Could have each failed job write to a file in some temp directory and have waitjobs read from there at the end. Would really help in not having to scroll through all the logs.