You are viewing an old version of this page. View the current version.

Compare with Current View Page History

Version 1 Next »

This document guides the cloud users how to create a kubernetes cluster and deploy a simple application.


Step 1. Create a Kubernetes (k8s) cluster

  • Note that a separate cluster template for k8s needs to be created for each project.
    • Use "Calico" as a network driver. "flannel" seem to have a issue.
    • It is handy to have a ingress controller. Add a label of 'ingress_controller="octavia"' to enable the octavia ingress controller.  Other type of ingress controllers are not tested yet.
  • From the Container Infra section, use the template to create the k8s cluster of the size you want.
    • A single master is only configured and tested for now.
    • You can add additional worker nodes later (you can delete some of them), but I don't think you can change the flavor of the nodes.
    • creating a k8s cluster may take tens of minutes and it may fail. Just delete the failed one and recreate one.
    • The ssh key-pair is usually not important. Just use the default.
    • Use the existing network and keep cluster api private.
    • Note that a cluster will be also be created under the Cluster Infra → KASI cluster  section.
Fig. 1-1 General setup tab for creating k8s from templateFig. 1-2 "size" tab

Fig. 1-3 "Network" tabFig. 1-4 "Management" tabFig. 1-5 "Advanced" tab

Step 2. Setup connection to the created k8s cluster

  • It is best to work from the gateway node of your project, i.e., we assume that you have a direct network connection to each k8s nodes.
  • It seems that it is best to use the openstack cli to generate kube config files.
    • To use the openstack cli commands, you need to setup the envirments. You can download the rc file from the openstack dashboard (API access → Download OpenStack RC file → OpenStack RC file )
    • Download the rc file to the gateway node.
  • Make sure you have "kubectl" command installed on the gateway : https://kubernetes.io/ko/docs/tasks/tools/install-kubectl-linux/


Setup the k8s connection
# the name of the rc file will reflect your project name. In this case, the project name is 'spherex'
> source spherex-openrc.sh  # this will ask for a password. Use the same password that yo use for the dashboard.
# The rc file need to be loaded before you use openstack cli tools.
Installing openstack cli tools
# somehow, openstack cli packages from ubuntu does not work. Instead, we install them via pip command.
> sudo apt-get install python-dev python3-pip # install pip
> sudo pip install python-openstackclient python-magnumclient
# now it's time to fetch the kube config for your cluster
> openstack coe cluster config YOUR-K8S-CLUSTER-NAME
# The above command will create a file named "config" under the current directory. This is basically a kube config file that you can use with kubectl command.
# You may environment variable of "KUBECONFIG" to this file, as suggested by the above command. You may instead copy this file under "~/.kube"


Now your "kubectl" command is connected to your newly created cluster.


kubectl get nodes
> kubectl get nodes
NAME                                          STATUS   ROLES    AGE    VERSION
spherex-k8s-calico100-4giy6vd2vahl-master-0   Ready    master   7d2h   v1.18.2
spherex-k8s-calico100-4giy6vd2vahl-node-0     Ready    <none>   7d2h   v1.18.2
spherex-k8s-calico100-4giy6vd2vahl-node-1     Ready    <none>   7d2h   v1.18.2
spherex-k8s-calico100-4giy6vd2vahl-node-2     Ready    <none>   7d2h   v1.18.2

> kubectl get pods -n kube-system
NAME                                         READY   STATUS    RESTARTS   AGE
calico-kube-controllers-795c4545c7-rl56t     1/1     Running   0          7d2h
calico-node-4rs5j                            1/1     Running   0          7d2h
calico-node-7bj8r                            1/1     Running   0          7d2h
calico-node-8slht                            1/1     Running   0          7d2h
calico-node-rxg5s                            1/1     Running   0          7d2h
coredns-5f98bf4db7-l6cd7                     1/1     Running   0          21h
coredns-5f98bf4db7-vzhn8                     1/1     Running   0          21h
dashboard-metrics-scraper-6b4884c9d5-p87b5   1/1     Running   0          7d2h
k8s-keystone-auth-n2dkh                      1/1     Running   0          7d2h
kube-dns-autoscaler-75859754fd-fd99t         1/1     Running   0          7d2h
kubernetes-dashboard-c98496485-wl4r4         1/1     Running   0          7d2h
npd-4bw99                                    1/1     Running   0          7d2h
npd-5sg2c                                    1/1     Running   0          7d2h
npd-cg6pc                                    1/1     Running   0          7d2h
octavia-ingress-controller-0                 1/1     Running   0          7d2h
openstack-cloud-controller-manager-796tr     1/1     Running   0          7d2h


Step 3. Setup Storage class

  • Openstack Container Platform is capable of provisioning persistent volumes (PVs) using the Container Storage Interface (CSI) driver for OpenStack Cinder. To use this, you need to create a StorageClass object.


sc-cinder-yaml
apiVersion: storage.k8s.io/v1beta1
kind: StorageClass
metadata:
  name: standard
  annotations:
    storageclass.beta.kubernetes.io/is-default-class: "true"
  labels:
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: EnsureExists
provisioner: kubernetes.io/cinder
  • Once you have a file with above contents (the file is named as 'sc-cinder.yaml' in this example)
> kubectl apply -f sc0cinder.yaml
> kubectl get storageclass
NAME                 PROVISIONER            RECLAIMPOLICY   VOLUMEBINDINGMODE   ALLOWVOLUMEEXPANSION   AGE
standard (default)   kubernetes.io/cinder   Delete          Immediate           false                  7d1h


Step 4. Deploy your app

  • simple example with Helm 3

  • ste-by-step example 


  • No labels