Slow disk detection processing method based on distributed storage

The invention belongs to the technical field of distributed storage disk detection, and particularly relates to a slow disk detection processing method based on distributed storage, which comprises the following steps of: firstly, dividing each storage node into a master node and a slave node, and regularly registering the slave node to the master node; reading the maintained related information of each node from the main node, notifying each node to complete disk data collection within a certain time, and completing a data collection process after summarizing; after a certain amount of data is reached, grouping the disks according to different service scenes based on a clustering algorithm; identifying whether a slow disk exists in the groups of different service scenes or not; and finally, accumulating the slow disk condition by adopting an integral mode, and identifying as a slow disk when an integral threshold value is reached. Through the slow disk detection method, the accuracy of slow disk recognition