A block storage node data repair method and storage medium

Abstract

The invention discloses a method for recovering data of a block storage node, which includes that the node in the process of fault recovery initiates a request for starting data recovery to the master node, and the master node accepts the request for data repair, and returns the latest log ID to the node in the process of fault recovery; the fault recovery The middle node and the master node perform log synchronization, and mark whether the current page needs to be repaired according to the synchronous log situation; for the pages marked as needing repair, the fault repairing node registers with the QOS controller and applies for the number of pages required for repair ; During fault recovery, the node sends a data recovery request to the master node for data recovery. The invention also provides a storage medium, which can ensure that the reading and writing service to the client is not interrupted while the data is being restored.

Description

technical field [0001] The invention belongs to the field of distributed storage, and more specifically relates to a block storage node data repair method and a storage medium. Background technique [0002] With the rapid development of the Internet and the advent of the era of big data, enterprises are increasingly dependent on storage, and the cost of a large number of high-end hosts and traditional storage is very high; while low-end blade machines and cheap disks, combined with distributed storage software Together, it is the preferred storage architecture for more and more enterprises. In the context of large-scale data storage nodes, host failures and disk failures of storage nodes are not accidental events, but normal. How to solve the situation of normalized hardware failures and provide highly available and highly secure storage services is a problem that all distributed storage service providers need to consider. [0003] At present, the restoration technologies ...

AI Technical Summary

Problems solved by technology

[0005] The disadvantage of this technology is that: nodes need to store a large amount of data, and each storage node needs to store a copy of a file, and the storage redundancy is large, resulting in a large waste of storage resources

[0007] However, there is a problem with erasure codes when repairing damaged data nodes: repairing M1-sized data blocks requires a total of k×M1-sized data blocks to be downloaded from k different nodes through network connections, so the cost of repairing bandwidth is high

[0010] When repairing the damaged part, the storage node of the damaged part needs to be isolated from the system, and the storage node is repaired separately. During this process, the distributed storage system cannot respond to the client's read request. After that, the distributed storage system will give normal feedback to the client's read request, that is, the distributed storage system will experience business interruption during the data restoration process, which will affect the business performance of the distributed storage system.

[0011] The present invention improves the copy-based data repair technology to overcome the business interruption in the process of data repair in the prior art

Images