...

The document provides guidance to install the ote-stack platform on AI Edge Blueprint, which covers the information of hardware requirement, prerequisite software and the detailed step. The ote-stack platform can be installed on Kubernetes or on a system with Docker support. The IEC cluster, which is based on Kubernetes, is chosen to be the edge infrastructure for AI Edge. The installation guide of IEC R2 can be got in here. 

The guide shows step by step to deploy ote-stack platform with one edge cluster as well as add a new edge cluster to management.

License

Apache 2.0 license

How to use this document

The guide covers the details of pre-installation requirements, installation steps and uninstallation guide. The installation steps should be executed before the prerequisites and pre-installation requirements are ready. In addition, one IEC edge cluster with 3 nodes needs to be prepared in advance.

...

Software Prerequisites

• Available harbor (>=1.7, <2.0) with admin user and password
• At least one IEC edge cluster with healthy container network. 

...

When ote-stack runs, a super administrator will be created. The username and password can be modified by changing parameters `superAdminName` and `superAdminPassword` in file deployments/open-api/open-api.yml.  Note that the initial username and password of admin, only works for the first time when ote-stack starts.

Enable edge autonomy feature in your cluster

1. Deploy EdgeHub

For k8s cluster

• Modify the kubelet-bootstrap.conf file of kubelet, the server field points to the listening address of EdgeHub's edge server, the default is: https://127.0.0.1:8778.And kubelet should add startup parameters: --kube-api-content-type=application/json

• Modify the kube-proxy.conf file of kube-proxy, the server field points to the listening address of EdgeHub's edge server, the default is: https://127.0.0.1:8778. And kube-proxy adds startup parameters: --kube-api-content-type=application/json

• Copy the certificate of k8s apiserver, including three files ca.pem, kube-apiserver.pem, kube-apiserver-key.pem. Rename kube-apiserver.pem to edge-server.pem and rename kube-apiserver-key.pem to edge-server-key.pem

• Copy the kubeconfig file to the working directory, modify the server field of the kubeconfig file to point to the listening address of the EdgeHub load balancer, the default is: https://127.0.0.1:6888.

• Start EdgeHub./ote_edgehub k8s -v 2 --kube-config kubeconfig --init-server ${master_ip}:${master_port}

For k3s cluster

• Set k3s agent startup parameters

  Code Block
  --server=https://127.0.0.1:8778 --kubelet-arg=kube-api-content-type=application/json --kube-proxy-arg=kube-api-content-type=application/json

  
  

• Create the ssl directory under the working directory, and copy the k3s certificate files server-ca.crt, server-ca.key and token to the ssl directory. Rename server-ca.crt to edge-server.pem, and rename server-ca.key to edge-server-key.pem.

• Copy the kubeconfig file to the working directory, modify the server field of the kubeconfig file to point to the listening address of the EdgeHub load balancer, the default is: https://127.0.0.1:6888.

• Start EdgeHub

  Code Block
  ./ote_edgehub k3s -v 2 --kube-config kubeconfig --init-server ${master_ip}:${master_port}

  
  

2.Deploy EdgeController

• First need to create EdgeNode CRD

     kubectl apply -f edgenode-crd.yml

• Run edge-controller on a master node of the k8s or k3s cluster

     ./ote_edgecontroller -v 2 -kubeconfig /root/.kube/config

Add a new cluster to ote-stack for management (option)

...

Please refer the section "Pre-Installation Requirements" in wiki IEC Type1&2 Installation Guide for R2 and make sure the environment requirements met.

...

Please prepare three arm servers and build a edge cluster according to wiki IEC Type1&2 Installation Guide for R2.

Including Upstream Patches with Deployment

...