Bridging Infrastructure and Applications (BIA) - vSphereCompute

A very rough K8s operator and CRD scheme for creating and deleting vSphere compute types (VMs).

Setup

Config

Requires Kubernetes 1.13+

Update the value field for the following in deploy/operator.yaml to reflect your environment:

- name: GOVC_URL
  value: "https://username:password@vCenterFQDNorIP"

Import the vSphereCompute CRDs

kubectl create -f deploy/crds/bia_v1alpha1_vspherecompute_crd.yaml

Import the operator

kubectl create -f deploy/

Usage

Create a vSphereCompute object reflecting the VM(s) you want created, an example can be found at deploy/crds/bia_v1alpha1_vspherecompute_cr.yaml

Import the vSphereCompute object into K8s:

kubectl create -f deploy/crds/bia_v1alpha1_vspherecompute_cr.yaml

A VM will be spun up on the environment as specified in deploy/operator.yaml.

Get the current BIA deployed and managed VMs

$ kubectl get vc
NAME          VMNAME        STATUS      CPUS   MEMORY   IP    HOST
hello-world   hello-world   poweredOn   2      2048           host-28

Cleanup

kubectl delete -f deploy/crds/
kubectl delete -f deploy/

Known Issues

• Lots of hard coded values. Tons.
• Uses a hammer-solution of govc instead of govmomi to interact with VC via os.exec
• Doesn't use the VMware VCP settings set on the cluster - relies on it's own ENV vars
• Only CPU, Memory and VMName implemented from CR spec so far.
• Doesnt resolve VC moref-ids to human-relatable names.
• Does not update VMs with patched changes.
• Likely uses the wrong call for deleting VMs.

Future work

• Deploy nodes with Cluster API?
• FCD and SPBM policy
• Create a network and use Cluster API to bootstrap.