cinder volume type not being used, bug?
I have an image uploaded to openstack with a cinder_volume_type.
I am finding when using the horizon dashboard, things work great. The instance is built and the volume uses the cinder backend as configured. However when launching an instance via cli using the same image and same other values such as network etc, the instance is launched with a volume on the compute nodes disk within /var/lib/ directory.
to try and fix the problem, I set the default_volume_type in cinder.conf, as well as removed the default value of enabled_backend=tripleo_iscsi but still the same problem: CLI launched instances do not get a volume on the backend.
Additionally, when you look at the instance either in the cli or horizon, it says: | volumes_attached | |
So it's showing no volume attached as well.
How do I proceed on this?
I guess your CLI command doesn’t specify that the root disk should be on a volume. How do you launch the instance?
Not sure if the openstack CLI client allows this. For an example using the nova client, see https://ask.openstack.org/en/question....
Thanks again for your quick response :) The cli command, and subsequently our ansible script is wrong! See the answer below. :)