Setting up Zep in a Kubernetes Cluster

Setup a Kubernetes Cluster

Setup a Kubernetes Cluster (e.g. using Docker Desktop for local testing on your laptop).

Make sure cluster is running...

kubectl cluster-info

Create an OpenAI API Key secret

Create a Secret to store the OPENAI_API_KEY

kubectl create secret generic zep-secret --from-literal=ZEP_OPENAI_API_KEY=<your-api-key>

Deploy Zep

Run the deployment yaml

kubectl apply -f zep-deployment.yaml

Access Zep

Make sure you have setup port forwarding (default listen port in the config is localhost:8000)

kubectl port-forward service/zep 8000:8000

Point your Zep Python or JavaScript client to http://localhost:8000 to access the Zep API.

Installing on EKS, GKE, and other managed Kubernetes services

The base instructions above will bring up a local Kubernetes deployment of a Zep server and related containers, with a very limited configuration. For production or managed Kubernetes deployment, this template should be edited to include resource requests and limits, and other Zep configuration options and secrets.

For resource requirements to consider, see the production deployment guide.