nscale is caching repository info between systems

[AdrianRossouw]

We have a situation here where we have multiple instances of the system deployed using the same nscale master, with the only thing different between them being the repository url.

Because all of the instances share the 'server' name for the container, settings for the previously deployed system seem to bleed through to the new ones.

Example :

1. you have 2 systems alpha and beta
2. each of those systems have a server container.
3. alpha/server points to github.com/project/alpha.
4. beta/server points to github.com/project/beta
5. nscale system deploy alpha works as expected.
6. nscale system deploy beta tries to redeploy alpha.

The only workaround we have discovered is to restart the nscale process in between deploys, which limits our ability for different teams to be able to deploy alpha and beta at the same time.

The communication overhead of having to make sure nobody is deploying anything is causing us to lose productivity.