摘要 |
PURPOSE:To improve the error recoverableness and to reduce the overhead due to input/output interrupt in comparison with recovery dependent upon OS by requesting reconnection to a channel while avoiding the route where error occurs. CONSTITUTION:The controller which controls each route refers to and updates a control memory to connect channels again while avoiding the route where error occurs, and a retry command is executed to recover the error. That is, in a step 103, error is detected in a disk controller, a controller, or a disk driving device. In a step 104, it is discriminated whether the route should be limited to retry a command for this error or not. If the route should be limited, it is discriminated in a step 105 whether the frequency in error occurrence exceeds a threshold or not. Thus, the command is retried to switch the route independently of OS. |