The NameNode and the JobTracker are single points of failure.
If your Hadoop version 0.21 or higher. then you can configure checkpoint node for back up. If your Hadoop version is lesser then it then secondary name node is there for it.
You can create back up note instead for master failure. Back up node, checkpoint node hardware configuration should be the same as name node.
Hope this answer helps :)