You need parity, which is cheap, or lockstep duplexing, which isn't. Or, you know, sometimes you can just restart malfunctioning processes and repair corrupted

You need parity, which is cheap, or lockstep duplexing, which isn't. Or, you know, sometimes you can just restart malfunctioning processes and repair corrupted filesystems while you run the failed tasks again on another node.