Been experimenting with Instant Clones on Horizon 7.0.1, and have noticed I seem to have 24 cp-parent machines in a 4-node cluster. This is no problem in itself, but I only have 3 pools so why do I have so many cp-parents? I also have 9 cp-replica and 3 cp-template machines. Is there no way of being able to figure out which pool of VMs the cp-parent applies to or which is the current cp-parent, and then be able to remove the cp-parents that aren't required?
Cheers
JD
Just to confuse matters further - I've disabled the pool, killed the clones so in theory the entire cluster should have no replicas, parents or anything. Unfortunately that is not the case.
I have three pools, all disabled and with the clones removed. I have 4 nodes, each node has 6x cp-parent VMs. The cluster has a total of 24x cp-parent VMs, 9x cp-replica and 2x cp-template machines. Why? Surely if I've turned off the cloning, disabled the pools and removed the desktops there should be none of these and instead they should all be re-created once I re-enable cloning?
What's happening? Am I doomed to have my infrastructure fill up with orphaned parent VMs?
JD
Hi JD,
The number of internal VMs depend on how many pools, hosts, datastores, images you have. You mentioned 3 pools and 4 hosts. How many datastores? and how many master images do you have in each of the pools?
After you've deleted the pools, the internal VMs still hang around in a protected folder. You can unprotect the folder and remove the VMs (I believe there's a KB on how to do that, let me try to find it). We have gotten similar feedback that there should be an easier way to delete those internal VMs, and we are investigating.
Thanks for trying instant clone! Let me know if you have any other questions / feedback.
-Angela Ge, instant clone PM
Hi JD,
Pool disablement and clone deletion will not clean up these VMs, you will have to delete the pools in order for these VMs to be removed.
For each unique image that you create (and looks like you have 3), there is one cp-template VM. cp-replica is created per image per datastore so that each datastore has a base disk, so my guess is that you have 3 datastores. For a given cluster, 2 cp-parent is created (on different hosts) per datastore per image to load balance between hosts (since those are quisced vm from which vmFork clone will be created).
Regards,
Oswald