I got caught out last week by something that I had not come across before. A customer has an environment comprised of vSphere, vRA, vRO.. the usual suspects. vRA and vRO and associated IAAS roles were all duplicated and sat behind a load balancer.

Anyway, whilst re-writing some vRO workflows and adjusting some blueprints for them, I noticed that randomly when I destroyed test VMs through vRA they weren’t always deleted from the vSphere inventory. Randomly, VMs would be deleted or would be moved to a folder in the inventory and have the current date and time stamp appended as a suffix to the VM name. I have to say that this baffled me.

Anyway, after a bit of digging about I found the following in the Configure the vSphere Agent section of the vRA Install and Configure guide  [editor: Paul does read all the product guides, really he does!]

At this point my forehead met with the desk and we become one.

With a bit of further digging and it quickly became obvious what was going on. As mentioned earlier, the customer has multiple windows boxes with the IAAS/Web/Dem roles seperated across them. They had this setup duplicated and then sat behind their load balancer. The setting doDeletes had been configured on just one of the IAAS role boxes. So depending on which box serviced the request, the VM either got deleted or got moved.

In the end it was a simple fix, using the information above, but definately a short term head scratcher!