Generic error messages in Solum when trove is down

[Dimitry Ushakov]

Whenever Trove is non responsive, Solum is unusable and throws back HTTP 500 errors for most requests. We should have better error representation to our end users.

As a complement to this issue, we also have a single point of failure that's outside of our control (Trove) for something as simple as data storage. We should have a backup/redundant storage that would allow us to service requests even if downstream services are temporarily unavailable.

[Adrian Otto (aotto)]

Dimitry, are you referring to the Trove used for storing the Solum database, or Trove used as part of a Heat orchestration?

[Devdatta Kulkarni (devdatta-kulkarni)]

Dimitry is referring to Solum's control plane database.