Add retry logic to database calls

[mrheinen]

Right now a failure is a failure. Add a retry mechanism to make the database interaction more robust. As a rule of thumb the backend should be robust enough to continue without errors if a database restart happens.