`juju remove-application --force --no-wait` removes storage but not its reference

[Alexandre Gomes]

When an application with storage is removed with `juju remove-application --force --no-wait`, its storage is removed but the reference remains. All commands after that failed with `ERROR getting details for storage <storage_name>: filesystem for storage instance "<storage_name>" not found`.

It can be fixed by `juju remove-storage <name> --force` though.

I've experienced this issue with the jenkins charm. Just deploy jenkins and then delete the application with the command above. After that, juju status and other commands fails with the error listed above.

[Ian Booth (wallyworld)]

--force is an option of last resort to be used in specific circumstances, and may have rough edges. It's supposed to only be used in situations where normal app removal cannot complete due to things like cloud resources being removed out of band or other similar scenarios. Is the use case here one of those scenarios? We strongly recommend not using --force as part of normal operating procedure, because things need to be removed and cleaned up in an orderly manor.

Having said that, if there's a model artefact left behind when using --force, that is something that should be fixed. Can you share what charm / cloud / storage is involved and anything else to allow the issue to be reproduced?

[Alexandre Gomes (alejdg)]

There wasn't anything different from deploying the charm and removing it. Here are the info requested:
- charm: ch:jenkins rev42
- cloud: openstack
- storage: cinder