phoenixai-kubernetes-operator

Deploy Multiple Clusters

If you have deployed a PhoenixAI cluster by YAML manifests, you can write a new PhoenixAICluster CR YAML to deploy another PhoenixAI cluster.

We have split the kube-anywhere chart into two subcharts: operator and phoenixai. Installing kube-anywhere is equivalent to installing both operator and phoenixai subcharts, and uninstalling kube-anywhere is equivalent to uninstalling both operator and phoenixai subcharts.

If you have deployed a PhoenixAI cluster by operator + phoenixai helm chart, you can deploy another PhoenixAI cluster by the phoenixai helm chart.

If you have deployed a PhoenixAI cluster by kube-anywhere helm chart, you have two ways to deploy another PhoenixAI cluster.

1. Deploy another PhoenixAI cluster by phoenixai helm chart.
2. Deploy another PhoenixAI cluster by kube-anywhere Helm chart.

This document will guide you through the process of deploying multiple PhoenixAI clusters by kube-anywhere helm chart.

Deploy another PhoenixAI cluster by kube-anywhere Helm chart

By default, the operator will watch all namespaces. If you want to deploy another PhoenixAI cluster by kube-anywhere, you should limit each operator to watch a specific namespace.

operator:
  phoenixAIOperator:
    watchNamespace: "your-namespace"

you can also add --set operator.phoenixAIOperator.watchNamespace="your-namespace" to the helm command which has higher priority.

So, the steps to deploy multiple PhoenixAI clusters by kube-anywhere are:

1. update values.yaml file of the first deployed PhoenixAI cluster to limit the operator to watch a specific namespace.
2. upgrade the first PhoenixAI cluster.
3. install the second PhoenixAI cluster by the same kube-anywhere chart, and do not forget to specify the namespace.