Released is awsome, but tests are better

Just before lunch, we moved the last item from our Stable list to Released list. In other words, we released! Nice and cozy. We all went for our lunch. Well back, the spiders are not running properly. The scrum master says it's because we're releasing an empty database. And I said, 'Hey, isn't that the problem we had last sprint. And the sprint before. And when I come to think about it: the sprint before.'
'No', says the scrum master. It's another problem but it manifests itself in the same manor. So, how come we don't have an automated test for this? The errors might differ between sprints but the test must always work.