Process, apparatus, and program for system management

A computer system comprising:

• a first disk array that redundantly stores distributed data distributed into blocks so as to recover a failed block of distributed data from other blocks of the same distributed data;
• a detection unit that detects a block in the first disk array in which a readout failure has occurred;
• a request unit that requests, when a detected readout failure goes beyond a redundancy of the first disk array, another computer comprising a second disk array that stores a duplicate of the distributed data stored in the first disk array to send a block that is stored in said second disk array and corresponds to the failed block;
• an overwriting unit that overwrites the failed block with a block received from said another computer;
• a reception unit which receives, from said another computer, a request for supplementation of one of the blocks of data stored in said second disk array in which a readout failure occurs;
• a block readout unit which reads out one of the blocks of said distributed data stored in said first disk array corresponding to said request for supplementation when said reception unit receives said request for supplementation;
• and
• a transmission unit which transmits to said another computer said one of the blocks read out by said block readout unit;
• wherein, when said request for supplementation is received by said reception unit and a readout failure occurring in the blocks of said distributed data stored in the first disk array is detected by said failure detection unit, said transmission unit transmits to said another computer said one of the blocks read out by said block readout unit, and said request unit requests said another computer to supplement said one of the blocks in which the readout failure irrecoverable by use of only the data stored in said first disk array occurs.

A system management method for recovering a first disk array that redundantly stores distributed data distributed into blocks from a readout failure, the method comprising:

• performing a failure detecting operation for detecting a readout failure occurring in any of the blocks in the first disk array storing the distributed data;
• requesting another computer including a second disk array that stores a duplicate of the distributed data stored in the first disk array to supplement a block of the first disk array in which a readout failure irrecoverable from other blocks of the same distributed data stored in the first disk array has occurred;
• receiving a block corresponding to a failed block, among the failed blocks stored in the first disk array, from said another computer;
• overwriting the failed block with the received block;
• recovering another failed block from the received block and other blocks of the same distributed data that have not failed;
• overwriting said another failed block with the recovered block;
• receiving, from said another computer, a request for supplementation of one of the blocks of data stored in said second disk array in which a readout failure occurs;
• reading out one of the blocks of said distributed data stored in said first disk array corresponding to said request for supplementation;
• and
• transmitting to said another computer said one of the blocks that is read out of said first disk array;
• wherein, when said request for supplementation is received and a readout failure occurring in the blocks of said distributed data stored in the first disk array is detected, the method requests said another computer to supplement said one of the blocks in which the readout failure irrecoverable by use of only the data stored in said first disk array occurs, in addition to said transmitting of said one of the blocks read out of the first disk array to said another computer.