Note: This demo is not a complete implementation.
To follow this demo, you need:
- A Kubernetes cluster with each node having minimum 8gb ram.
For this demo, we will use KinD tool to create cluster, although any Kubernetes cluster should work.
Click here for KinD cluster creation instructions.
Download Kind on linux:
curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.11.0/kind-linux-amd64
chmod +x ./kind
mv ./kind /bin/kind
Now create cluster:
kind create cluster --image kindest/node:v1.21.1
Note: You can try this demo in a VM on any public cloud like AWS, GCP, If you do not want to try on your system.
Install crds
First we need to install the required crds for Kupenstack.
kubectl apply -f https://raw.githubusercontent.com/Kupenstack/kupenstack/main/config/demo2/crds.yamlCreate kupenstack-config
Before deploying KupenStack we need to deploy its configurations as configmap into the cluster. The following is a sample configuration file:
kubectl apply -f https://raw.githubusercontent.com/Kupenstack/kupenstack/main/config/demo2/config.yamlapiVersion: v1
kind: ConfigMap
metadata:
name: kupenstack-config
namespace: kupenstack
data:
config.yaml: |
apiVersion: kupenstack.io/v1alpha1
kind: KupenstackConfiguration
metadata:
name: configfile
spec:
defaultProfile:
name: occp
namespace: default
# List of k8s nodes to disable for osk cluster.
nodes:
- name: kind-control-plane
type: control,computeThis is a pretty straightforward configuration. Here we have said that the default OpenstackCloudConfiguration Profile is occp in the default namespace. We will discuss more on this next. But before that please note that KupenStack automatically creates a containerized Openstack cluster for us. For creating this cluster it assumes k8s control nodes to be openstack control nodes and k8s compute nodes to be openstack compute nodes. But we can override this behaviour. For example, in this sample configuration, we have configured to use k8s node named kind-control-plane as both control and compute. We can also disable certain nodes to be not used for deploying OpenStack cluster pods.
OpenstackCloudConfiguration Profile
We can configure OpenStack cluster using OpenstackCloudConfigurationProfile(Occp) Crds. An Occp is required before creating an OpenStack cluster.
kubectl apply -f https://raw.githubusercontent.com/Kupenstack/kupenstack/main/config/demo2/occp.yamlapiVersion: cluster.kupenstack.io/v1alpha1
kind: OpenStackCloudConfigurationProfile
metadata:
name: occp
spec:
keystone:
replicas:
api: 1
glance:
replicas:
registry: 1
api: 1
horizon:
disable: true
replicas:
server: 1
nova:
replicas:
osapi: 1
conductor: 1
metadata: 1
conf:
ceph:
enabled: false
nova:
libvirt:
virt_type: qemu
cpu_mode: none
neutron:
conf:
neutron:
DEFAULT:
interface_driver: linuxbridge
dhcp_agent:
DEFAULT:
interface_driver: linuxbridge
l3_agent:
DEFAULT:
interface_driver: linuxbridge
placement:
replicas:
api: 1Installing Kupenstack
Now, as a final step, we can install KupenStack.
kubectl apply -f https://raw.githubusercontent.com/Kupenstack/kupenstack/main/config/demo2/kupenstack-controller-manager.yamlAs soon as Kupenstack starts it will first look at KupenstackConfiguration and k8s cluster to create required OpenstackNode objects. (Try kubectl get osknodes). Next, these various controllers in Kupenstack will try to spin up an Openstack cluster.
To look at the deployment while it is happening, try: watch kubectl get pods -n kupenstack. (On a machine with good compute and internet connection it should take about ~13mins)
Once deployment is complete. The output will look something like this:
glance-api-7bcdc786b6-wktzr 1/1 Running 0 41m
glance-bootstrap-hkhk7 0/1 Completed 0 31m
glance-db-init-q95l8 0/1 Completed 0 41m
glance-db-sync-6z7zj 0/1 Completed 0 37m
glance-ks-endpoints-kphtm 0/3 Completed 0 33m
glance-ks-service-dtd4q 0/1 Completed 0 34m
glance-ks-user-vnncg 0/1 Completed 0 33m
glance-metadefs-load-w99r2 0/1 Completed 0 33m
glance-rabbit-init-jh76b 0/1 Completed 0 36m
glance-storage-init-jzt4p 0/1 Completed 0 32m
horizon-56b578d677-pzhfq 1/1 Running 0 41m
horizon-db-init-fwkm2 0/1 Completed 0 41m
horizon-db-sync-wb5hb 0/1 Completed 0 41m
ingress-d5d8b9bdc-cz9lk 1/1 Running 0 41m
ingress-error-pages-6d54c96946-jbb96 1/1 Running 0 41m
keystone-api-b4f58cdd9-pc6kr 1/1 Running 0 41m
keystone-bootstrap-trqxz 0/1 Completed 0 33m
keystone-credential-setup-pkd4p 0/1 Completed 0 41m
keystone-db-init-t7jxd 0/1 Completed 0 38m
keystone-db-sync-6w448 0/1 Completed 0 37m
keystone-domain-manage-qgh2c 0/1 Completed 0 34m
keystone-fernet-setup-c9m2d 0/1 Completed 0 37m
keystone-rabbit-init-94l7d 0/1 Completed 0 36m
kupenstack-controller-678c66b58c-zrhwx 1/1 Running 0 42m
libvirt-libvirt-default-xpfbv 1/1 Running 0 41m
mariadb-ingress-56bbb59f78-dqs8b 1/1 Running 0 41m
mariadb-ingress-error-pages-77c55c9cd7-c2klm 1/1 Running 0 41m
mariadb-server-0 1/1 Running 0 41m
memcached-memcached-9978cc66c-gvnkd 1/1 Running 0 41m
neutron-db-init-bj986 0/1 Completed 0 41m
neutron-db-sync-ks6v2 0/1 Completed 0 37m
neutron-dhcp-agent-default-mpsjz 1/1 Running 0 41m
neutron-ks-endpoints-bkq2q 0/3 Completed 0 33m
neutron-ks-service-8rs4t 0/1 Completed 0 34m
neutron-ks-user-4v4nw 0/1 Completed 0 33m
neutron-l3-agent-default-f27kz 1/1 Running 0 41m
neutron-lb-agent-default-7pcj7 1/1 Running 0 41m
neutron-metadata-agent-default-bd68j 1/1 Running 0 41m
neutron-netns-cleanup-cron-default-8gw4j 1/1 Running 0 41m
neutron-rabbit-init-fpln2 0/1 Completed 0 37m
neutron-server-b56796d79-vhcl9 1/1 Running 0 41m
nova-api-metadata-566785c56c-hdh6s 1/1 Running 2 41m
nova-api-osapi-66466bd88-b9b9p 1/1 Running 0 41m
nova-bootstrap-xgfj2 1/1 Running 0 41m
nova-cell-setup-9xthp 1/1 Running 0 41m
nova-compute-default-fcz4z 1/1 Running 0 41m
nova-conductor-56bf96fb47-bjnpr 1/1 Running 0 41m
nova-consoleauth-7989ff966b-pqmjr 1/1 Running 0 41m
nova-db-init-gm6jk 0/3 Completed 0 41m
nova-db-sync-lp2lp 0/1 Completed 0 37m
nova-ks-endpoints-bf2cp 0/3 Completed 0 33m
nova-ks-service-df9cl 0/1 Completed 0 34m
nova-ks-user-rv9wn 0/1 Completed 0 33m
nova-novncproxy-5894489d56-grm7w 1/1 Running 0 41m
nova-rabbit-init-4xtp9 0/1 Completed 0 34m
nova-scheduler-85d7f5d5bc-p2xsr 1/1 Running 0 41m
placement-api-694dccf9f-hvl7x 1/1 Running 0 41m
placement-db-init-bf2h8 0/1 Completed 0 41m
placement-db-sync-cm5jh 0/1 Completed 0 41m
placement-ks-endpoints-tp2lb 0/3 Completed 0 34m
placement-ks-service-dd8xp 0/1 Completed 0 37m
placement-ks-user-hlswb 0/1 Completed 0 35m
rabbitmq-cluster-wait-7k5h5 0/1 Completed 0 41m
rabbitmq-rabbitmq-0 1/1 Running 0 41m
Next we are ready to use Openstack in kubernetes:
Let us create a Virtual Machine using KupenStack.
Apply following yaml file to Kubernetes:
apiVersion: kupenstack.io/v1alpha1
kind: Image
metadata:
name: image-sample
spec:
src: http://download.cirros-cloud.net/0.5.1/cirros-0.5.1-x86_64-disk.img
format: raw
---
apiVersion: kupenstack.io/v1alpha1
kind: Flavor
metadata:
name: flavor-sample
spec:
vcpu: 2
ram: 500
disk: 1
rxtx: "1.0"
---
apiVersion: kupenstack.io/v1alpha1
kind: VirtualMachine
metadata:
name: virtualmachine-sample
spec:
image: image-sample
flavor: flavor-sampleApply:
kubectl apply -f https://raw.githubusercontent.com/Kupenstack/kupenstack/main/config/demo/vm-sample.yamlNow, let us explore what we did.
kubectl get virtualmachinesthis will show
NAME NODE STATE AGE
virtualmachine-sample kind-control-plane BUILD 12s
Our Virtual machine is in build state, let us wait for some time and check again.
kubectl get vm -o wide
NAME NODE STATE NETWORKS(IP) AGE
virtualmachine-sample kind-control-plane Running default(10.10.1.190) 62s
Now, our Virtual machine is in running state. We can also see that VM is in default network and has IP 10.10.1.190 . Remember we have not deployed any networks. KupenStack by default keeps all VM in default network if network is not specified in VM Manifest. Since, we had not created any network called default KupenStack automatically created it for us. We can check it by:
kubectl get networks
NAME IN-USE AGE
default true 56s
kubectl get flavorsgives us following output:
NAME IN-USE AGEflavor-sample true 38s
kubectl get imagesgive us output:
NAME IN-USE READY AGEimage-sample true true 43s
Note: our Image is in Ready state. It means it has been downloaded and uploaded to glance successfully.
kubectl get keypairs -o wide
NAME IN-USE AGE PRIVATE-KEYkeypair-sample true 50s keypair-sample-542zr
Note: we didn't specified any public key while creating key-pair so KupenStack automatically generated one and is exposing it through a secret. We can copy our private-key from this secret and then delete this secret.
- If we look at our Namespaces, then we can see there are annotations for corresponding Project in OpenStack. For example:
kubectl describe ns defaultgive us:
Name: defaultLabels: <none>Annotations: kupenstack.io/external-project-id: 4c7c184c14314cb08244a1fcc47d0bf5 kupenstack.io/external-project-name: default-5fg8kStatus: ActiveNo resource quota.No LimitRange resource.
In the annotations we can see a Project named default-5fg8k has been created in OpenStack which maps to this Namespace in Kubernetes. Every resource that is created in this Namespace will be created in corresponding Project at OpenStack.
We can cross-verify our above deployments in Horizon also, Go to port 32020 on the node's ip in browser:
Now, you can login and check, (Credentials are: Domain: "default", user: "admin", password: "password")
