Skip to content

HPA setup and configuration

It is possible to use cpu-utilization-based HPA for API server (aka waldur-mastermind-api) and Celery executor (aka waldur-mastermind-worker and waldur-mastermind-beat) pods.

Setup

If you use minikube, you need to enable metrics-server using next command: minikube addons enable metrics-server

Configuration

In values.yaml file you can configure HPA for:

  1. API server (hpa.api prefix):

    1.1 enabled - flag for enabling HPA. Possible values: true for enabling and false for disabling.

    1.2 resources - custom resources for server. requests.cpu param is mandatory for proper HPA work.

    1.3 cpuUtilizationBorder - border percentage of average CPU utilization per pod for deployment.

  2. Celery (hpa.celery prefix):

    2.1 enabled - flag for enabling HPA, the same possible values as for API server.

    2.2 workerResources - custom resources for celery worker. requests.cpu param is mandatory for proper HPA work.

    2.3 beatResources - custom resources for celery beat. requests.cpu param is mandatory for proper HPA work.

    2.4 cpuUtilizationBorder - border percentage of average CPU utilization per pod for deployment.