I work with JUJU and vSphere to deploy machines, applications and so via JUJU; in version 2.6.6, when I deployed new machines or bootstrap controllers the process is as follows:
1.- JUJU copies a VMDK file to the designated datastore, to directory juju-vmdks. This VMDK is the base for deploying new machines
2.- JUJU creates a temporary VM and then clone the machine with the designated VMDK, expand the HD and power on the machine
3.- Then the agent is installed and the machine is prepared to deploy applications on it.
JUJU has updated to version 2.6.8, and the process now is different, but with errors or strange things:
1.- JUJU creates a VM template instead of copying only the VMDK
2,. JUJU clones the template to deploy a machine, but the network adapters are duplicated on the newly created machine, I mean, if the template has 1 network adapter the new machine has 2 (connected to the same networks as template); if the base template has 2 network adapters (one connected to Network01 and the other to Network02) the new machines is created with 4 network adapters (2 for Network01 and 2 for Network02).
This produces errors on communications and waste of IPs from the DHCP.
Is this now the normal behavior or is it a bug? I had to downgrade the JUJU client to version 2.6.6 again in order to keep working correctly with my models.