Due to the way recipe status updates are now handled (to avoid race conditions on the server), the reservesys task may now fail to notice that the immediately preceding task actually failed. This can't be fixed directly, but increase the importance of adding the more robust conditional reservation mechanism described in bug 639938