Routing management method and apparatus of node controller

Abstract

The embodiment of the invention provides a routing management method and apparatus of a node controller. The method comprises the following steps: judging whether each link where a target NC is located is faulty according to the state information of the target NC connected with a substrate controller; if one faulty link exists, sending first routing configuration information to the target NC, andsending second routing configuration information to a to-be-configured NC corresponding to the faulty link, wherein the first routing configuration information is used by the target NC to obtain firstrouting information of the to-be-configured NC, the second routing configuration information is used by the to-be-configured NC to obtain second routing information of the target NC, and skipping NCsindicated by the first routing information and the second routing information are different. According to the routing management method and apparatus provided by the embodiment of the invention, as the skipping NCs indicated by the first routing information and the second routing information are different, the phenomenon of node locking caused by the resource competition between the CPU connectedwith the target NC and the CPU connected with the to-be-configured NC is avoided, therefore the possibility of system downtime is reduced.

Description

technical field [0001] The present application relates to the computer field, in particular to a routing management method and device for a node controller. Background technique [0002] With the development of technology, the expansion capability of the central processing unit (Central Processing Unit, CPU) can be increased to more than 8P (that is, to realize the interconnection of more than 8 CPUs) through an external node controller (Node Controller, referred to as NC). ), such as the current mainstream 16P and 32P high-performance servers, as well as the higher-spec 64P products that can be supported in the future, can use the CPU system connected by NC. [0003] At present, at least one NC group including 4 NCs is generally used to realize the interconnection of multiple CPUs. The NCs in the NC group are connected in pairs, and each NC is connected to at least one group of CPUs. A group of CPUs is called a high-speed interconnection protocol (Quick- PathInterconnect (...

AI Technical Summary

Problems solved by technology

[0004] Under the above interconnection architecture, if the NC link corresponding to the shortest path for mutual access between CPUs in different QPI domains is abnormal, it needs to be forwarded through other NCs, that is, there are multiple jump NCs to choose from, which is prone to resource competition. Cause node deadlock, causing a major risk of system downtime

Images