Manage Clusters

Managing Kubernetes clusters created in the Hostman control panel

Dashboard

This tab displays CPU usage and memory consumption charts for master nodes. In the Information section on the right, you can download the configuration file for connecting to the cluster, as well as view the Kubernetes version and cluster configuration parameters.

CPU Core Load

• Capacity: The total number of CPU cores on the node. Measured in cores or millicores.
• Allocatable: The number of CPU cores available for scheduling applications on the node.
• Resource Limit: The total CPU limit set for pods. This indicates the maximum CPU usage allowed for all pods on the node.
• Resource Request: The total amount of CPU requested by all pods on the node.

Memory

• Capacity: The total amount of memory on the node.
• Allocatable: The amount of memory available for scheduling applications.
• Resource Limit: The total memory usage limit set for all pods on the node.
• Resource Request: The total amount of memory requested by all pods.

Pods Capacity

• Allocatable: The number of additional pods that can still be scheduled on the node.
• Requested: The number of pods currently scheduled on the node.

Pods Status

• Running: The pod has been successfully deployed and is running. All containers within the pod are up and running.
• Failed: An error occurred during the execution of one or more containers in the pod. For example, a container exited with a non-zero status code or failed to access required resources.
• Pending: The pod has not started yet because Kubernetes is still scheduling it to a node and has not yet allocated the necessary resources (e.g., node space, networking, or image pulling). To find out the reason, run the command:

kubectl describe node <node_name>

• Unknown: Kubernetes cannot determine the current status of the pod, possibly due to lost connection with the node where the pod was running. This may indicate connectivity issues or system failures.
• Succeeded: All containers in the pod have successfully completed their work. This status typically applies to pods that are designed for one-time execution (e.g., a Job).

Cluster Resources

You can add or remove nodes and node groups in the Resources tab in the control panel.

To add a node:

1. Hover over the node group and click Add nodes.

2. You can then select the number of nodes to add.

To remove a node or node group:

1. Click on the three dots next to the group or node and select Delete node / group.

Updating Kubernetes Version

You can update the Kubernetes version in cluster settings. You can only select a newer version; rolling back to one of the older version isn't possible.

Changing the version might cause performance issues in the cluster. We recommend doing this during a time when the project has the least activity.

To update the version:

1. In the Management tab, click Update version.

2. Select the version you need and click Change.

Restarting Worker Nodes

You can restart any worker node from the control panel. To do this:

1. In the Resources tab, expand the list of nodes in the group.

2. Click on the three dots next to the desired node and select Reboot node.

Cluster Maintenance Window

A cluster maintenance window is a period during which system updates are allowed. There are three available options:

• Never
• Any time
• Specific time

During the maintenance window, certificates may be updated, and security patches may be installed. Updates can affect both master and worker nodes. During this time, operations on the cluster will be blocked. Short service interruptions may occur, for example, during node restarts.

You can set the maintenance window when creating a cluster, and you can manage it later in the cluster settings.

To change the maintenance window:

1. In the Kubernetes section, click on the cluster.
2. On the Management tab, click Change next to Cluster maintenance window.

3. Select the preferred maintenance window and save the changes.

Autohealer

The system automatically checks the status of worker nodes every 10 minutes. If a node stops responding, the autohealer attempts to restart it. Each attempt can take up to 2 minutes. If the node does not recover, the attempt is repeated. The autohealer will continue trying for up to 14 minutes. If all attempts fail, the node will be recreated.

Autohealing is enabled by default for all nodes. To disable it for a specific node, add the following label:

kube-healer.kubernetes.io/healing-disabled: true

No more than 10 autohealing tasks can be processed at the same time.

Action History

The control panel logs all cluster-related actions, which can be viewed in the History tab.

The following events are logged:

• Cluster creation
• Creation of a group in the cluster
• Deletion of a group from the cluster
• Adding nodes to a group
• Removing nodes from a group
• Cluster deletion

Deleting a Cluster

You can delete a cluster in two ways:

• From the specific cluster's page by clicking on the delete icon.

• From the cluster list, click the three dots next to the cluster you want to delete and select Delete cluster.

In the next window, enter the name of the cluster to be deleted and confirm the deletion.