state_resourcequota

README

Kube-state-metrics/ResourceQuota

Version history

• September 2019, v1.7.0

Resources

Docs for 1.7 release of kube-state-metrics ResourceQuota can be found here: https://github.com/kubernetes/kube-state-metrics/blob/release-1.7/docs/resourcequota-metrics.md

Metrics insight

- kube_resourcequota{namespace,resourcequota,resource,type} Gauge

    Info about existing `ResourceQuota` and current status        

- kube_resourcequota_created{namespace,resourcequota} Gauge

    Creation time for `ResourceQuota`

Setup environment for manual tests

• Setup kubernetes environment for beats testing

https://github.com/elastic/beats/tree/master/metricbeat/module/kubernetes/_meta/test

• Install kube-state-metrics

As part of the referred document above, follow these instructions

https://github.com/elastic/beats/tree/master/metricbeat/module/kubernetes/_meta/test#testing-kubernetes-loads

• Create ResourceQuota objects

The manifest are found at this location, not only creates the ResourceQuota objects, but also other resources that will fail because of the existence of the quota at the namespace:

https://github.com/elastic/beats/tree/master/metricbeat/module/kubernetes/_meta/test/docs/02_objects/resourcequota.yaml

It will create

• named rqtest namespace, which will be assigned the resource quotas

• named resources resource quota, which will limit the ammount of CPU and memory that can be assigned to the namespace. (This settings won't be put to test)

• objects resource quota, which will limit the quantity of objects that can be created at this namespace:

  • 3 Pods
  • 1 Configmap
  • 0 PersistentVolumeClaims
  • 1 ReplicaController
  • 1 Secret
  • 2 Services
  • 1 Service type LoadBalancer

• It will also create regular objects at that same namespace

  • 1 Service type LoadBalancer, that will succeed
  • 1 Service type LoadBalancer, that will fail due to exceeding Quota

• Copy binary and metricbeat assets to the playground pod. The module file targeting ResourceQuota should look like this:

- module: kubernetes
  enabled: true
  metricsets:
    - state_resourcequota
  period: 10s
  hosts: ["kube-state-metrics.kube-system:8080"]
  in_cluster: true

• Execute metricbeat from the playground

You should see at elasticsearch/kibana:

Events that indicate a hard limit on services of type LoadBalancer

• dataset: kubernetes.resourcequota
• kubernetes.resourcequota.name: objects
• kubernetes.resourcequota.resource: services.loadbalancers
• kubernetes.resourcequota.quota: 1
• kubernetes.resourcequota.type: hard

Events that indicate the number of service type LoadBalancer used

• dataset: kubernetes.resourcequota
• kubernetes.resourcequota.name: objects
• kubernetes.resourcequota.resource: services.loadbalancers
• kubernetes.resourcequota.quota: 1
• kubernetes.resourcequota.type: used

Documentation

Index

• func NewResourceQuotaMetricSet(base mb.BaseMetricSet) (mb.MetricSet, error)
• type ResourceQuotaMetricSet
  • func (m *ResourceQuotaMetricSet) Fetch(reporter mb.ReporterV2)

Functions

func NewResourceQuotaMetricSet

func NewResourceQuotaMetricSet(base mb.BaseMetricSet) (mb.MetricSet, error)

NewResourceQuotaMetricSet returns a prometheus based metricset for ResourceQuotas

Types

type ResourceQuotaMetricSet

type ResourceQuotaMetricSet struct {
	mb.BaseMetricSet
	// contains filtered or unexported fields
}

ResourceQuotaMetricSet is a prometheus based MetricSet that looks for mb.ModuleDataKey prefixed fields and puts then at the module level

Copying the code from other kube state metrics, this should be improved to avoid all these ugly tricks

func (*ResourceQuotaMetricSet) Fetch

func (m *ResourceQuotaMetricSet) Fetch(reporter mb.ReporterV2)

Fetch prometheus metrics and treats those prefixed by mb.ModuleDataKey as module rooted fields at the event that gets reported