Previous

Next

Kueue Overview (light mode)

Kueue Overview (dark mode)

Overview

Kueue is a Kubernetes workload queueing system that manages and schedules workloads on your Kubernetes cluster. It prioritizes workloads and schedules them fairly and efficiently.

This integration collects metrics from the Kueue controller manager and API server to help you monitor the health and performance of your Kueue cluster. It also collects events for Workload lifecycle transitions, including admission, eviction, and completion.

Setup

Prerequisites

Kueue is a cluster-level service. Configure this integration as a Cluster Agent cluster check so only one Agent instance scrapes the Kueue metrics endpoint.

Installation

The Kueue check is included in the Datadog Agent package. No additional installation is required on your server.

Configuration

1. To collect optional ClusterQueue resource metrics, such as kueue.cluster_queue.resource_usage.gpu, configure Kueue with metrics.enableClusterQueueResources: true and restart the Kueue controller manager.
2. Provide a cluster check configuration to the Cluster Agent. For file or ConfigMap based configuration, set cluster_check: true in the instance:

clusterAgent:
  confd:
    kueue.yaml: |-
      cluster_check: true
      init_config:
      instances:
      - openmetrics_endpoint: http://kueue-controller-manager-metrics-service.kueue-system.svc:8080/metrics

Kueue Workload lifecycle events are collected by default. The Agent running the check needs get and list permissions on the workloads resource in the kueue.x-k8s.io API group. Set collect_workload_events: false to disable event collection. 3. Alternatively, annotate the Kueue metrics service with Autodiscovery cluster check annotations:

ad.datadoghq.com/endpoints.checks: |
  {
    "kueue": {
      "instances": [
        {
          "openmetrics_endpoint": "http://%%host%%:%%port%%/metrics"
        }
      ]
    }
  }

Log collection

The Kueue controller manager writes logs to its container output, which Kubernetes captures as container logs. Log collection is disabled by default in the Datadog Agent. To enable it, see Kubernetes log collection in the Datadog documentation. The node Agent running on the node that hosts the Kueue controller manager collects the logs, not the Cluster Agent that runs this cluster check.

After you enable log collection, set the Kueue log configuration as an Autodiscovery annotation on the controller manager’s pod template so it persists across pod restarts. Add it under spec.template.metadata.annotations of the kueue-controller-manager deployment, or set controllerManager.manager.podAnnotations if you install Kueue with the Helm chart:

ad.datadoghq.com/manager.logs: |
  [
    {
      "source": "kueue",
      "service": "<SERVICE_NAME>"
    }
  ]

This annotation targets the container named manager, which is the container name used by both the Kueue release manifests and the Helm chart. If your Kueue container uses a different name, replace manager with that name (.spec.containers[i].name).

Validation

Run the Cluster Agent’s clusterchecks subcommand and look for kueue under the Checks section.

Data Collected

Metrics

Uninstallation

To disable the integration, rename the configuration file from kueue.yaml to kueue.yaml.example. Alternatively, if you are running a containerized environment, you can remove the annotation used to enable the integration.

Support

Need help? Contact Datadog Support.