Kubernetes Resource Usage

Sysdig follows the Prometheus-compatible naming convention for both metrics and labels as opposed to the previous statsd-compatible, legacy Sysdig naming convention. This page shows metrics in the legacy Sysdig naming convention. See Metrics and Label Mapping for the mapping between Sysdig legacy and Prometheus naming conventions.

Compatibility Mapping

Before using Kubernetes resource metrics, review their compatibility with Sysdig components. The newly supported Kubernetes metrics are not available to older versions of Sysdig Agent.

Note also that you must edit the agent config file, dragent.yaml, to enable these metrics. See Enable Kube State Metrics Collection with K8s_extra_resources.

Metric NameAgentPlatform
PVC metrics0.89.3 and beyondRelease 2172
Resource Quota metrics0.87.1 and beyondRelease 2172
HPA metrics0.79.0 and beyondRelease 2172

Kubernetes Resource Metrics

Metric Name

Metric Description

Metric Type

Segment By

kubernetes.persistentvolumeclaim.storage

The storage capacity requested by the persistent volume claim.

kubernetes.persistentvolumeclaim.storage provides Sysdig users with a single overarching metric for persistent volume claims (PVCs), rather than a series of metrics that often repeat/duplicate information. Each Kubernetes PVC metric is mapped to a kubernetes.persistentvolumeclaim.storage label, which can then be used to segment the overarching metric.

See Using Labels for more information on segmenting metrics.

Gauge

  • kubernetes.namespace.name

  • kubernetes.persistentvolumeclaim.label.accessmode

  • kubernetes.persistentvolumeclaim.label.app

  • kubernetes.persistentvolumeclaim.label.status.phase

  • kubernetes.persistentvolumeclaim.label.storage

  • kubernetes.persistentvolumeclaim.label.storageclassname

  • kubernetes.persistentvolumeclaim.label.volumename

kubernetes.pod.restart.count

The cumulative number of container restarts for the pod over its lifetime.

This metric is not useful for alerts. Sysdig recommends using kubernetes.pod.restart.rate instead.

Counter - Integer

Kubernetes

kubernetes.pod.restart.rate

The number of container restarts for the pod within the defined scope/time period.

Gauge - Integer

Kubernetes

kubernetes.replicaSet.replicas.desired

The number of replica pods the replicaSet is configured to maintain.

Gauge - Integer

Kubernetes

kubernetes.replicaSet.replicas.running

The current number of replica pods running in the replicaSet.

Gauge - Integer

Kubernetes

kubernetes.replicationController.replicas.desired

The number of replica pods the replicationController is configured to maintain.

Gauge - Integer

Kubernetes

kubernetes.replicationController.replicas.running

The current number of replica pods running in the replication controller.

Gauge - Integer

Kubernetes