Skip to main content

How to add and modify a node group

Node groups let you segment the nodes of your Kubernetes cluster according to the needs of your workloads. This guide explains how to add, modify and delete node groups from the Hikube console.

Prerequisites​

  • A deployed Hikube Kubernetes cluster (see the quick start)
  • The cluster kubeconfig downloaded from the console (Kubeconfig button, reserved for project administrators), to check the nodes with kubectl

Steps​

1. Understand the instance types​

Hikube offers three instance series suited to different use cases:

SeriesCPU:RAM ratioUse case
Standard (S)1:2Economical use, development, testing
Universal (U)1:4General use: web servers, applications
Memory (M)1:8Memory-optimized: databases, caches

The flavors of each series are detailed in the concepts.

2. Open the cluster edit page​

  1. In the console, open Infrastructure > Kubernetes.
  2. Open the cluster's Actions menu and choose Edit (or click Edit on the cluster detail page).

The edit page shows the General information, Node groups and Extensions & Addons sections, as well as the project quota gauges.

3. Add a node group​

  1. In the Node groups section, click Add node group. A new card opens.
  2. Fill in the fields:
    • Group name: for example compute (3 to 16 characters: lowercase letters, digits and hyphens);
    • Node disk size: for example 100 GB;
    • Minimum nodes and Maximum nodes: for example 1 and 10;
    • Instance type: for example Universal (U) series, size 4XLarge (u1.4xlarge);
    • Exposed on the internet (Public IP): enable only if this group must receive incoming traffic (Ingress NGINX);
    • GPU: if needed, see Add GPUs.
  3. Click Save.
tip

Choose a descriptive name for your groups (compute, web, monitoring, gpu) to make cluster management easier.

4. Modify an existing group​

In the Node groups section, expand the group's card, change the desired fields (instance type, node disk size, minimum or maximum number of nodes, exposure), then click Save.

warning

Changing the instance type or the disk size, whether up or down, gradually replaces all the nodes in the group, one at a time: the platform creates the new nodes and removes the old ones one by one, without waiting for each replacement to be ready. Data stored locally on the nodes, outside persistent volumes, is lost. A single-node group is therefore unavailable during the replacement (several minutes): plan at least two nodes for workloads that cannot tolerate interruption. The project quota must be able to accommodate the additional nodes during the replacement.

note

Avoid renaming an existing group: a renamed group is treated as a new group.

5. Add GPUs​

A card's GPU section only appears if GPUs are available for your project.

  • For a new group, select the GPU model and number per node. The GPU Operator addon is then enabled automatically and can no longer be unchecked.
  • A group created without GPUs cannot receive any: add a new GPU node group.
  • A group created with GPUs can change model or number, but must keep at least one GPU: to go back to nodes without GPUs, add a new group without GPUs instead.

See also Provision GPUs in Kubernetes.

6. Delete a node group​

warning

Before deleting a group, make sure the workloads running on it can be rescheduled on other groups. Use kubectl drain on the relevant nodes if necessary.

  1. In the Node groups section, click the Remove this group icon on the relevant card.
  2. Click Save.

The first group of the cluster cannot be deleted; a cluster must always keep at least one node group.

Verification​

After saving, the console shows "Cluster updated" and returns to the detail page. The Node Pools section lists each group with its instance type and its number of active nodes.

In the cluster, watch the new nodes arrive:

export KUBECONFIG=~/Downloads/kubeconfig-<cluster-name>.yaml
kubectl get nodes -w

Expected result:

NAME                        STATUS   ROLES    AGE   VERSION
my-cluster-general-xxxxx    Ready    <none>   10m   v1.xx.x
my-cluster-compute-yyyyy    Ready    <none>   2m    v1.xx.x

Going further​