When rebuilding the live workspace of a site that has no validated content, the site node ends up without a content root.
Calling site.getContents() after that throws an exception.