Skip to content

Latest commit

 

History

History
189 lines (158 loc) · 5.56 KB

USAGE.md

File metadata and controls

189 lines (158 loc) · 5.56 KB
Raw

Usage

This will show the basic usage of the Custom Pod Autoscaler Operator, for more indepth examples check out the Custom Pod Autoscaler repo.

Simple Custom Pod Autoscaler

apiVersion: custompodautoscaler.com/v1
kind: CustomPodAutoscaler
metadata:
  name: python-custom-autoscaler
spec:
  template:
    spec:
      containers:
      - name: python-custom-autoscaler
        image: python-custom-autoscaler:latest
        imagePullPolicy: Always
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hello-kubernetes
  config:
    - name: interval
      value: "10000"

This is a simple Custom Pod Autoscaler, using an image called python-custom-autoscaler:latest.

It provides the configuration option interval with a value of 10000 as an environment variable injected into the container.

The target of this CPA is defined by scaleTargetRef - it targets a Deployment called hello-kubernetes.

For more indepth examples check out the Custom Pod Autoscaler repo.

Using Custom Resources

apiVersion: v1
kind: ServiceAccount
metadata:
  name: python-custom-autoscaler
  annotations:
    myCustomAnnotation: test
---
apiVersion: custompodautoscaler.com/v1
kind: CustomPodAutoscaler
metadata:
  name: python-custom-autoscaler
spec:
  template:
    spec:
      containers:
      - name: python-custom-autoscaler
        image: python-custom-autoscaler:latest
        imagePullPolicy: Always
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hello-kubernetes
  provisionServiceAccount: false
  config:
    - name: interval
      value: "10000"

This is a Custom Pod Autoscaler that is similar to the basic one defined above, except that it uses a custom ServiceAccount, with the annotation myCustomAnnotation.

Take note of the option inside the CPA provisionServiceAccount: false, which informs the CPAO that the user will be providing their own ServiceAccount, so it should not override it with its own provisioned ServiceAccount.

This custom resource provision is supported for all resources the CPAO manages:

  • provisionRole - determines if a Role should be provisioned.
  • provisionRoleBinding - determines if a RoleBinding should be provisioned.
  • provisionServiceAccount - determines if a ServiceAccount should be provisioned.
  • provisionPod - determines if a Pod should be provisioned.

Automatically Provisioning a Role with Access to the Kubernetes Metrics Server

Note: this feature is only available in Custom Pod Autoscaler Operator v1.1.0 and above

apiVersion: custompodautoscaler.com/v1
kind: CustomPodAutoscaler
metadata:
  name: python-custom-autoscaler
spec:
  template:
    spec:
      containers:
      - name: python-custom-autoscaler
        image: python-custom-autoscaler:latest
        imagePullPolicy: Always
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hello-kubernetes
  roleRequiresMetricsServer: true
  config:
    - name: interval
      value: "10000"

This is a Custom Pod Autoscaler that is similar to the ones defined above, except it provisions a role with access to the Kubernetes metrics server.

Take note of the option inside the CPA roleRequiresMetricsServer: true which informs the CPAO that the CPA requires access to the metrics server, so the role that is provisioned should include these accesses.

Automatically Provisioning a Role that Supports Argo Rollouts

Note: this feature is only available in Custom Pod Autoscaler Operator v1.2.0 and above

apiVersion: custompodautoscaler.com/v1
kind: CustomPodAutoscaler
metadata:
  name: python-custom-autoscaler
spec:
  template:
    spec:
      containers:
      - name: python-custom-autoscaler
        image: python-custom-autoscaler:latest
        imagePullPolicy: Always
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hello-kubernetes
  roleRequiresArgoRollouts: true
  config:
    - name: interval
      value: "10000"

This is a Custom Pod Autoscaler that is similar to the ones defined above, except it provisions a role with access to the ability to manage Argo Rollouts.

Take not of the option inside the CPA roleRequiresArgoRollouts: true which informs the CPAO that the CPA requires the ability to manage Argo Rollouts, so the role that is provisioned should include these accesses.

Pausing autoscaling

Note: this feature is only available in Custom Pod Autoscaler Operator v1.4.0 and above

If you want to disable an autoscaler from autoscaling (e.g. during maintenance) you can do so by using the v1.custompodautoscaler.com/paused-replicas annotation on the Custom Pod Autoscaler.

When this annotation is supplied the autoscaler pod will be deleted, and the resource will be set to whatever value is set in the annotation.

For example:

apiVersion: custompodautoscaler.com/v1
kind: CustomPodAutoscaler
metadata:
  name: python-custom-autoscaler
  annotations:
    "v1.custompodautoscaler.com/paused-replicas": "42"
spec:
  template:
    spec:
      containers:
      - name: python-custom-autoscaler
        image: python-custom-autoscaler:latest
        imagePullPolicy: Always
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hello-kubernetes
  config:
    - name: interval
      value: "10000"

This autoscaler will be paused, with the replica count for the resource being managed set to 42.

If you want to re-enable the autoscaler after, just remove the annotation.