The invention discloses a fault-tolerance method for a high-reliability
disk array. The fault-tolerance method comprises the following steps: when a data write error occurs, confirming storage space of a disk in which data with the write error is, and carrying out bad block marking on the storage space; when a data read error occurs, judging if the
disk array is complete, if so, working out current position data by checking, writing the data in a corresponding position of the disk, and otherwise, carrying out bad block marking on the storage space with the data read error; and when a next write operation is carried out, carrying out the write operation on the storage space subjected to bad block marking, if the storage space is successfully written and the subsequent read operation is also successful, removing a bad block mark of the storage space, and on the contrary, remaining the bad block mark. According to the fault-tolerance method, the technical problems that only bad block information of read-write errors is recorded and bad blocks are not repaired in a
disk array system and the disk array
system is limited by a fault-tolerance range in the prior art are solved; and the fault-tolerance method is in particular suitable for occasions with low
data accuracy requirements.