Hi,
I am trying to bootstrap a controller on a vmware-vsphere environment.
I have followed this guide : https://juju.is/docs/olm/vmware-vsphere, and I am stuck in the section "Create a controller".
For some reasons, I had to stop the creation of the controller at my first try. The command I run was "juju bootstrap vsp-cloud vsp-controller"
I then deleted it using the command "juju kill-controller vsp-controller", because "juju destroy-controller vsp-controller" was returning an error.
When I tried to create a controller again with some added paramaters (namely: "--to zone=[...]", and "--config datastore=[...]"), the task get stuck. With the "--debug" option, here are the latest log lines:
15:35:29 DEBUG juju.environs imagemetadata.go:124 obtained image datasource "default ubuntu cloud images"
15:35:30 DEBUG juju.provider.vmware createvm.go:441 Creating import spec: pool="ResourcePool:resgroup-124290", datastore="Datastore:datastore-168 @ /DC-dco/datastore/LABESX_dco", entity="juju-template-bef2d86bdd8c1f9f3714e8f442313c02485dcd980f16d918e0111790205b6dba"
15:35:30 DEBUG juju.provider.vmware createvm.go:444 Fetching OVF manager
15:35:30 DEBUG juju.provider.vmware createvm.go:459 CreateImportSpec succeeded
15:35:30 DEBUG juju.provider.vmware createvm.go:199 creating template VM in folder Folder:group-v140908"
On the Vcenter console, I see an "Create folder" task in error, with Status: " The name 'Juju Controller (f4a25761-061b-41df-8fd3-6105c417e414)' already exists." See screenshot attached (some names are redacted).
Any idea how to fix this issue? Is this a known bug from juju?
Thanks.
On the surface of it, it's surprising that a new controller has a UUID clash like that. Anyway, if you want to start fresh, try deleting those juju controller folders from vsphere and also check "juju controllers" to make sure there's nothing left behind there also. If there is, run "juju unregister" to get rid of it. Only do this for old controllers that did not start, don't do it for running controllers you want to keep.