search string side menu

ELMA365 On-Premises > ELMA365 On-Premises Enterprise > Install add-on components for ELMA365 / Install Keda

Install Keda

KEDA is an event-driven autoscaler that complements the functionality of ELMA365 applications with advanced scaling capabilities. With KEDA, you can manage the scaling of any container in Kubernetes based on the number of events that need to be processed.

The installation consists of three steps:

  1. Download the Helm chart and configuration file.
  2. Fill out the configuration file.
  3. Install the KEDA chart using Helm in the Kubernetes cluster.

Step 1: Download the Helm chart and configuration file

To install via the internet, obtain the configuration file values-keda.yaml by executing the command:

helm repo add elma365 https://charts.elma365.tech
helm repo update
helm show values elma365/keda > values-keda.yaml

Getting the configuration file for installation in a closed network without internet access

  1. On a computer with internet access, download the ELMA365 images and upload them to the local image registry by executing the following command:

helm repo add elma365 https://charts.elma365.tech
helm repo update
helm pull elma365/keda

For more details, see Download ELMA365 images.

  1. Copy the downloaded chart archive keda-X.Y.Z.tgz to the server where the installation will be carried out.
  1. Unpack this chart and copy the default configuration file values.yaml to values-keda.yaml:

tar -xf keda-X.Y.Z.tgz
cp keda/values.yaml values-keda.yaml

Step 2: Fill out the configuration file

Fill out the configuration file values-keda.yaml for the KEDA service installation.

For the parameter watchNamespace, specify namespace in which the ELMA365 application is installed. The default namespace is ELMA365. For multiple instances of ELMA365, specify all required namespaces, appending to the array watchNamespace.

If high availability is required, set the number of replicas to 3 in parameters operator.replicaCount and metricsServer.replicaCount.

## keda settings
keda:
  ## number of replicas for high availability
  operator:
    replicaCount: 1
  metricsServer:
    replicaCount: 1
  ## namespace that KEDA should monitor
  watchNamespace:
    - elma365
  ## install crds
  crds:
    install: true
...

Filling out the connection parameters to a private registry for installation in a closed loop without internet access

To connect to a private registry, you need to:

  1. Download ELMA365 images and upload them to the local image registry. For more details, see Download ELMA365 images.
  2. Set the address and path in the parameters image.keda.repository and image.metricsApiServer.repository.
  1. Specify the name of the secret with access rights to the private registry in the parameter imagePullSecrets. The secret must be created manually and encrypted in Base64.

## keda settings
keda:
...
  ## Connection parameters to a private registry
  image:
    ## address and path for the private registry
    keda:
      repository: registry.example.com/kedacore/keda
    metricsApiServer:    
      repository: registry.example.com/kedacore/keda-metrics-apiserver
  ## The secret with access rights to the private registry must be created manually, encrypted in Base64
  imagePullSecrets:
    - name: myRegistryKeySecretName

Where the repository format is:

  • address is registry.example.com;
  • path is /kedacore/keda (for metricsApiServer — /kedacore/keda-metrics-apiserver).

Step 3: Install the KEDA chart using Helm in the Kubernetes cluster

Run the installation of the KEDA chart in a separate namespace, for example, keda. The namespace will be created during installation if it was not created earlier.

For online installation:

helm upgrade --install keda elma365/keda -f values-keda.yaml -n keda --create-namespace 

For offline installation without internet access, navigate to the directory with the downloaded chart and execute the command:

helm upgrade --install keda ./keda -f values-keda.yaml -n keda --create-namespace

Начало внимание

The installation of the KEDA add-on component does not automatically include support for advanced autoscaling on the ELMA365 application side.

After installation, do not forget to change the ELMA365 application settings and configure advanced autoscaling on the ELMA365 application side.

Read more about enabling advanced scaling of services on the ELMA365 application side in Enable service autoscaling in ELMA365 Enterprise.

Конец внимание

Delete the KEDA chart using Helm in the Kubernetes cluster

Начало внимание

Before removing the KEDA add-on component, reconfigure the ELMA365 application to use the HPA autoscaling tool.

Конец внимание

To delete the KEDA chart in namespace keda, execute the command:

helm uninstall keda -n keda

install-descheduler.html install-kyverno.html

Found a typo? Highlight the text, press ctrl + enter and notify us