new unit installation fails when last unit removed

[Erik Lönroth]

I do:

$ juju deploy postgresql

I then mess up the server, by performing some configs.

$ juju config postgresql wal_e_storage_uri="file://localhost/var/backups/pg"

So I need to remove the unit.

$ juju remove-unit postgresql/0

The application is still there, so I want to install a new unit.

$ juju add-unit postgresql

..... Now, the installation fails by getting stucp in maintenance. See attached screenshot.

Consequence: I cant restore my database without removing the application and thereby also losing my configs etc.

I don't know if this is the expected and desired behaviour. Should I need to remove the application as whole in this case to be able to re-deploy a new master?