Kubernetes with Helm

The presented example uses Helm charts, if you haven’t used the tool before, we highly recommend getting familiar with the Helm quick-start docs.

The repository provides Helm charts for two setups: CKEditor AI On-Premises standalone, with its own MySQL database and Redis, and CKEditor AI On-Premises together with Collaboration Server On-Premises, where both services share one MySQL database and Redis.

Helm v3 and Kubernetes 1.19+ are required to run examples from the section.

To fetch a Git repository with a pre-configured Helm chart, git needs to be installed.

To fetch required docker images you will need to authorize to the CKEditor AI On-Premises Docker Registry.
Please prepare the download token from the Customer Portal.

1. Clone the cksource/ckeditor-cs-on-premises-infrastructure Git repository.

git clone git@github.com:cksource/ckeditor-cs-on-premises-infrastructure.git

2. Go to the kubernetes/helm directory.

cd ckeditor-cs-on-premises-infrastructure/kubernetes/helm

3. Create imagePullSecret for pulling images from the CKEditor container registry, replace the placeholder with the download token.

kubectl create secret docker-registry docker-cke-cs-com \
	--docker-username "ai-service" \
	--docker-server "https://docker.cke-cs.com" \
	--docker-password="<PASSWORD>"

The standalone scenario runs CKEditor AI On-Premises with a MySQL and single instance of Redis.

Create the dev.values.yaml file in the ai-service-development-stack directory. Replace the placeholders with your CKEditor AI On-Premises license key, which can be found in the Customer Portal, and with your LLM provider and API key. See LLM providers for available providers.

ai-service:
  secret:
	data:
	  LICENSE_KEY: '<LICENSE_KEY>'
	  PROVIDERS: '{"openai":{"type":"openai","apiKeys":["<API_KEY>"]}}'

Install the chart:

cd ai-service-development-stack &&
helm dependency update &&
helm install ai-service . --values dev.values.yaml

then expose the CKEditor AI On-Premises port

kubectl port-forward service/ai-service 8000:8080

This scenario runs CKEditor AI On-Premises and Collaboration Server On-Premises on a shared MySQL and single instance of Redis. See Collaboration Server integration for details.

Create the dev.values.yaml file in the cs-with-ai directory. Replace the placeholders with your Collaboration Server On-Premises and CKEditor AI On-Premises license keys, and with your LLM provider API key.

sharedDatabase: &sharedDatabase
  DATABASE_HOST: cs-with-ai-mysql
  DATABASE_USER: root
  DATABASE_PASSWORD: root-password
sharedRedis: &sharedRedis
  REDIS_HOST: cs-with-ai-redis-master

ckeditor-cs:
  server:
	secret:
	  data:
		<<: [*sharedDatabase, *sharedRedis]
		ENVIRONMENTS_MANAGEMENT_SECRET_KEY: secret
		LICENSE_KEY: '<COLLABORATION_SERVER_LICENSE_KEY>'

ai-service:
  secret:
	data:
	  <<: [*sharedDatabase, *sharedRedis]
	  ENVIRONMENTS_MANAGEMENT_SECRET_KEY: secret
	  LICENSE_KEY: '<AI_SERVICE_LICENSE_KEY>'
	  PROVIDERS: '{"openai":{"type":"openai","apiKeys":["<API_KEY>"]}}'

mysql:
  enabled: true
redis:
  enabled: true

Install the chart:

cd cs-with-ai &&
helm dependency update &&
helm install cs-with-ai . --values dev.values.yaml

then expose the Collaboration Server On-Premises and CKEditor AI On-Premises ports

kubectl port-forward service/cs-with-ai-ckeditor-cs-server 8000 &
kubectl port-forward service/cs-with-ai-ai-service 8001:8080

Note: This is not a production-ready solution. To expose the application to the public in a production environment use a proper proxy configuration.

The command below will stop and remove all pods which were started during the tutorial.

helm delete ai-service

For the scenario with Collaboration Server, use helm delete cs-with-ai.

You can adjust the example configuration to your needs by modifying the values.yaml file in the .../kubernetes/helm/ai-service/ directory.