Manage workloads

Workloads are resources used to manage pod replicas, responsible for overseeing multiple containerized application instances. This section guides you on how to manage pods under different workloads (applications or services) within Control Hub.

Note

Olares members can only access their own namespaces, while Olares administrators can access all user and system namespaces.

Projects and namespaces

In Olares' Browse view, Projects and Namespaces combine to provide a two-level organizational structure for resources:

• Project: Categorize namespaces by "user" or "system" to quickly locate the resources belonging to a specific user or system module.
• Namespace: A project is made up of multiple namespaces, which are native Kubernetes isolation units used to distinguish between different applications, components, or environments.

Category	Namespace Prefix	Description
User projects	app-<olares-id>	Community applications installed from the Market, such as steamheadless-chenglin106, ollama-chenglin106, etc.
	user-space-<olares-id>	Built-in system applications such as Files, Market, Control Hub, Dashboard, Vault, etc.
	user-system-<olares-id>	User-related system services such as runtime components, schedulers, cross-application proxies, etc.
System projects	System	Cluster-level dependencies such as Kubernetes core, KubeSphere, Olares platform components, and necessary hardware drivers.

Manage workloads

In Olares, a workload represents an application or service running in the Olares cluster, responsible for managing one or more pods of an application. All workloads are deployed within a specific namespace. There are three workload types:

Type	Typical Scenarios	Characteristics
Deployment	Manages stateless applications.	Most common; automatically creates ReplicaSets and pods; supports horizontal scaling.
StatefulSet	Manages stateful applications like databases, distributed storage, and cache clusters.	Ensures pods start/terminate in order; provides a fixed network identity and persistent storage.
DaemonSet	Workloads for node-level monitoring, log collection, GPU drivers, etc.	Runs one pod per node, which is automatically added or deleted as nodes join or leave.

View workloads

To view a specific workload:

1. In the left navigation, click Browse,

2. In the first column, click the target namespace to expand it.

3. In the second column, click to expand the Deployment/StatefulSet/DaemonSet list, and then select the desired workload item.

4. View the workload details un the third column:

   Section	Description
   Basic information	Metadata such as cluster, project, creation/update time, and creator.
   Pods	Lists associated pods with info such as node, IP, and real-time CPU/memory charts. Click on the pods name to view the details.
   Ports	The container ports, protocols, and listener port numbers exposed by the pods.
   Environment Variables	A list of environment variables defined in the pod template.
   Labels	Configured in workload metadata for resource scheduling and filtering.
   Annotations	Defined in metadata.annotations, which functions similarly to labels, allowing for flexible management by controllers.
   Events	Events related to the workload within the last hour, such as scheduling, restarts, or image pulls.

Edit YAML configuration

In some advanced maintenance scenarios, you may need to directly adjust the number of pod replicas, add environment variables, modify probes, or update storage volume claims. In such cases, you can use the Edit YAML function for fine-grained configurations.

Proceed with caution

Directly modifying system resource YAML can affect cluster stability and performance. Operate with extreme caution and ensure you have a backup or are guided by a professional.

1. On the workload details page, click the edit_square button in the top-left corner.
2. Modify the configuration in the pop-up YAML editor.
3. Click OK to apply the new configuration immediately.

Stop or restart workloads

When you need to quickly troubleshoot an application or service, free up system resources, or reload a configuration, you can use the Stop/Start and Restart functions.

Function	Description
Stop/Start	Reduces the number of replicas to 0 and terminate the pods. To resume, click the Start button or adjust the number of replicas in YAML.
Restart	Stops and then immediately starts new pods based on the original replica count, used for quickly refreshing configurations or resolving transient failures.

1. On the workload details page, click the Stop or Restart button in the top-right corner.
2. In the confirmation dialog, type the pod name as prompted and click Confirm.
3. Observe the pod termination/start progress in the pods list to confirm that the status has returned to normal.

Monitor resource usage

You can monitor the cluster resource usage through two views:

View	Use case	Description
Namespaces	When you need to horizontally compare source usage across multiple applications or system modules.	Summarizes key metrics like CPU, memory, and traffic for a namespace to quickly locate resource hotspots and evaluate quota usage.
Pods	When you need to drill down into a single pod for troubleshooting or performance analysis.	Provides real-time status, resource charts, and event logs for each pod and its containers, enabling fine-grained maintenance.

You can use the Namespaces view to first identify "who" is consuming too many resources, and then use the Pods view to pinpoint "which specific pod/container" is causing the issue. This helps you monitor and maintain the cluster from a global to a detailed level.

Namespaces view

The Namespaces view aggregates key metrics like CPU, memory, and traffic by namespace, allowing you to quickly discover resource hotspots and compare quota usage for efficient performance tuning and troubleshooting.

1. In the left navigation bar, click Namespaces.

2. Use the dropdown menu at the top to switch between different users' namespaces, or use the search box for a precise search by keyword.

   Column	Description
   Namespace	The name of the namespace.
   CPU usage	Current CPU usage (Supports sorting).
   Memory usage	Current memory usage (Supports sorting).
   Pods	The number of pods running in the namespace.
   Outbound traffic	The outbound traffic rate.
   Inbound traffic	The inbound traffic rate.

Clicking on any namespace will take you to its resource details page:

Section	Description
Quota card	The used percentage of CPU/memory quotas.
Pods list	List of Pods under the namespace; Supports searching by Pods name and sorting by CPU/memory/traffic.
Pod	View image, node, ports, and real-time resource charts of the Pods. Click the Pod name to view the container charts.

Pods view

The Pods view provides an aggregated display of statues of all pods and resource usage within the cluster.

1. In the left navigation bar, click Pods.

2. Use the dropdown menu at the top to switch between nodes, Pods status, or use the search box for a precise search by keyword.

   

   Column	Description
   Name	The full name of the pod.
   Status	Running/Completed/Abnormal/Error.
   Node	The node where the pod is located, and its private IP.
   Pod IP	The IP address of the pod.
   Creation time	Pod creation time. Can be sorted in ascending or descending order.

Click on a specific pod name to drill into the Pod details.

Resource trends

Control Hub provides two distinct views to monitor real-time and historical resource trends, helping you quickly assess resource usage from different perspectives.

• View by user: Under the Namespaces > Resources tab, you can view the overall CPU, memory, and pod dynamics for a specific user. This allows you to quickly assess an individual user's resource footprint over a given period.

• View by cluster: Under the Pods > Resources tab, you can view the overall CPU, memory, storage, and pod dynamics from a cluster-wide perspective. This provides a clear, high-level overview of your cluster's resource trends and health.