Fault tolerance method and device for disk arrays

Abstract

The invention provides a fault tolerance method and device for disk arrays, wherein the method comprises the following steps: when a disk in a disk array goes wrong, adding a hot spare into the disk array so as to replace the disk which goes wrong, and carrying out reconstruction on the disk array into which the hot spare is added in strips; when a reconstructed current strip has a reconstruction read error, recording the identifier of the current strip into a non-volatile random access memory (NVRAM), and skipping over the current strip, and continuing to carry out reconstruction from the next strip until the reconstruction on the disk array is completed; and aiming at the identifier of each strip recorded in the NVRAM, repairing the reconstruction read error of the strip corresponding to the identifier of the strip by writing, and deleting the identifier of the strip from the NVRAM after the repairing is completed. By using the method and device provided by the invention, the occurrence of problems caused by the reconstruction read errors or service read errors of the disk array in a degraded state can be avoided.

Description

technical field [0001] The invention relates to the field of storage, in particular to a fault-tolerant method and device of a disk array. Background technique [0002] Redundant Array of Independent Disks (RAID: Redundant Array of Independent Disks), referred to as a disk array, combines multiple independent disks into an array to provide good redundancy and higher storage performance than a single disk. In the field of storage, data is directly or indirectly stored on multiple individual disks through the redundancy of the disk array itself, so that data will not be lost when one or more disks fail, that is, data fault tolerance is realized. [0003] Wherein, when the redundancy of the disk array is lost due to some reasons such as failure of a disk in the disk array, the disk array will be in a degraded state. Taking a disk failure in a disk array to cause the disk array to lose its redundancy and make the disk array in a degraded state as an example, in the prior art, i...

AI Technical Summary

Problems solved by technology

However, during the rebuilding process, if a disk rebuilding read error occurs again, where the rebuilding read error is a read error caused by rebuilding I / O during the rebuilding process, then the rebuilding is stopped. At this time, the disk array can only Stuck in degraded state, unable to go back to redundant state

Once other disks in the disk array fail again, the entire disk array will fail, that is, the I / O channel will be closed, which will not only cause the disk array to stop providing services, but also cause the data stored in the disk array to be lost

[0004] In addition, when a business read is performed on a disk array in a degraded state, if a business read error occurs, the business read error is: a read error caused by business I / O on the disk during the business read and write process, then, at this time, the The disk array fails, that is, the I / O channel is closed, which causes the disk array to stop providing services and leads to the loss of previously stored data

Images