Scaling

In the context of Magnum and Kubernetes, there are two types of scaling:

1. Scaling nodes

2. Scaling containers

Scaling nodes

For redundancy and to increase the number of resources available to your cluster, you can scale up the number of nodes (instances). You can also scale them down.

• When you scale up, Magnum creates a new Nova instance and adds it to the cluster

• When you scale down, Magnum deletes a node in the cluster. Pods might be deleted without draining, so it is a good idea to build your services to automatically recover from the loss of a node

Worker nodes in a cluster belong to node groups. Every cluster starts with a default-worker node group, which you can scale up and down. You can also create additional node groups, for example to add workers with a different flavour, and scale those independently.

Scaling the default worker node group

To scale up and down, use the openstack coe cluster resize command. The default nodegroup should be default-worker:

openstack coe nodegroup list mycluster
+--------------------------------------+----------------+-----------+--------------------------------------+------------+-----------------+--------+
| uuid                                 | name           | flavor_id | image_id                             | node_count | status          | role   |
+--------------------------------------+----------------+-----------+--------------------------------------+------------+-----------------+--------+
| 5884f8c8-1975-4f7a-9fcd-9d765638d442 | default-master | m3.small  | 7db23d42-5246-4b61-a667-696bb4aa9010 |          1 | CREATE_COMPLETE | master |
| ce7dc214-6dd7-4ddf-b83b-c5f784ae346b | default-worker | m3.small  | 7db23d42-5246-4b61-a667-696bb4aa9010 |          1 | CREATE_COMPLETE | worker |
+--------------------------------------+----------------+-----------+--------------------------------------+------------+-----------------+--------+

These can be scaled as so:

openstack coe cluster resize mycluster --nodegroup default-worker 2

Creating a new node group in the Nectar Dashboard

You can create and scale additional node groups from the Nectar Dashboard. Go to Project > Container Infra > Clusters, click on your cluster and select the Node Groups tab. This lists the node groups in your cluster, including the default-worker node group.

1. Click + Create Node Group. In the Details step, enter a name for the node group, and select a flavour and the initial node count.

   

2. In the Advanced step, you can optionally:

   • Tick Enable Autoscaling and set a minimum and maximum node count, so the node group scales automatically with demand
   • Tick Boot from volume to boot the nodes from a volume, and set the volume size and type
   • Apply Kubernetes node labels and taints to every node in the node group

   Click Submit to create the node group.

   

Once created, the new node group appears in the Node Groups tab and can be scaled in the same way as the default worker node group, using the Actions menu in the dashboard or openstack coe cluster resize with the --nodegroup option. If autoscaling is enabled, the node count is adjusted automatically within the minimum and maximum you defined.

Scaling containers

Adjusting the number of running containers is a standard operation in Kubernetes and can be performed using the kubectl command-line tool. For instance, you can scale the number of Pods in a Deployment using kubectl.

kubectl scale deployment <deployment-name> --replicas=<number> -n <namespace>

More information

For more information, refer to:

• Magnum scaling documentation.