Comment 5 for bug 1650651

Version:

fengxia@local-charmdev:~/workspace/wss$ juju --version
2.0.2-xenial-amd64

Steps I took:

1. create a vanilla `trusty` LXD container named `gold`
2. ssh into `gold` container, `apt install python python-dev...`
3. in terminal, `$ lxc snapshot gold`, this created a snapshot, can be viewed from `$ lxc info gold`.
4. in terminal, `$ lxc publish gold/snap0 --alias gold-image` (according to https://insights.ubuntu.com/2016/04/01/lxd-2-0-image-management-512/). This create a new image `gold-image` based on container snapshot (created in #3).
5. `$ lxc image alias create juju/trusty/amd64`, based on recommendation on this thread. Verify using `$ lxc image info gold-image`. Similarly, I added alias `ubuntu-trusty`, `trusty`.
6. `$ lxc image delete ubuntu-trusty`. This is to delete the default ubuntu-trusty image downloaded by LXD.

So at this point, there should be one image (`$ lxc image list`) that has alias of `trusty` related name/alias.

Expect:

1. When `$ juju deploy [charm] --series trusty`, it will use `gold-image` as `trusty` image to create new machine.

What happened:

1. it will download `ubuntu-trusty` image (`$ lxc image list` will show this new image after a few minutes of downloading), then use this image to create a new machine.