluster	Cluster name or ID
--count	Target number of nodes the pool should have (1-8).

Behaviour change in v1.4.2-beta — --count is now the target node count, not a delta. This mirrors kubectl scale deployment foo --replicas=N. The CLI reads the current node count, computes the add / remove delta automatically, and sends the right request to the backend.

Example: If the pool currently has 1 node and you run ... --count 3, the CLI adds 2. If the pool has 3 nodes and you run ... --count 1, the CLI removes 2.

Optional Flags

Flag	Description
--delta	Legacy mode — treat --count as a relative delta (+N to add, -N to remove). Provided for backward compatibility with scripts from older CLI versions. New scripts should use the default target-count semantics.

Examples

Scale pool to 3 nodes (target count):

ace k8s cluster nodepool scale pool1 \
  --cluster my-cluster \
  --count 3

Scale back down to 1 node:

ace k8s cluster nodepool scale pool1 \
  --cluster my-cluster \
  --count 1

Legacy delta mode (add 2 nodes regardless of current size):

ace k8s cluster nodepool scale pool1 \
  --cluster my-cluster \
  --count 2 --delta

---

Delete

Delete a node pool from a cluster.

ace k8s cluster nodepool delete <nodepool-id> --cluster <cluster-name-or-id>

Aliases: rm

Example:

ace k8s cluster nodepool delete pool2 --cluster my-cluster

---

Name Resolution

All node pool commands that accept a cluster reference support both cluster names and cluster IDs:

• If the value starts with c-, it is treated as a cluster ID

• Otherwise, the CLI lists all clusters and matches by name

Similarly, node pool get accepts either the pool name (hostnamePrefix) or pool ID.

---

Global Flags

Flag	Description
-o, --output	Output format: table (default) or json
-v, --verbose	Show full JSON response
--project	Override project ID
--region	Override region