Installing The vSphere OpenStack Virtual Appliance (VOVA) On Your Laptop, Part 2

(Diagram by Rawlinson Rivera from VMware)

Introduction

In part 1 of this short series, I walked though setting up a vSphere environment, including ESXi and vCenter, using VMware Fusion. In part 2, I will walk through deploying VOVA and using vSphere with OpenStack. Before proceeding, I encourage the reader to review the official VOVA manual if you haven’t already done so.

Preparing For VOVA

Log on to the VCSA from your browser, using the static IP address you set earlier and going through port 9443. Log on with username: root and password: vmware.
Create a new virtual datacenter (I created one called “dc”) and underneath the datacenter, create a new cluster (I created one called “openstack”); connect to your ESXi host and assign it to your new vdc and cluster.

It’s important that VOVA is in time sync with VCSA. If you aren’t able to sync both either to a NTP server, you can get around it by finding the date and time of the VCSA and inputting the same for the ESXi server that will be hosting VOVA.

As per the VOVA manual, we need to create a port group or port groups called “br100” for every ESXi host that will part of our OpenStack environment. This can be done per host when using vSphere Standard Switches or by creating at single br100 port group when using the vSphere Distributed Switch. The instructions are in the VOVA manual and I also have a series of screenshots below that walk though setting up the br100 port group on a vSphere Standard Switch using vmnic1.

Next, download the VOVA_Havana OVA to your laptop or workstation.

Deploying VOVA

From the vSphere Web Client, deploy a new OVF template to your cluster and choose VOVA_HAVANA.ova (typically, you would deploy VOVA on a vSphere cluster separate from the one it is managing but here we are trying to minimize resource usage on your laptop); confirm that you want to allow VMware Client Integration access. If you have not previously used the VMware Client Integration client, you will have to install the plugin on your laptop.

Walk though the OVF Template settings and when you get to the “Select storage” section, select “Thin Provision” for your virtual disk format. This will allow you to move forward with the deployment.

When you get to the “Customize template” section, fill out the fields using the information relevant to your setup. The following screenshots provide an example of what settings I used for my setup.

If you didn’t choose to automatically power up VOVA after finishing setup, do so now and you are ready to go.

Using OpenStack To Launch vSphere Hosted Instances

log on to the OpenStack Horizon dashboard using the static IP address you set for VOVA and using user name: demo and password: vmware. Note that the VOVA dashboard uses a VMware skin for the portal but is still the Horizon dashboard.

From the dashboard, launch a new instance using the Debian image that was preloaded on to VOVA.

After a few minutes, you will have a running instance in your OpenStack environment; from the vSphere client, you should also see a new vm spun up on your ESXi host.

Click on the instance in the OpenStack dashboard to see the instance detail; you should see that the instance ID matches the name of the associated vm as shown in the vSphere client (You may have to refresh the vSphere Web Client).

Creating VMDK Backed OpenStack Cinder Volumes

In a previous post, I detailed how vSphere has contributed code to the OpenStack Cinder project to enable the use of VMDK files to function as the backing storage for an OpenStack block volume. Let’s walk through how this looks using VOVA.

Go ahead and create a new Cinder volume using the OpenStack dashboard.

Once the volume is created, go to “Manage Volume attachments” and attach it to the instance you created.

At this point, you can see your new volume attached to your new instance in the OpenStack dashboard; from the vSphere client, you will see that a new vm, with a server name starting with “volume-…” was created that does not show up on your OpenStack dashboard.

If you recall from my post on vSphere Integration With Cinder, a “shadow” vm is created whenever a VMDK backed Cinder volume is created in OpenStack; this “shadow” vm does not show up in the OpenStack dashboard and cannot be directly managed from that dashboard.

Now detach and delete the Cinder volume using the OpenStack dashboard; you will see that the “shadow” vm is deleted along with the Cinder volume.

Finally terminate your instance using the OpenStack dashboard; you will see in the vSphere Client that the vm has been removed.

At this point, if you have enough free resources on your laptop or workstation, go ahead and install new ESXi host vm in Fusion and add it to your cluster. Assuming you have Distribute Resource Scheduler (DRS) enabled and fully automated, you should see all new instances be balanced evenly across both ESXi hosts.

I hope this little exercise will help any VMware administrators interested in OpenStack and its integration with vSphere get started. As always, feedback and corrections are welcomed.