CN108632142A - The route management method and device of Node Controller - Google Patents

The route management method and device of Node Controller Download PDF

Info

Publication number
CN108632142A
CN108632142A CN201810262549.6A CN201810262549A CN108632142A CN 108632142 A CN108632142 A CN 108632142A CN 201810262549 A CN201810262549 A CN 201810262549A CN 108632142 A CN108632142 A CN 108632142A
Authority
CN
China
Prior art keywords
target
cpu
link
routing
redirects
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201810262549.6A
Other languages
Chinese (zh)
Other versions
CN108632142B (en
Inventor
陈旭灿
刘钢
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
XFusion Digital Technologies Co Ltd
Original Assignee
Huawei Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huawei Technologies Co Ltd filed Critical Huawei Technologies Co Ltd
Priority to CN201810262549.6A priority Critical patent/CN108632142B/en
Publication of CN108632142A publication Critical patent/CN108632142A/en
Application granted granted Critical
Publication of CN108632142B publication Critical patent/CN108632142B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/28Routing or path finding of packets in data switching networks using route fault recovery
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/10Packet switching elements characterised by the switching fabric construction
    • H04L49/109Integrated on microchip, e.g. switch-on-chip
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/25Routing or path finding in a switch fabric

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Hardware Redundancy (AREA)

Abstract

The embodiment of the present application provides a kind of route management method and device of Node Controller, and this method includes:According to the status information for the target NC being connect with baseboard controller, whether each link where judging target NC breaks down;If there are a faulty links, the first routing configuration information is sent to target NC, secondary route configuration information is sent to the corresponding NC to be configured of faulty link, first routing configuration information is obtained for target NC to the first routing iinformation of NC to be configured, secondary route configuration information is obtained for NC to be configured to the secondary route information of target NC, and the NC that redirects of the first routing iinformation and the instruction of secondary route information is differed.The embodiment of the present application, the NC that redirects respectively indicated by the first routing iinformation and secondary route information is differed, avoiding the when of exchanging visits with target the NC CPU connecting and the CPU being connect with NC to be configured carries out resource contention, and the phenomenon that cause node deadlock, to reduce system failure odds.

Description

The route management method and device of Node Controller
Technical field
This application involves computer realm more particularly to the route management methods and device of a kind of Node Controller.
Background technology
It, can will be central with the development of science and technology, by external Node Controller (Node Controller, abbreviation NC) The extended capability of processor (Central Processing Unit, abbreviation CPU) promoted to 8P or more (namely realize 8 with On CPU interconnection), such as current main-stream 16P and 32P high-performance servers, and the more high standard that can subsequently support 64P products in cpu system by NC connections can be used.
It includes that the NC groups of 4 NC realize that the interconnection of multiple CPU, the NC in NC groups connect two-by-two generally to use at present at least one It connects, each NC is connect at least one set CPU, and one group of CPU is known as a high speed interconnection protocol (Quick-Path Interconnect, abbreviation QPI) domain, it can be achieved that CPU cross-domain access.
Under above-mentioned interconnection architecture, if the corresponding NC links of shortest path that the CPU in the domains different Q PI exchanges visits occur It when abnormal, needs to be forwarded by other NC, that is, NC is redirected it can be selected that being susceptible to resource contention with multiple Node deadlock is caused, the material risk of system failure is caused.
Invention content
The embodiment of the present application provides a kind of route management method and device of Node Controller, the CPU in the domains different Q PI When the corresponding NC links of shortest path exchanged visits occur abnormal, realizes the CPU in the domains different Q PI and can still provide for exchanging visit Meanwhile also reducing system failure odds.
In a first aspect, the embodiment of the present application provides a kind of route management method of Node Controller, Node Controller NC groups Including multiple NC, the NC in same NC groups is connected two-by-two, and each CPU groups include multiple CPU, and each NC in NC groups is at least with one A CPU groups connection, which is characterized in that including:
Baseboard controller judges the target NC institutes according to the status information for the target NC being connect with the baseboard controller Each link whether break down;The status information is used to indicate the working condition of each link where the target NC;
If there are the faulty link of a cisco unity malfunction, for the corresponding target NC of the faulty link and NC to be configured, the baseboard controller send the first routing configuration information to the target NC, send secondary route configuration information To the NC to be configured, the first routing configuration information obtains the NC to the first via of the NC to be configured for the target NC By information, secondary route configuration information obtains the NC to be configured to the secondary route of the target NC for the NC to be configured The NC that redirects that information, the first routing iinformation and secondary route information respectively indicate is differed.
In the program, by respectively self-configuring to the routing letter up to other side for faulty link corresponding target NC and NC to be configured Breath so that can still provide for exchanging visits with the target NC CPU connecting and with the NC to be configured CPU connecting;By controlling faulty link pair The target NC answered reach NC to be configured it is corresponding redirect NC and NC to be configured and reach the corresponding NC that redirect of target NC differ, avoid Resource contention is carried out when exchanging visits with target the NC CPU connecting and the CPU being connect with NC to be configured, and causes showing for node deadlock As to reduce system failure odds.
It is described described for the corresponding target NC and NC to be configured of faulty link in a kind of possible design Baseboard controller sends the first routing configuration information to the target NC, sends secondary route configuration information to the NC to be configured Before, the method further includes:
The baseboard controller determines first routing configuration information and secondary route configuration according to dimension sequence routing algorithm Information.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle;
The dimension sequence routing algorithm is based on preset condition, and it is to work as described in the arrival of the first message to enable to the preset condition After target NC, side of first message in the virtual rectangle horizontal direction is sent to first and redirected by the target NC NC, described first redirect NC side of first message in the virtual rectangle vertical direction is sent to it is described to be configured NC;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle horizontal direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle vertical direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle;
The dimension sequence routing algorithm is based on preset condition, and the preset condition is when the first message reaches the target NC Afterwards, it enables to the target NC that side of first message in the virtual rectangle vertical direction is sent to first to redirect NC, described first redirect NC side of first message in the virtual rectangle horizontal direction is sent to it is described to be configured NC;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle vertical direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle horizontal direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, if The NC that redirects of the first routing iinformation instruction is the first NC, then the NC that redirects of the secondary route information instruction is the 2nd NC, If the NC that redirects of the first routing iinformation instruction is the 2nd NC, the NC that redirects of the secondary route information instruction is first NC;
First NC is any NC other than the target NC and the NC to be configured in 4 NC, described 2nd NC is any NC other than the target NC and the NC to be configured in 4 NC.
In a kind of possible design, the status information according to NC, each link where judging the target NC is No failure, including:
For each association NC being connect with the target NC, if being used to indicate in the status information of the target NC described Target NC and the working condition of the link of the association NC compositions are identified as first identifier, it is determined that the target NC with it is described Link between association NC can work normally, if being used to indicate the work of the target NC and the link of the association NC compositions State is identified as second identifier, it is determined that the link failure between the target NC and the association NC.
In a kind of possible design, first routing iinformation and the secondary route information include the enabled mark of routing Know, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is the NC to be configured, the purpose of the secondary route information instruction NC is the target NC.
In a kind of possible design, each CPU in same group is connected two-by-two or annular connection.
Second aspect, the embodiment of the present application provide a kind of route management method of Node Controller, Node Controller NC groups Including multiple NC, multiple NC in NC groups are connected two-by-two, and each CPU groups include multiple CPU, and each NC in NC groups is at least with one A CPU groups connection, including:
Target NC sends status information to first substrate controller BMC, and the status information is used to indicate the target NC The working condition of each link at place;
If the first link failure in multilink where the target NC, receive the first via that the BMC is sent by Configuration information, and according to first routing configuration information, the target NC is obtained to the corresponding another NC of first link The first routing iinformation;Wherein, the NC that redirects indicated in secondary route information redirects NC with what first routing iinformation indicated It differs, the secondary route information is that another NC is obtained according to the 2nd BMC secondary route configuration informations sent.
In a kind of possible design, first routing iinformation and the secondary route information include the enabled mark of routing Know, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is another NC, the purpose NC of the secondary route information instruction For the target NC.
The third aspect, the embodiment of the present application also provide a kind of routing management device of Node Controller, Node Controller NC Group includes multiple NC, and the NC in same NC groups is connected two-by-two, and each CPU groups include multiple CPU, each NC in NC groups at least with One CPU groups connection, including:
Judgment module judges the target for the status information according to the target NC being connect with the baseboard controller Whether each link where NC breaks down;The status information is used to indicate the work shape of each link where the target NC State;
Sending module corresponds to the faulty link for if there are the faulty links of a cisco unity malfunction The target NC and NC to be configured, send the first routing configuration information to the target NC, send secondary route configuration information To the NC to be configured, the first routing configuration information obtains the NC to the first via of the NC to be configured for the target NC By information, secondary route configuration information obtains the NC to be configured to the secondary route of the target NC for the NC to be configured The NC that redirects that information, the first routing iinformation and secondary route information respectively indicate is differed.
In a kind of possible design, described device further includes determining module, is used for:
According to dimension sequence routing algorithm, first routing configuration information and secondary route configuration information are determined.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle, the dimension sequence routing algorithm is based on pre- If condition;
The preset condition is to enable to the target NC by described first after the first message reaches the target NC Side of the message in the virtual rectangle horizontal direction is sent to first and redirects NC, and described first redirects NC by first message Side in the virtual rectangle vertical direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle horizontal direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle vertical direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle, the dimension sequence routing algorithm is based on pre- If condition;
The dimension sequence routing algorithm is based on preset condition, and the preset condition is when the first message reaches the target NC Afterwards, it enables to the target NC that side of first message in the virtual rectangle vertical direction is sent to first to redirect NC, described first redirect NC side of first message in the virtual rectangle horizontal direction is sent to it is described to be configured NC;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle vertical direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle horizontal direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, if The NC that redirects of the first routing iinformation instruction is the first NC, then the NC that redirects of the secondary route information instruction is the 2nd NC, If the NC that redirects of the first routing iinformation instruction is the 2nd NC, the NC that redirects of the secondary route information instruction is first NC;
First NC is any NC other than the target NC and the NC to be configured in 4 NC, described 2nd NC is any NC other than the target NC and the NC to be configured in 4 NC.
In a kind of possible design, the judgment module is specifically used for:
For each association NC being connect with the target NC, if being used to indicate in the status information of the target NC described Target NC and the working condition of the link of the association NC compositions are identified as first identifier, it is determined that the target NC with it is described Link between association NC can work normally, if being used to indicate the work of the target NC and the link of the association NC compositions State is identified as second identifier, it is determined that the link failure between the target NC and the association NC.
In a kind of possible design, first routing iinformation and the secondary route information include the enabled mark of routing Know, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is the NC to be configured, the purpose of the secondary route information instruction NC is the target NC.
In a kind of possible design, each CPU in same group is connected two-by-two or annular connection.
Fourth aspect, the embodiment of the present application also provide a kind of routing management device of Node Controller, Node Controller NC Group includes multiple NC, and multiple NC in NC groups are connected two-by-two, and each CPU groups include multiple CPU, each NC in NC groups at least with One CPU groups connection, which is characterized in that including:
Sending module, for sending status information to first substrate controller BMC, the status information is used to indicate target The working condition of each link where NC;
Routing information acquisition module receives if for the first link failure in the multilink where the target NC The first routing configuration information that the BMC is sent, and according to first routing configuration information obtains the target NC to described The first routing iinformation of the corresponding another NC of first link;Wherein, what is indicated in secondary route information redirects NC and described first The NC that redirects of routing iinformation instruction is differed, and the secondary route information is second that another NC is sent according to the 2nd BMC What routing configuration information obtained.
In a kind of possible design, first routing iinformation and the secondary route information include the enabled mark of routing Know, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is another NC, the purpose NC of the secondary route information instruction For the target NC.
5th aspect, the embodiment of the present application also provide a kind of computer readable storage medium, the computer-readable storage Media storage has computer program, the computer program that processor is made to execute first aspect and any possibility of first aspect Design described in method.
6th aspect, the embodiment of the present application also provide a kind of baseboard controller, including:Memory and processor;
The memory, for storing program instruction;
The processor, for calling the described program stored in the memory instruction to realize first aspect and the On the one hand the method described in any possible design.
7th aspect, the embodiment of the present application also provide a kind of computer readable storage medium, the computer-readable storage Media storage has computer program, the computer program that processor is made to execute second aspect and any possibility of second aspect Design described in method.
Eighth aspect, the embodiment of the present application also provide a kind of Node Controller, including:Memory and processor;
The memory, for storing program instruction;
The processor, for calling the described program stored in the memory instruction to realize second aspect and the Method described in the two any possible designs of aspect.
The route management method of Node Controller provided by the embodiments of the present application, by for the corresponding target NC of faulty link The routing iinformation up to other side is respectively self-configured to NC to be configured so that is connect with the target NC CPU connecting and with NC to be configured CPU can still provide for exchanging visits;By control the corresponding target NC of faulty link reach NC to be configured it is corresponding redirect NC with it is to be configured NC reaches the corresponding NC that redirect of target NC and differs, and avoids mutual with target the NC CPU connecting and the CPU being connect with NC to be configured The phenomenon that carrying out resource contention when visit, and causing node deadlock, to reduce system failure odds.
Description of the drawings
Fig. 1 is the connection diagram one of same group of each CPU in the embodiment of the present application;
Fig. 2 is the connection diagram two of same group of each CPU in the embodiment of the present application;
Fig. 3 is CPU interconnection architectures schematic diagram one provided by the embodiments of the present application;
Fig. 4 is CPU interconnection architectures schematic diagram two provided by the embodiments of the present application;
Fig. 5 is CPU interconnection architectures schematic diagram three provided by the embodiments of the present application;
Fig. 6 is the signaling interaction diagram one of the route management method of Node Controller provided by the embodiments of the present application;
Fig. 7 is the virtual rectangle schematic diagram of NC provided by the embodiments of the present application compositions;
Fig. 8 is the virtual rectangle schematic diagram after being converted after the link of the A and B compositions in Fig. 8 break down;
Fig. 9 is the virtual rectangle schematic diagram after being converted after the link of the A and C compositions in Fig. 8 break down;
Figure 10 is the structural schematic diagram one of the routing management device of Node Controller provided by the embodiments of the present application;
Figure 11 is the structural schematic diagram two of the routing management device of Node Controller provided by the embodiments of the present application;
Figure 12 is the structural schematic diagram of baseboard controller provided by the embodiments of the present application;
Figure 13 is the structural schematic diagram of Node Controller provided by the embodiments of the present application.
Specific implementation mode
The corresponding application scenarios of NC in the present embodiment are illustrated first.
In Non Uniform Memory Access access architectures (Non-Uniform Memory Access Architecture, abbreviation NUMA) in system, between central processing unit (Central Processing Unit, abbreviation CPU) can by interconnection into Row interconnection composition multi-CPU system.A kind of common interconnection be high speed interconnection protocol (Quick-Path Interconnect, Abbreviation QPI), therefore, the domains QPI can be described as by the QPI one group of CPU interconnected.One CPU group or a domain QPI generally have 4 A CPU can also have the CPU of other quantity, such as 8 CPU, is not restricted in the embodiment of the present application;Wherein, in same group CPU connect two-by-two or annular connection.
In addition, can also be connected by super transmission (Hyper-Transport, abbreviation HT) bus between CPU.
Fig. 1 is the connection diagram one of same group of each CPU in the embodiment of the present application, and Fig. 2 is in the embodiment of the present application Same group of each CPU connection diagram two.
Referring to Fig. 1, CPU0, CPU1, CPU2, CPU3 in figure are connected two-by-two, that is to say, that any two CPU in one group Between interconnect.
Referring to Fig. 2, annular connection between CPU0, CPU1, CPU2, CPU3 in figure, that is, CPU0, CPU1, CPU2, CPU3 is sequentially connected, and CPU0 and CPU3 are also connected.
Below to the invention relates to multigroup CPU between interconnection architecture illustrate.
Fig. 3 is CPU interconnection architectures schematic diagram one provided by the embodiments of the present application, and Fig. 4 is CPU provided by the embodiments of the present application Interconnection architecture schematic diagram two, Fig. 5 are CPU interconnection architectures schematic diagram three provided by the embodiments of the present application.
Referring to Fig. 3, in CPU interconnection architectures shown in Fig. 3, NC0, NC1, NC2, NC3 form a NC group, NC0, NC1, It is connected two-by-two between NC2, NC3, (full mesh) is interconnected entirely between alternatively referred to as NC.
NC0 is connect with the first CPU groups 11, NC1 with and the 2nd CPU groups 12 connect, NC2 is connect with the 3rd CPU groups 13, NC3 It is connect with the 4th CPU groups 14.The connection type of CPU in CPU groups can be the connection type in Fig. 1 or Fig. 2.
By interconnection architecture shown in Fig. 3, it can be achieved that the interconnection of 16 CPU.
When the CPU-a in the first CPU groups 11 needs to access the CPU-b in the 2nd CPU groups 12, CPU-a accesses CPU-b's Path can be:CPU-a-NC0-NC1-CPU-b (path 1) realize that the cross-domain access of CPU, the path access for CPU-a The shortest path of CPU-b.
The path that CPU-a accesses CPU-b can also be:CPU-a-NC0-NC2-NC1-CPU-b (path 2) or CPU- A-NC0-NC3-NC1-CPU-b (path 3);NC2 in path 2 is known as redirecting NC, and the NC3 in path 3 is known as redirecting NC; Redirect NC refer to for the CPU of two different groups, not with any CPU is directly connected in two CPU NC.
Normal in the link that NC0 and NC1 is formed, CPU-a, which accesses CPU-b, will not select two paths, only It can select above-mentioned shortest path.
Link in the present embodiment refer to be connected with each other two NC composition link, including two NC of interconnection and Connection cables between two NC.
Referring to Fig. 4, in CPU interconnection architectures shown in Fig. 4, NC0, NC1, NC2, NC3 form a NC group, NC0, NC1, It is connected two-by-two between NC2, NC3.
NC0 is connect with the first CPU groups 11 and the 2nd CPU groups 12, NC1 with and the 3rd CPU groups 13 and the 4th CPU groups 14 company It connects, NC2 is connect with the 5th CPU groups 15 and the 6th CPU groups 16, and NC3 is connect with the 7th CPU groups 17 and the 8th CPU groups 18.CPU groups The connection type of interior CPU can be the connection type in Fig. 1 or Fig. 2.
By interconnection architecture shown in Fig. 4, it can be achieved that the interconnection of 32 CPU.
When the CPU-a in the first CPU groups 11 needs to access the CPU-b in the 2nd CPU groups 12, CPU-a accesses CPU-b's Path can be:CPU-a-NC0-CPU-b realize the cross-domain access of CPU.
When the CPU-a in the first CPU groups 11 needs to access the CPU-c in the 3rd CPU groups 13, CPU-a accesses CPU-c's Path is:CPU-a-NC0-NC1-CPU-c, the path are the shortest path that CPU-a accesses CPU-c.
CPU-a accesses the non-shortest path of CPU-c with reference to above-mentioned explanation, and details are not described herein again.
Referring to Fig. 5, in CPU interconnection architectures shown in Fig. 5, NC0, NC1, NC2, NC3 form a NC group, referred to herein as First NC groups, connect two-by-two between NC0, NC1, NC2, NC3, i.e., full interconnection;NC4, NC5, NC6, NC7 form a NC group, this Place is known as the 2nd NC groups, is connected two-by-two between NC4, NC5, NC6, NC7, i.e., full interconnection.
CPU interconnection architectures shown in fig. 5 are the full interconnection architectures of double NC, can improve CPU interconnection architectures shown in Fig. 4 in this way Reliability.
When the CPU-a in the first CPU groups 11 needs to access the CPU-d in the 4th CPU groups 14, CPU-a accesses CPU-d's Path can be:CPU-a-NC0-NC1-CPU-d (path 1) can also be CPU-a-NC4-NC5-CPU-d (path 2), road Diameter 1 and path 2 are shortest path.If that is failure occurs in the link of NC0-NC1 compositions, CPU-a accesses CPU-d Path can be shortest path 2, if failure occurs in the link of NC4-NC5 compositions, the path that CPU-a accesses CPU-d can be most Short path 1.
CPU-a accesses the non-shortest path of CPU-d with reference to above-mentioned explanation, and details are not described herein again.
As seen from the above, CPU interconnection architectures shown in Fig. 5 improve the reliable of system by using Redundancy Design Property.
Based on the above-mentioned description to CPU interconnection architectures, below to the routing management side of Node Controller provided by the present application Method illustrates.
Fig. 6 is the signaling interaction diagram one of the route management method of Node Controller provided by the embodiments of the present application, referring to figure 6, the method for the present embodiment may include:
Step S101, target NC sends status information to baseboard controller (Baseboard Management Controller, abbreviation BMC), status information is used to indicate the working condition of each link where target NC;
Whether the step S102, status information that BMC is sent according to target NC, each link where judging target NC there is event Barrier;
If there are the faulty link of a cisco unity malfunction step S103, in each link where target NC, the is sent One routing configuration information sends secondary route configuration information and waits matching to faulty link is corresponding to the corresponding target NC of faulty link Set NC;
Step S104, target NC obtains target NC to the first via of NC to be configured by believing according to the first routing configuration information Breath;
Step S105, NC to be configured obtains NC to be configured to the secondary route of target NC according to secondary route configuration information Information;The NC that redirects indicated in first routing iinformation and secondary route information is differed.
Specifically, the target NC in the present embodiment is any NC in CPU interconnection architectures.
BMC is totally independent of CPU, BIOS, operating system, the on-off state of server, is one and is completely independent operation System.Wherein, BMC is typically integrated on server master board, and BMC and corresponding firmware provide outband management work(for server Energy.There are one BMC for each NC connections.
Each link in the present embodiment where target NC refer to include target NC link, for example, CPU shown in fig. 5 Interconnection architecture, each link where NC0 include:The link of NC0 and NC1 compositions, the link of NC0 and NC2 compositions, NC0 and NC3 groups At link;Each link where NC1 includes:The link of NC1 and NC0 compositions, the link of NC1 and NC2 compositions, NC1 and NC4 groups At link;Each link where NC2 includes:The link of NC2 and NC0 compositions, the link of NC2 and NC1 compositions, NC2 and NC3 groups At link;Each link where NC3 includes:The link of NC3 and NC0 compositions, the link of NC3 and NC1 compositions, NC3 and NC2 groups At link.
For the link of any two NC composition, link failure may be one of in two NC NC with it is another NC connections port used is broken down, or the cable of two NC of connection breaks down.
Such as:For the link of NC0 and NC1 compositions, the cable between NC0 and NC1, the company of port 0 of one end and NC0 are connected It connects, the port 0 of one end and NC1 connect, if the link of NC0 and NC1 compositions breaks down, it may be possible to which event occurs in the port 0 of NC0 Barrier, it may be possible to break down the port 0 of NC1, it is also possible to be the cable failure between connecting NC0 and NC1.
Each NC can send state-detection message to NC connected to it, if having received other NC's in preset duration Response message, then the link for illustrating the NC and sending the NC compositions of response message is normal, on the contrary then illustrate the NC and do not send There is failure in the link of the NC compositions of response message.
Such as:NC0 in Fig. 5 periodically can send state-detection message to NC1, NC2, NC3 respectively, if when default The response message of NC1 is had received in length, then illustrates that the link that NC0 and NC1 is formed is normal;If not received in preset duration The response message of NC2 then illustrates the link failure of NC0 and NC2 compositions;If having received the response message of NC3 in preset duration, Then illustrate that the link that NC0 and NC3 is formed is normal.
The instruction information of the state between the NC and other each NC is stored in the status register of each NC.For NC groups For interior any NC, for each of connecting association NC with the NC, if being used to indicate the NC in the status information of the NC and closing The working condition of the link of connection NC compositions is identified as first identifier, it is determined that the NC can be normal with the link being associated between NC Work is identified as second identifier, it is determined that the NC if being used to indicate the NC and be associated with and marking the working condition of the link that NC is formed With the link failure being associated between NC.
For example, for the NC0 in Fig. 5, corresponding target NC is NC1, NC2 and NC3.
Three bits can be used and indicate respectively the chain that working condition, NC0 and the NC2 of the link of NC0 and NC1 compositions are formed The working condition on road, the working condition of the link of NC0 and NC3 compositions, if three links work normally, status information can be 000, if the link of itself and NC1 composition failure occurs, status information can be 100, if its going out with the link of NC3 compositions Failure is showed, then status information can be 001, if failure occurs in each link where NC0, status information can be 111.This When, first identifier 0, second identifier 1.
It will be appreciated by persons skilled in the art that if the first CPU being connect with target NC needs to access and NC to be configured 2nd CPU of connection, and there is no when the shortest path of message forwarding, then target NC is in the message for receiving the first CPU transmissions Afterwards, which is sent to the 2nd CPU by the path that the first NC is indicated according to the first routing iinformation.
If the 2nd CPU being connect with NC to be configured needs to access the first CPU connecting with target NC, and there is no messages to turn When the shortest path of hair, then NC to be configured is after the message for receiving the 2nd CPU transmissions, and NC to be configured is according to secondary route information The message is sent to the first CPU by the path of instruction.
Step S101~step S105 is illustrated with reference to several specific examples.
The link that first example corresponds in Fig. 5 breaks down, remaining link is normal, and target NC is NC0.
Status information can be sent to BMC by NC0 after the order for the read status information for receiving BMC.
BMC receives the status information that NC0 is sent, and according to the status information that NC0 is sent, each link where judging NC0 is No failure.
Specifically, for example, the status information that the NC0 that BMC is received is sent is 010, under framework shown in Fig. 5, three The working condition of first in the bit link 1 for being used to indicate NC0 and NC1 compositions, second are used to indicate NC0 and NC2 groups At link 2 working condition, third position is used to indicate the working condition of the link 3 of NC0 and NC3 composition.It is by status information 010 it is found that first is 0, illustrates that the state of the link 1 of NC0 and NC1 composition is normal operating conditions, second 1, explanation The state of the link 2 of NC0 and NC2 compositions is failure, and third position is 0, illustrates that the state of the link 3 of NC0 and NC3 compositions is normal Working condition.
In summary, event has occurred in the link 2 where failure, that is, NC0 has occurred in the link 2 of NC0 and NC2 compositions Barrier, the second NC (namely the 3rd NC) to be configured is NC2 at this time.
Then BMC sends the first configuration order to 2 corresponding first NC-NC0 to be configured of faulty link, sends the second configuration It orders to 2 corresponding second NC-NC2 to be configured of faulty link.
After NC0 receives the first configuration order, according to the first configuration order, the first routing iinformation of NC0 to NC2 is obtained;NC2 After receiving the second configuration order, according to the second configuration order, the secondary route information of NC2 to NC0 is obtained;
Wherein, may include in routing iinformation:The enabled mark of routing, redirects the mark of NC, the mark of purpose NC;Routing iinformation It is stored in the routing register of corresponding NC.
Framework is it is found that when the link of NC0 and NC2 compositions breaks down according to figure 5, NC0 can by NC1 or NC3 reaches NC2, and the CPU in CPU groups being connect in this way with NC0 can passage path NC0-NC1-NC2 or NC0-NC3- NC2 accesses the CPU in the CPU groups being connect with NC2.At this point, NC1, NC3 are known as redirecting NC.
Equally, NC2 can reach NC0 by NC1 or NC3, and the CPU in CPU groups being connect in this way with NC2 can pass through Path NC2-NC1-NC0 or NC2-NC3-NC0 access the CPU in the CPU groups being connect with NC0.
If the CPU-a passage paths NC0-NC1-NC2 in the CPU groups of NC0 connections is accessed in the CPU groups being connect with NC2 CPU-b passage paths NC2-NC1-NC0 in the CPU groups of CPU-b, NC2 connection accesses the CPU-a in the CPU groups being connect with NC0 When, if CPU-a issues CPU-b messages (namely CPU-a accesses CPU-b), the cache resources of the NC1 of occupancy do not discharge, So when CPU-b replys message (namely CPU-b access CPU-a) to CPU-a, just not enough cache resources make NC1 With, therefore there is resource contention, it is be easy to cause node deadlock, so as to cause the material risk of system failure.
Therefore, in order to solve the above-mentioned technical problem, the first routing iinformation in the application and secondary route information include The NC that redirects differ.If indicated in the first routing iinformation redirect NC be NC1 when, secondary route information instruction redirect NC For NC3, if when to redirect NC be NC3 of secondary route information instruction, the NC that redirects of secondary route information instruction is NC1.
Wherein, since the first routing iinformation is the routing iinformation of NC0 to NC2, then the purpose indicated in the first routing iinformation NC is NC2;Due to secondary route information instruction be NC2 to NC0 routing iinformation, then the purpose indicated in secondary route information NC is NC0.
In conclusion the path of the corresponding first routing iinformation instructions of NC0 is under the example:When NC0-NC1-NC2, NC2 The path of corresponding secondary route information instruction is:The path of the corresponding first routing iinformation instruction of NC2-NC3-NC0, NC0 is: When NC0-NC3-NC2, the path of the corresponding secondary route information instructions of NC0 is NC0-NC1-NC2.
For routeing enabled mark in routing iinformation, after routing iinformation obtains, the enabled mark of routing of the routing iinformation It is identified for third, that is to say, that the routing iinformation is effective, wherein third mark can be 1.
Further, since CPU interconnection architectures shown in fig. 5 include two NC groups, that is, there are Redundancy Designs, such as CPU upper described, that the CPU in CPU groups connect with NC0 can also be in the CPU groups that are connect with NC2 of passage path NC4-NC6 access, That is passage path NC4-NC6 realizes the cross-domain access of above-mentioned CPU, and path is shorter, that is, shortest path;The CPU being connect with NC2 CPU in group can also passage path NC6-NC4 access the CPU in the CPU groups being connect with NC0, i.e. passage path NC6-NC4 realizations The cross-domain access of above-mentioned CPU, path is shorter, that is, shortest path.
So, in the case where the link 2 of above-mentioned NC0 and NC2 compositions breaks down, if the link between NC4 and NC6 does not have Break down, according to shortest path principle, CPU-a in the CPU groups that are connect with NC0 can passage path NC4-NC6 access with CPU-b in the CPU groups of NC2 connections.That is, when there are shortest path, although being configured in the routing register of NC0 NC0 reaches the first routing iinformation of NC2, but in actual access process, the CPU groups that connects with NC0 (and with NC4 companies The CPU groups connect) in the NC0 that can't be indicated by the first routing iinformation of CPU reach the CPU that NC2 path access is connect with NC2 CPU in group (and the CPU groups being connect with NC6), but accessed in the CPU groups being connect with NC2 by shortest path NC4-NC6 CPU;Similarly, when there are shortest path, although being configured with the secondary route letter that NC2 reaches NC0 in the routing register of NC2 Breath, what the CPU in CPU groups (and the CPU groups being connect with NC6) being connect with NC2 can't be indicated by secondary route information NC2 reaches the CPU in the CPU groups (and the CPU groups being connect with NC4) that NC0 path access is connect with NC0, but by most short Path NC6-NC4 accesses the CPU in the CPU groups being connect with NC2.
That is, if the first CPU in the first CPU groups 11 being connect with NC0 needs to access the 5th connect with NC2 The 2nd CPU in CPU groups 15, and there are when the shortest path of message forwarding, then message is sent to and the first CPU by the first CPU NC4 in 2nd NC groups of 11 connection of group reaches the 2nd CPU in the 5th CPU groups 15 through the NC6 in the 2nd NC groups, that is, Message can be forwarded to purpose CPU through shortest path.
If the 2nd CPU in the 5th CPU groups 15 being connect with NC2 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there are when the shortest path of message forwarding, then message is sent to the connect with the 5th CPU groups 15 by the 2nd CPU NC6 in two NC groups reaches the first CPU in the first CPU groups 11 through the NC4 in the 2nd NC groups, that is, message can be through most short Path is forwarded to purpose CPU.
So, in the case where the link 2 of above-mentioned NC0 and NC2 compositions breaks down, if the link between NC4 and NC6 Failure has occurred, that is to say, that there is no when the shortest path of message forwarding, the CPU in CPU groups that is connect with NC0 can be by the The NC0 of one routing iinformation instruction reaches the CPU in the CPU groups that NC2 path access is connect with NC2;Similarly, the CPU being connect with NC2 CPU in group can reach the CPU in the CPU groups that the path access of NC0 is connect with NC0 by the NC2 that secondary route information indicates.
That is, the NC that redirects when the instruction of the first routing iinformation is NC1, the NC that redirects of secondary route information instruction is When NC3, if the first CPU in the first CPU groups 11 being connect with NC0 needs to access the in the 5th CPU groups 15 being connect with NC2 Two CPU, and there is no message forwarding shortest path when, then, can be through the first via by believing after message is sent to NC0 by the first CPU Breath instruction redirects NC-NC1, reaches target NC-NC2, and the 2nd CPU is forwarded to through NC2, that is, message is after reaching NC0, Can be according to the path of the first routing iinformation instruction configured in the routing register of NC0, achieve the goal CPU.
If the 2nd CPU in the 5th CPU groups 15 being connect with NC2 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there is no message forwarding shortest path when, then, can be through secondary route after message is sent to NC2 by the 2nd CPU Information instruction redirects NC-NC3, reaches target NC-NC0, is forwarded to the 2nd CPU through NC0, that is, message is reaching NC2 Afterwards, can be according to the path of the secondary route information instruction configured in the routing register of NC2, achieve the goal CPU.
Equally, when the first routing iinformation instruction redirect NC be NC3, secondary route information instruction redirect NC be NC1 when, If the first CPU in the first CPU groups 11 being connect with NC0 needs to access the 2nd CPU in the 5th CPU groups 15 connecting with NC2, And it there is no when the shortest path of message forwarding, then after message is sent to NC0 by the first CPU, can be indicated through the first routing iinformation Redirect NC-NC3, reach target NC-NC2, the 2nd CPU be forwarded to through NC2, that is, message is after reaching NC0, can be according to The path of the first routing iinformation instruction configured in the routing register of NC0, achieve the goal CPU.
If the 2nd CPU in the 5th CPU groups 15 being connect with NC2 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there is no message forwarding shortest path when, then, can be through secondary route after message is sent to NC2 by the 2nd CPU Information instruction redirects NC-NC1, reaches target NC-NC0, is forwarded to the first CPU through NC0, that is, message is reaching NC2 Afterwards, can be according to the path of the secondary route information instruction configured in the routing register of NC2, achieve the goal CPU.
Above-mentioned elaboration is that " under CPU interconnection architectures shown in Fig. 5, target NC is NC0 to example, and between NC0 and NC2 Link when breaking down " corresponding step S101~step S107 specific implementation.
The link that second example corresponds in Fig. 4 breaks down, and target NC is NC3.
Status information can be sent to BMC by NC3 after the order for the read status information for receiving BMC.
BMC receives the status information that NC3 is sent, and according to the status information that NC3 is sent, each link where judging NC3 is No failure.
Specifically, for example, the status information that the NC3 that BMC is received is sent is 100, under framework shown in Fig. 4, three The working condition of first in the bit link 1 for being used to indicate NC3 and NC0 compositions, second are used to indicate NC3 and NC1 groups At link 2 working condition, third position is used to indicate the working condition of the link 3 of NC3 and NC2 composition.It is by status information 100 it is found that first is 1, illustrates that the state of the link 1 of NC3 and NC0 composition is failure, second 1 illustrates NC3 and NC1 The state of the link 2 of composition is normal work, and third position is 0, illustrates that the state of the link 3 of NC3 and NC2 compositions is normal work State.
In summary, illustrate that the link 1 where failure, that is, NC3 has occurred in the link 1 of NC3 and NC0 compositions has occurred Failure.
Then BMC sends the first configuration order to 1 corresponding first NC-NC3 to be configured of faulty link, sends the second configuration It orders to 1 corresponding second NC-NC0 to be configured of faulty link.
After NC3 receives the first configuration order, according to the first configuration order, the first routing iinformation of NC3 to NC0 is obtained;NC0 After receiving the second configuration order, according to the second configuration order, the secondary route information of NC0 to NC3 is obtained;
Wherein, may include in routing iinformation:The enabled mark of routing, redirects the mark of NC, the mark of purpose NC;Routing iinformation It is stored in the routing register of corresponding NC.
Framework according to Fig.4, it is found that when the link of NC3 and NC0 compositions breaks down, NC3 can by NC1 or NC2 reaches NC0, and the CPU in CPU groups being connect in this way with NC3 can passage path NC3-NC1-NC0 or NC3-NC2- NC0 accesses the CPU in the CPU groups being connect with NC0.At this point, NC1, NC2 are known as redirecting NC.
Equally, NC0 can reach NC3 by NC1 or NC2, and the CPU in CPU groups being connect in this way with NC0 can pass through Path NC0-NC1-NC3 or NC0-NC2-NC3 access the CPU in the CPU groups being connect with NC3.
The generation of the phenomenon that in order to avoid the appearance resource contention described in a upper example occur, causing node deadlock, this The NC that redirects that the first routing iinformation and secondary route information in example include is differed.If being indicated in the first routing iinformation Redirect NC be NC1 when, then secondary route information instruction redirect NC be NC2, if secondary route information instruction redirect NC be NC2 When, then the NC that redirects of secondary route information instruction is NC1.
Wherein, since the first routing iinformation is the routing iinformation of NC3 to NC0, then the purpose indicated in the first routing iinformation NC is NC0;Due to secondary route information instruction be NC0 to NC3 routing iinformation, then the purpose indicated in secondary route information NC is NC3.
In conclusion the path of the corresponding first routing iinformation instructions of NC3 is under the example:When NC3-NC1-NC0, NC0 The path of corresponding secondary route information instruction is:The path of the corresponding first routing iinformation instruction of NC0-NC2-NC3, NC3 is: When NC3-NC2-NC0, the path of the corresponding secondary route information instructions of NC0 is NC3-NC1-NC0.
For routeing enabled mark in routing iinformation, after routing iinformation obtains, the enabled mark of routing of the routing iinformation It is identified for third, that is to say, that the routing iinformation is effective, wherein third mark can be 1.
Further, since CPU interconnection architectures shown in Fig. 4 include a NC group, Redundancy Design is not present, works as NC3 When breaking down with the link 1 of NC0 compositions, the CPU in CPU groups being connect with NC3 accesses the CPU in the CPU groups being connect with NC0 Shortest path also will no longer be present, CPU in the CPU groups being connect with NC0 accesses the CPU's in the CPU groups being connect with NC3 Shortest path is also not present.
So, in the case where the link 1 of above-mentioned NC0 and NC3 compositions breaks down, that is to say, that there is no messages to forward Shortest path when, the CPU in CPU groups that is connect with NC0 can reach the path of NC3 by the NC0 that secondary route information indicates Access the CPU in the CPU groups being connect with NC3;Similarly, the CPU in CPU groups being connect with NC3 can be referred to by secondary route information The CPU in CPU groups that the path access of the NC3 arrival NC0 shown is connect with NC0.
That is, the NC that redirects when the instruction of the first routing iinformation is NC1, the NC that redirects of secondary route information instruction is When NC2, if the first CPU in the 7th CPU groups 17 being connect with NC3 needs to access the in the first CPU groups 11 being connect with NC0 When two CPU, then after message is sent to NC3 by the first CPU, what can be indicated through the first routing iinformation redirects NC-NC1, reaches target NC-NC0 is forwarded to the 2nd CPU through NC0, that is, message is after reaching NC3, can be according to being configured in the routing register of NC3 The first routing iinformation instruction path, achieve the goal CPU.
If the 2nd CPU in the first CPU groups 11 being connect with NC0 needs to access in the 7th CPU groups 17 connecting with NC3 When one CPU, then after message is sent to NC0 by the 2nd CPU, what can be indicated through secondary route information redirects NC-NC2, reaches mesh NC-NC3, be forwarded to the first CPU through NC3, that is, message is after reaching NC0, can be according to matching in the routing register of NC0 The path for the secondary route information instruction set, achieve the goal CPU.
Equally, when the first routing iinformation instruction redirect NC be NC2, secondary route information instruction redirect NC be NC1 when, If the first CPU in the 7th CPU groups 17 being connect with NC3 needs to access the 2nd CPU in the first CPU groups 11 connecting with NC0 When, then after message is sent to NC3 by the first CPU, what can be indicated through the first routing iinformation redirects NC-NC2, reaches target NC- NC0 is forwarded to the 2nd CPU through NC0, that is, message is after reaching NC3, can be according to the configured in the routing register of NC3 The path of one routing iinformation instruction, achieve the goal CPU.
If the 2nd CPU in the first CPU groups 11 being connect with NC0 needs to access in the 7th CPU groups 17 connecting with NC3 When one CPU, then after message is sent to NC0 by the 2nd CPU, what can be indicated through secondary route information redirects NC-NC1, reaches mesh NC-NC3, be forwarded to the first CPU through NC3, that is, message is after reaching NC0, can be according to matching in the routing register of NC0 The path for the secondary route information instruction set, achieve the goal CPU.
Above-mentioned elaboration is that " under CPU interconnection architectures shown in Fig. 4, target NC is NC3 to example, and between NC3 and NC0 Link when breaking down " corresponding step S101~step S107 specific implementation.
It will be appreciated by persons skilled in the art that because each NC corresponds to a BMC, step S101~step S102 is every Interaction between a BMC NC connected to it is all carrying out, and step S103~step S107 is answered in the link pair to break down NC and corresponding BMC between existing interactive process, that is to say, that if only a link breaks down, only there are one NC The interactive process of step S101~step S107 is carried out between corresponding BMC.
The example that uniform link breaks down in above-mentioned two example, below in CPU interconnection architectures, both links go out When existing failure, the route management method of Node Controller illustrates.
If failure has occurred in the link where first object NC, failure also has occurred in the link where the second target NC, the Link where one target NC is differed with the link where the second target NC, and first object NC and the second target NC are not same , then there is following interactive process in one NC:
For first object NC and the first BMC:
A1, first object NC send first state information to first substrate controller, and first state information is used to indicate the The working condition of each link where one target NC;
According to first state information, whether each link where judging first object NC breaks down by b1, the first BMC;
If there are the Fisrt fault link of a cisco unity malfunction in each link where c1, first object NC, for Fisrt fault link sends first object routing configuration information to the corresponding first object NC of Fisrt fault link, sends second Target routing configuration information is to the corresponding NC to be configured of Fisrt fault link;The corresponding NC to be configured of Fisrt fault link is known as Three target NC;
D1, first object NC obtain first object NC to Fisrt fault link pair according to first object routing configuration information The first object routing iinformation of the second NC to be configured answered;
E1, third target NC obtain the of third target NC to first object NC according to the second target routing configuration information Two target routing iinformations;The NC that redirects indicated in first object routing iinformation and the second target routing iinformation is differed.
It will be appreciated by persons skilled in the art that if the first CPU being connect with first object NC needs access and third 2nd CPU of target NC connections, and there is no when the shortest path of message forwarding, then first object NC is receiving the first CPU After the message of transmission, the first NC is sent to the 2nd CPU according to first object routing iinformation, by the message;If with third target NC 2nd CPU of connection needs to access the first CPU that connect with first object NC, and when the shortest path that message forwards is not present, Then third target NC is after the message for receiving the 2nd CPU transmissions, and third target NC is according to the second target routing iinformation, by the report Text is sent to the first CPU.
For the second target NC and the 2nd BMC:
A2, the second target NC send the second status information to second substrate controller, and the second status information is used to indicate the The working condition of each link where two target NC;
According to the second status information, whether each link where judging the second target NC breaks down by b2, the 2nd BMC;
If there are the second faulty link of a cisco unity malfunction in each link where c2, the second target NC, for Second faulty link is sent where sending third target routing configuration information to the corresponding second target NC of the second faulty link The corresponding NC to be configured of 4th target routing configuration information to the second faulty link;The corresponding NC to be configured of second faulty link claims For the 4th target NC;
D2, the second target NC obtain the of the second target NC to the 4th target NC according to third target routing configuration information Three target routing iinformations;
E2, the 4th target NC obtain the of the 4th target NC to the second target NC according to third target routing configuration information Four target routing iinformations;The NC that redirects indicated in third target routing iinformation and the 4th target routing iinformation is differed;
It will be appreciated by persons skilled in the art that if the 3rd CPU being connect with the second target NC needs to access and the 4th 4th CPU of target NC connections, and there is no when the shortest path of message forwarding, then the second target NC is receiving the 3rd CPU After the message of transmission, the second target NC is sent to the 4th CPU according to third target routing iinformation, by the message;If with the 4th mesh 4th CPU of mark NC connections needs to access the 3rd CPU connecting with the second target NC, and there is no the shortest paths of message forwarding When, then for the 4th target NC after the message for receiving the 4th CPU transmissions, the 4th target NC, will according to the 4th target routing iinformation The message is sent to the 3rd CPU.
Specifically, the first object NC in the present embodiment is any NC in CPU interconnection architectures, and the second target NC is CPU Any NC in interconnection architecture, first object NC and the second target NC are not the same NC.
Each link in the present embodiment where first object NC refer to include first object NC link, for example, Fig. 5 institutes The CPU interconnection architectures shown, each link where NC0 include:The link of NC0 and NC1 compositions, the link of NC0 and NC2 compositions, NC0 With the link of NC3 compositions;Each link where NC1 includes:The link of NC1 and NC0 compositions, the link of NC1 and NC2 compositions, NC1 With the link of NC4 compositions;Each link where NC2 includes:The link of NC2 and NC0 compositions, the link of NC2 and NC1 compositions, NC2 With the link of NC3 compositions;Each link where NC3 includes:The link of NC3 and NC0 compositions, the link of NC3 and NC1 compositions, NC3 With the link of NC2 compositions.Similarly, each link in the present embodiment where the second target NC refer to include the second target NC chain Road.
Above-mentioned two interactive process is illustrated with reference to specific example.
The both links for belonging to same NC groups that the example corresponds in Fig. 5 break down, and first object NC is NC0, the Two target NC are NC3.
The first state information of itself can be sent to BMC by NC0 after the order for the read status information for receiving BMC.
BMC receives the status information that NC0 is sent, and according to the status information that NC0 is sent, each link where judging NC0 is No failure.
Specifically, for example, the status information that the NC0 that BMC is received is sent is 100, under framework shown in Fig. 5, three The working condition of first in the bit link 1 for being used to indicate NC0 and NC1 compositions, second are used to indicate NC0 and NC2 groups At link 2 working condition, third position is used to indicate the working condition of the link 3 of NC0 and NC3 composition.It is by status information 100 it is found that first is 1, illustrates that the state of the link 1 of NC0 and NC1 composition is failure, second 0 illustrates NC0 and NC2 The state of the link 2 of composition is normal operating conditions, and third position is 0, illustrates that the state of the link 3 of NC0 and NC3 compositions is normal Working condition.
In summary, event has occurred in the link 1 where failure, that is, NC0 has occurred in the link 1 of NC0 and NC1 compositions The link 1 of barrier, NC0 and NC1 composition is known as Fisrt fault link, and the corresponding NC to be configured of Fisrt fault link is (namely at this time Third target NC) it is NC1.
Likewise, NC3 can be after the order for the read status information for receiving BMC, the second status information of itself is sent To BMC.
BMC receives the status information that NC3 is sent, and according to the status information that NC3 is sent, each link where judging NC3 is No failure.
Specifically, for example, the status information that the NC3 that BMC is received is sent is 001, under framework shown in Fig. 5, three The working condition of first in the bit link 1 for being used to indicate NC3 and NC0 compositions, second are used to indicate NC3 and NC1 groups At link 2 working condition, third position is used to indicate the working condition of the link 3 of NC3 and NC2 composition.It is by status information 001 it is found that first is 0, illustrates that the state of the link 1 of NC3 and NC0 composition is normal operating conditions, second 0, explanation The state of the link 2 of NC3 and NC1 compositions is normal operating conditions, and third position is 1, illustrates the shape of the link 3 of NC3 and NC2 compositions State is normal operating conditions.
In summary, event has occurred in the link 3 where failure, that is, NC3 has occurred in the link 3 of NC3 and NC2 compositions The link 3 of barrier, NC3 and NC2 composition is known as the second faulty link, and the corresponding NC to be configured of the second faulty link is (namely at this time 4th target NC) it is NC2.
BMC sends first object routing configuration information to the corresponding first NC-NC0 to be configured of Fisrt fault link, sends Second target routing configuration information is to corresponding second NC (the 3rd the NC)-NC1 to be configured of Fisrt fault link.
After NC0 receives first object routing configuration information, according to first object routing configuration information, NC0 is obtained to NC1's First object routing iinformation;After NC1 receives the second target routing configuration information, according to the second target routing configuration information, obtain The second tunnel target of NC1 to NC0 is by information;
Wherein, may include in routing iinformation:The enabled mark of routing, redirects the mark of NC, the mark of purpose NC;Routing iinformation It is stored in the routing register of corresponding NC.
For routeing enabled mark in routing iinformation, after routing iinformation obtains, the enabled mark of routing of the routing iinformation It is identified for third, that is to say, that the routing iinformation is effective, wherein third mark can be 1.
What first object routing iinformation and the second target routing iinformation included redirect, and NC differs that (concrete reason is referring to upper One elaboration implemented).If indicated in first object routing iinformation redirect NC be NC2 when, the second target routing iinformation instruction Redirect NC be NC3, if the second target routing iinformation instruction redirect NC be NC3 when, the second target routing iinformation instruction It is NC2 to redirect NC.Since first object routing iinformation is the routing iinformation of NC0 to NC1, then indicated in first object routing iinformation Purpose NC be NC1;Due to the second target routing iinformation instruction be NC1 to NC0 routing iinformation, then the second target routing letter The purpose NC indicated in breath is NC0.
Namely when the link of NC0 and NC1 compositions breaks down, NC0 can reach NC1 by NC3 or NC2, in this way The CPU in CPU groups being connect with NC0 can passage path NC0-NC2-NC1 or NC0-NC3-NC1, access connect with NC1 CPU groups in CPU;NC1 can reach NC0 by NC3 or NC2, and the CPU in CPU groups being connect in this way with NC1 can lead to Path NC1-NC2-NC0 or NC1-NC3-NC0 are crossed, the CPU in the CPU groups being connect with NC0 is accessed.At this point, NC2, NC3 are known as Redirect NC.
Similarly, BMC sends third target routing configuration information to corresponding first NC-to be configured of the second faulty link NC3 sends corresponding second NC (the 4th the NC)-NC2 to be configured of the 4th target routing configuration information to the second faulty link.
After NC3 receives third target routing configuration information, according to third target routing configuration information, NC3 is obtained to NC2's Third target routing iinformation;After NC2 receives the 4th target routing configuration information, according to the 4th target routing configuration information, obtain The 4th target routing iinformation of NC2 to NC3.
What third target routing iinformation and the 4th target routing iinformation included redirect, and NC differs that (concrete reason is referring to upper One elaboration implemented).If indicated in third target routing iinformation redirect NC be NC0 when, the 4th target routing iinformation instruction Redirect NC be NC1, if the 4th target routing iinformation instruction redirect NC be NC1 when, the 4th target routing iinformation instruction It is NC0 to redirect NC.Since third target routing iinformation is the routing iinformation of NC3 to NC2, then indicated in first object routing iinformation Purpose NC be NC2;Due to the 4th target routing iinformation instruction be NC2 to NC3 routing iinformation, then the 4th target routing letter The purpose NC indicated in breath is NC3.
Namely when the link of NC3 and NC2 compositions breaks down, NC3 can reach NC2 by NC0 or NC1, in this way The CPU in CPU groups being connect with NC3 can passage path NC3-NC0-NC2 or NC3-NC1-NC2, access connect with NC2 CPU groups in CPU;NC2 can reach NC3 by NC0 or NC1, and the CPU in CPU groups being connect in this way with NC2 can lead to Path NC2-NC0-NC3 or NC2-NC1-NC3 are crossed, the CPU in the CPU groups being connect with NC3 is accessed.At this point, NC0, NC1 are known as Redirect NC.
For the Fisrt fault link of NC0 and NC1 compositions:
Since the CPU interconnection architectures in Fig. 5 include two NC groups, the CPU groups being connect with NC0 are also connect with NC4, with NC1 The CPU groups of connection are also connect with NC5, and therefore, the CPU in CPU groups being connect in this way with NC0 can also passage path NC4-NC5 visits Ask the CPU in the CPU groups being connect with NC1.It is found that passage path NC4-NC5 realizes the cross-domain access of above-mentioned CPU, path is shorter, Namely shortest path.The CPU in CPU groups being connect with NC1 can also passage path NC5-NC4 access the CPU groups being connect with NC0 Interior CPU.It is found that passage path NC5-NC4 realizes the cross-domain access of above-mentioned CPU, path is shorter, that is, shortest path.
So, in the case where the link of above-mentioned NC0 and NC1 compositions breaks down, if the link between NC4 and NC5 does not have Break down, according to shortest path principle, CPU-a in the CPU groups that are connect with NC0 can passage path NC4-NC5 access with CPU-b in the CPU groups of NC1 connections.That is, when there are shortest path, although being configured in the routing register of NC0 NC0 reaches the first object routing iinformation of NC1, but in actual access process, the CPU groups that are connect with NC0 (and with The CPU groups of NC4 connections) in CPU can't reach NC1 path access and NC1 by NC0 that first object routing iinformation indicates CPU in the CPU groups (and the CPU groups being connect with NC5) of connection, but accessed by shortest path NC4-NC5 and connect with NC1 CPU groups in CPU;Similarly, when there are shortest path, although being configured with NC1 in the routing register of NC1 reaches NC0's Second target routing iinformation, the CPU in the CPU groups being connect with NC1 (and the CPU groups being connect with NC5) can't be by second The NC1 of target routing iinformation instruction is reached in the CPU groups (and the CPU groups being connect with NC4) that NC0 path access is connect with NC0 CPU, but the CPU in the CPU groups being connect with NC0 is accessed by shortest path NC5-NC4.
That is, if the first CPU in the first CPU groups 11 being connect with NC0 needs to access the 4th connect with NC1 The 2nd CPU in CPU groups 14, and there are when the shortest path of message forwarding, then message is sent to and the first CPU by the first CPU NC4 in second target NC groups of 11 connection of group, second in the 4th CPU groups 14 is reached through the NC5 in the second target NC groups CPU, that is, message can be forwarded to purpose CPU through shortest path.
If the 2nd CPU in the 4th CPU groups 14 being connect with NC1 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there are when the shortest path of message forwarding, then message is sent to the connect with the 4th CPU groups 14 by the 2nd CPU NC5 in two target NC groups reaches the first CPU in the first CPU groups 11, that is, message through the NC4 in the second target NC groups It can be forwarded to purpose CPU through shortest path.
So, in the case where the link 2 of above-mentioned NC0 and NC1 compositions breaks down, if the link between NC4 and NC5 Failure has occurred, that is to say, that there is no when the shortest path of message forwarding, the CPU in CPU groups that is connect with NC0 can be by the The NC0 of one target routing iinformation instruction reaches the CPU in the CPU groups that NC1 path access is connect with NC1;Similarly, it is connect with NC1 CPU groups in the NC1 that can be indicated by the second target routing iinformation of CPU reach the CPU groups that be connect with NC0 of NC0 path access Interior CPU.
That is, the NC that redirects when the instruction of first object routing iinformation is NC2, the jump of the second target routing iinformation instruction When to turn NC be NC3, if the 4th CPU groups 14 that are connect with NC1 of the first CPU needs access in the first CPU groups 11 connect with NC0 The 2nd interior CPU, and there is no message forwarding shortest path when, then, can be through first after message is sent to NC0 by the first CPU The instruction of target routing iinformation redirects NC-NC2, reaches target NC-NC1, is forwarded to the 2nd CPU through NC1, that is, message exists After reaching NC0, it can achieve the goal according to the path of the first object routing iinformation instruction configured in the routing register of NC0 CPU。
If the 2nd CPU in the 4th CPU groups 14 being connect with NC1 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there is no message forwarding shortest path when, then, can be through the second target after message is sent to NC1 by the 2nd CPU Routing iinformation instruction redirects NC-NC3, reaches target NC-NC0, is forwarded to the 2nd CPU through NC0, that is, message is reaching , can be according to the path of the second target routing iinformation instruction configured in the routing register of NC1 after NC1, achieve the goal CPU.
Equally, when the NC that redirects of first object routing iinformation instruction is NC3, the instruction of the second target routing iinformation redirects NC For NC2 when, if the first CPU in the first CPU groups 11 being connect with NC0 needs to access in the 4th CPU groups 14 connecting with NC1 2nd CPU, and there is no message forwarding shortest path when, then, can be through first object after message is sent to NC0 by the first CPU Routing iinformation instruction redirects NC-NC3, reaches target NC-NC1, is forwarded to the 2nd CPU through NC1, that is, message is reaching , can be according to the path of the first object routing iinformation instruction configured in the routing register of NC0 after NC0, achieve the goal CPU.
If the 2nd CPU in the 4th CPU groups 14 being connect with NC1 needs to access in the first CPU groups 11 connecting with NC0 First CPU, and there is no message forwarding shortest path when, then, can be through the second target after message is sent to NC1 by the 2nd CPU Routing iinformation instruction redirects NC-NC2, reaches target NC-NC0, is forwarded to the 2nd CPU through NC0, that is, message is reaching , can be according to the path of the second target routing iinformation instruction configured in the routing register of NC1 after NC1, achieve the goal CPU.
For the second faulty link of NC3 and NC2 compositions:
Since the CPU interconnection architectures in Fig. 5 include two NC groups, the CPU groups being connect with NC3 are also connect with NC7, with NC2 The CPU groups of connection are also connect with NC6, and therefore, the CPU in CPU groups being connect in this way with NC3 can also passage path NC7-NC6 visits Ask the CPU in the CPU groups being connect with NC2, it is known that, passage path NC7-NC6 realizes the cross-domain access of above-mentioned CPU, and path is shorter, Namely shortest path.The CPU in CPU groups being connect with NC2 can also passage path NC6-NC7 access the CPU groups being connect with NC3 Interior CPU, it is known that, passage path NC6-NC7 realizes the cross-domain access of above-mentioned CPU, and path is shorter, that is, shortest path.
So, in the case where the link of above-mentioned NC3 and NC2 compositions breaks down, if the link between NC7 and NC6 does not have It breaks down, according to shortest path principle, the CPU in CPU groups being connect with NC3 can passage path NC7-NC6 access and NC2 CPU in the CPU groups of connection.That is, when there are shortest path, arrived although being configured with NC3 in the routing register of NC3 Up to the first object routing iinformation of NC2, but in actual access process, the CPU groups that are connect with NC3 (and connect with NC7 CPU groups) in CPU can't reach what NC2 path access be connect with NC2 by NC3 that first object routing iinformation indicates CPU in CPU groups (and the CPU groups being connect with NC6), but the CPU being connect with NC2 is accessed by shortest path NC7-NC6 CPU in group;Similarly, when there are shortest path, although being configured with the second mesh that NC2 reaches NC3 in the routing register of NC2 Routing iinformation is marked, the CPU in the CPU groups being connect with NC2 (and the CPU groups being connect with NC6) can't pass through the second target road CPU in the CPU groups (and the CPU groups being connect with NC7) that NC3 path access is connect with NC3 is reached by the NC2 of information instruction, But the CPU in the CPU groups being connect with NC3 is accessed by shortest path NC6-NC7.
That is, if the first CPU in the 8th CPU groups 18 being connect with NC3 needs to access the 6th connect with NC2 The 2nd CPU in CPU groups 16, and there are when the shortest path of message forwarding, then message is sent to and the 8th CPU by the first CPU NC7 in second target NC groups of 18 connection of group, second in the 6th CPU groups 16 is reached through the NC6 in the second target NC groups CPU, that is, message can be forwarded to purpose CPU through shortest path.
If the 2nd CPU in the 6th CPU groups 16 being connect with NC2 needs to access in the 8th CPU groups 18 connecting with NC3 First CPU, and there are when the shortest path of message forwarding, then message is sent to the connect with the 6th CPU groups 16 by the 2nd CPU NC6 in two target NC groups reaches the first CPU in the first CPU groups 11, that is, message through the NC7 in the second target NC groups It can be forwarded to purpose CPU through shortest path.
So, in the case where the link 2 of above-mentioned NC3 and NC2 compositions breaks down, if the link between NC6 and NC7 Failure has occurred, that is to say, that there is no when the shortest path of message forwarding, the CPU in CPU groups that is connect with NC3 can be by the The NC3 of one target routing iinformation instruction reaches the CPU in the CPU groups that NC2 path access is connect with NC2;Similarly, it is connect with NC2 CPU groups in the NC3 that can be indicated by the second target routing iinformation of CPU reach the CPU groups that be connect with NC3 of NC2 path access Interior CPU.
That is, the NC that redirects when the instruction of third target routing iinformation is NC0, the jump of the 4th target routing iinformation instruction When to turn NC be NC1, if the 6th CPU groups 16 that are connect with NC2 of the first CPU needs access in the 8th CPU groups 18 connect with NC3 The 2nd interior CPU, and there is no message forwarding shortest path when, then, can be through third after message is sent to NC3 by the first CPU The instruction of target routing iinformation redirects NC-NC0, reaches target NC-NC2, is forwarded to the 2nd CPU through NC2, that is, message exists After reaching NC3, it can achieve the goal according to the path of the third target routing iinformation instruction configured in the routing register of NC3 CPU。
If the 2nd CPU in the 6th CPU groups 16 being connect with NC2 needs to access in the 8th CPU groups 18 connecting with NC3 First CPU, and there is no message forwarding shortest path when, then, can be through the 4th target after message is sent to NC2 by the 2nd CPU Routing iinformation instruction redirects NC-NC1, reaches target NC-NC3, is forwarded to the 2nd CPU through NC3, that is, message is reaching , can be according to the path of the 4th target routing iinformation instruction configured in the routing register of NC2 after NC2, achieve the goal CPU.
Equally, when the NC that redirects of third target routing iinformation instruction is NC1, the instruction of the 4th target routing iinformation redirects NC For NC0 when, if the first CPU in the 8th CPU groups 18 being connect with NC3 needs to access in the 6th CPU groups 16 connecting with NC2 2nd CPU, and there is no message forwarding shortest path when, then, can be through third target after message is sent to NC3 by the first CPU Routing iinformation instruction redirects NC-NC1, reaches target NC-NC2, is forwarded to the 2nd CPU through NC2, that is, message is reaching , can be according to the path of the third target routing iinformation instruction configured in the routing register of NC3 after NC3, achieve the goal CPU.
If the 2nd CPU in the 6th CPU groups 16 being connect with NC2 needs to access in the 8th CPU groups 18 connecting with NC3 First CPU, and there is no message forwarding shortest path when, then, can be through the 4th target after message is sent to NC2 by the 2nd CPU Routing iinformation instruction redirects NC-NC0, reaches target NC-NC3, is forwarded to the 2nd CPU through NC3, that is, message is reaching , can be according to the path of the 4th target routing iinformation instruction configured in the routing register of NC2 after NC2, achieve the goal CPU.
Above-mentioned elaboration is that " under CPU interconnection architectures shown in Fig. 5, first object NC is NC0, the second target NC to example When the link between link and NC3 and NC2 for NC3, and between NC0 and NC1 breaks down " corresponding step S201~ The specific implementation of step 213.
BMC configures the CPU that two NC are respectively connected by corresponding two NC of link for failure in the present embodiment The routing iinformation of exchanging visit, and the NC that redirects of the respective routing iinformations of corresponding two NC of same faulty link is differed, and is existed in this way When having ensured that link breaks down, while can still provide for the exchanging visit of corresponding CPU, is also not in resource contention, causes node The phenomenon that deadlock, reduces system failure odds.
Further, BMC is to determine the first routing configuration information and second according to dimension sequence routing algorithm in above-described embodiment Routing configuration information, it can specifically be realized based on following methods:Construction belongs to the virtual architecture of multiple NC of same NC groups, Virtual architecture is the position relationship of the virtual polygon and each NC of multiple NC compositions in virtual polygon.Such as:If one NC groups include 4 NC, then the virtual architecture of multiple NC of same NC groups is virtual quadrangle and 4 NC in virtual quadrangle Position;If it is, NC group includes N number of NC, the virtual architecture of multiple NC of same NC groups for the virtual sides N shape and Positions of N number of NC in virtual polygon.
Wherein, virtual quadrangle can be virtual rectangle, pseudo-parallelogram.
If a NC group includes 4 NC, dimension sequence routing algorithm is based on preset condition, and preset condition is when the first message reaches After corresponding first NC to be configured of faulty link, enable to the first NC to be configured by the first message along destination virtual rectangular horizontal Side on direction is sent to first and redirects NC, and first, which redirects NC, sends out side of first message on destination virtual rectangular vertical direction It send to the second NC (the corresponding another NC of faulty link) to be configured, and after the second message reaches the second NC to be configured, it can So that side of second message on destination virtual rectangular horizontal direction is sent to second by the second NC to be configured redirects NC, second jumps Turn NC and side of second message on destination virtual rectangular vertical direction is sent to the first NC to be configured;
Alternatively, preset condition is after the first message arrival faulty link corresponding first NC to be configured, the is enabled to Side of the message on destination virtual rectangular vertical direction is sent to first and redirects NC by one NC to be configured, and first redirects NC by first Side of the message on destination virtual rectangular horizontal direction is sent to the second NC to be configured, and waits matching when the second message reaches second After setting NC, enable to the second NC to be configured that side of second message on destination virtual rectangular vertical direction is sent to the second jump Turn NC, second, which redirects NC, to be sent to the first NC to be configured by side of second message on destination virtual rectangular horizontal direction;
Wherein, the line of the first NC to be configured and the second NC to be configured are the diagonal line of destination virtual rectangle.
Fig. 7 is the virtual rectangle schematic diagram of NC provided by the embodiments of the present application compositions, and Fig. 8 is A and B compositions in Fig. 7 Virtual rectangle after link breaks down after transformation, Fig. 9 are the void after being converted after the link of the A and C compositions in Fig. 7 break down Quasi-moments shape.
Referring to Fig. 7, A, B, C, D in Fig. 7 are four NC, form virtual rectangle.When being not at diagonal line in virtual rectangle On two NC compositions link when breaking down, for example link between A and B breaks down, and B is moved to the diagonal line of A Position on, be transformed into new rectangle, that is, destination virtual rectangle, as shown in Figure 8.At this point, being calculated based on the routing of above-mentioned dimension sequence The path for the corresponding first routing iinformation instruction of the first routing configuration information that method obtains can be that the path of A to B is A-D-B, base It is in the path of the B to A for the corresponding secondary route information instruction of secondary route configuration information that above-mentioned dimension sequence routing algorithm obtains B-C-A;Alternatively, based on the corresponding first routing iinformation instruction of the first routing configuration information that above-mentioned dimension sequence routing algorithm obtains Path can be that the path of A to B is A-C-B, the secondary route configuration information corresponding obtained based on above-mentioned dimension sequence routing algorithm The path of the B to A of two routing iinformations instruction is B-D-A.
If the link of the A and D compositions in Fig. 7 break down, since A and D is natively in diagonal positions, so nothing The transformation of quadrangle need to be carried out, then rectangle shown in Fig. 7 is destination virtual rectangle.It is obtained based on above-mentioned dimension sequence routing algorithm The path of the corresponding first routing iinformation instruction of first routing configuration information can be that the path of A to D is A-B-D, be based on above-mentioned dimension The path of the D to A for the corresponding secondary route information instruction of secondary route configuration information that sequence routing algorithm obtains is D-C-A;Or The path of person, the corresponding first routing iinformation instruction of the first routing configuration information obtained based on above-mentioned dimension sequence routing algorithm can be The path of A to D is A-C-D, the corresponding secondary route letter of secondary route configuration information obtained based on above-mentioned dimension sequence routing algorithm The path for ceasing the D to A of instruction is D-B-A.
If the link of the A and C compositions in Fig. 7 break down, C is moved on cornerwise position of A, is transformed into new Rectangle, that is, destination virtual rectangle, as shown in figure 9, the first routing configuration letter obtained based on above-mentioned dimension sequence routing algorithm The path for ceasing corresponding first routing iinformation instruction can be that the path of A to C is A-B-C, be obtained based on above-mentioned dimension sequence routing algorithm The path of C to A of secondary route configuration information corresponding secondary route information instruction be C-D-A;Alternatively, being based on above-mentioned dimension sequence The path for the corresponding first routing iinformation instruction of the first routing configuration information that routing algorithm obtains can be that the path of A to C is A- D-C, the C of the corresponding secondary route information instruction of secondary route configuration information obtained based on above-mentioned dimension sequence routing algorithm is to A's Path is C-B-A.
The case where breaking down for each link where other NC other than A, with reference to the above method, herein no longer It repeats one by one.
That is:For destination virtual quadrangle, corresponding two NC of faulty link are in destination virtual quadrangle Diagonal positions are the routing iinformation of two NC configurations of faulty link it is required that the one of NC of message arrival is forwarded to During another NC, transmission rule on virtual quadrangle after the conversion reaches another NC with message and is forwarded to this During a wherein NC, the transmission rule on virtual quadrangle after the conversion is identical, that is, the jump based on dimension sequence routing Walk algorithm so that the corresponding NC that redirects in path that the CPU that corresponding two NC of faulty link are respectively connected exchanges visits is differed.In this way Also the purpose of the generation of the case where resource contention that can avoid the occurrence of being achieved that in above-described embodiment, node deadlock.
It is illustrated so that a NC group includes the CPU interconnection architectures of 4 NC as an example in the various embodiments described above, certainly, this Application embodiment is not intended to limit the number for the NC that a NC group includes, such as:It may include 5 NC in one NC group, need to only expire The corresponding NC that redirects in path that the CPU that corresponding two NC of sufficient faulty link are respectively connected exchanges visits is differed.
Figure 10 is the structural schematic diagram one of the routing management device of Node Controller provided by the embodiments of the present application, referring to figure 10, the device of the present embodiment includes:Judgment module 21, sending module 22 and determining module 23
Wherein, Node Controller NC groups include multiple NC, and the NC in same NC groups is connected two-by-two, and each CPU groups include more Each NC in a CPU, NC group is at least connected with a CPU group, and each CPU in same group is connected two-by-two or annular connection.
Judgment module 21 judges the mesh for the status information according to the target NC being connect with the baseboard controller Whether each link where mark NC breaks down;The status information is used to indicate the work of each link where the target NC State;
Sending module 22, if for there are the faulty links of a cisco unity malfunction, for the faulty link pair The target NC answered and NC to be configured sends the first routing configuration information to the target NC, sends secondary route and match confidence For breath to the NC to be configured, the first routing configuration information obtains the NC to the first of the NC to be configured for the target NC Routing iinformation, secondary route configuration information obtain the NC to be configured to the second tunnel of the target NC for the NC to be configured By information, the NC that redirects that the first routing iinformation and secondary route information respectively indicate is differed.
First routing iinformation and the secondary route information include the enabled mark of routing, redirect the mark and mesh of NC NC mark;The purpose NC for stating the instruction of the first routing iinformation is the NC to be configured, the mesh of the secondary route information instruction NC be the target NC.
Determining module 23, is used for:According to dimension sequence routing algorithm, determine that first routing configuration information and secondary route are matched Confidence ceases.
Device provided by the embodiments of the present application can execute above-mentioned corresponding embodiment of the method, realization principle and technology Effect is similar, and details are not described herein.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle, the dimension sequence routing algorithm is based on pre- If condition;
The preset condition is to enable to the target NC by described first after the first message reaches the target NC Side of the message in the virtual rectangle horizontal direction is sent to first and redirects NC, and described first redirects NC by first message Side in the virtual rectangle vertical direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle horizontal direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle vertical direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, it is described When four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle, the dimension sequence routing algorithm is based on pre- If condition;
The preset condition is to enable to the target NC by described first after the first message reaches the target NC Side of the message in the virtual rectangle vertical direction is sent to first and redirects NC, and described first redirects NC by first message Side in the virtual rectangle horizontal direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message edge Side in the virtual rectangle vertical direction is sent to second and redirects NC, and described second redirects NC will be described in second message edge Side in virtual rectangle horizontal direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
In a kind of possible design, when the NC groups where the target NC and the NC to be configured include 4 NC, if The NC that redirects of the first routing iinformation instruction is the first NC, then the NC that redirects of the secondary route information instruction is the 2nd NC, If the NC that redirects of the first routing iinformation instruction is the 2nd NC, the NC that redirects of the secondary route information instruction is first NC;
First NC is any NC other than the target NC and the NC to be configured in 4 NC, described 2nd NC is any NC other than the target NC and the NC to be configured in 4 NC.
In a kind of possible design, the judgment module 21 is specifically used for:
For each association NC being connect with the target NC, if being used to indicate in the status information of the target NC described Target NC and the working condition of the link of the association NC compositions are identified as first identifier, it is determined that the target NC with it is described Link between association NC can work normally, if being used to indicate the work of the target NC and the link of the association NC compositions State is identified as second identifier, it is determined that the link failure between the target NC and the association NC.
Device provided by the embodiments of the present application can execute above-mentioned corresponding embodiment of the method, realization principle and technology Effect is similar, and details are not described herein.
Figure 11 is the structural schematic diagram two of the routing management device of Node Controller provided by the embodiments of the present application, referring to figure 11, the device of the present embodiment includes:Sending module 31 and routing information acquisition module 32.
Wherein, Node Controller NC groups include multiple NC, and multiple NC in NC groups are connected two-by-two, and each CPU groups include more Each NC in a CPU, NC group is at least connected with a CPU group, and each CPU in same group is connected two-by-two or annular connection.
Sending module 31, for sending status information to first substrate controller BMC, the status information is used to indicate mesh Mark the working condition of each link where NC;
Routing information acquisition module 32 connects if for the first link failure in the multilink where the target NC The first routing configuration information that the BMC is sent is received, and according to first routing configuration information, obtains the target NC to institute State the first routing iinformation of the corresponding another NC of the first link;Wherein, what is indicated in secondary route information redirects NC and described the The NC that redirects of one routing iinformation instruction is differed, the secondary route information is another NC sent according to the 2nd BMC the What two routing configuration informations obtained.
First routing iinformation and the secondary route information include the enabled mark of routing, redirect the mark and mesh of NC NC mark;The purpose NC of the first routing iinformation instruction is another NC, the mesh of the secondary route information instruction NC be the target NC.
Device provided by the embodiments of the present application can execute above-mentioned corresponding embodiment of the method, realization principle and technology Effect is similar, and details are not described herein.
The embodiment of the present application also provides a kind of computer readable storage medium, and the computer-readable recording medium storage has Computer program, the computer program make processor execute the method that BMC is executed in above method embodiment.
Figure 12 is the structural schematic diagram of baseboard controller provided by the embodiments of the present application, including:Memory 41, processor 42 With communication bus 43;Communication bus 43 is for realizing the connection between each component.
The memory 41, for storing program instruction;
The processor 42, for calling the described program stored in the memory instruction to realize that the above method is implemented The method that BMC is executed in example.
Baseboard controller provided by the embodiments of the present application can execute above-mentioned corresponding embodiment of the method, realization principle Similar with technique effect, details are not described herein.
The embodiment of the present application also provides a kind of computer readable storage medium, and the computer-readable recording medium storage has Computer program, the computer program make processor execute the method that NC is executed in above method embodiment.
Figure 13 is the structural schematic diagram of Node Controller provided by the embodiments of the present application, including:Memory 51, processor 52 With communication bus 53;Communication bus 53 is for realizing the connection between each component.
The memory 51, for storing program instruction;
The processor 52, for calling the described program stored in the memory instruction to realize that the above method is implemented The method that NC is executed in example.
Node Controller provided by the embodiments of the present application can execute above-mentioned corresponding embodiment of the method, realization principle Similar with technique effect, details are not described herein.

Claims (22)

1. a kind of route management method of Node Controller, Node Controller NC groups include multiple NC, the NC two in same NC groups Two connections, which is characterized in that including:
Baseboard controller is according to the status information of the target NC being connect with the baseboard controller, where judging the target NC Whether each link breaks down;The status information is used to indicate the working condition of each link where the target NC;
If there are the faulty link of a cisco unity malfunction, for the corresponding target NC of the faulty link and wait matching NC is set, the baseboard controller sends the first routing configuration information to the target NC, sends secondary route configuration information to institute NC to be configured is stated, the first routing configuration information obtains the target NC to the first via of the NC to be configured for the target NC By information, secondary route configuration information obtains the NC to be configured to the secondary route of the target NC for the NC to be configured The NC that redirects that information, the first routing iinformation and secondary route information respectively indicate is differed.
2. according to the method described in claim 1, it is characterized in that, it is described for the corresponding target NC of faulty link and NC to be configured, the baseboard controller send the first routing configuration information to the target NC, send secondary route configuration information Before to the NC to be configured, the method further includes:
The baseboard controller determines that first routing configuration information and secondary route match confidence according to dimension sequence routing algorithm Breath.
3. according to the method described in claim 2, it is characterized in that, when the NC groups where the target NC and NC to be configured Including 4 NC, when four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle;
The dimension sequence routing algorithm is based on preset condition, and the preset condition is energy after the first message reaches the target NC Enough so that side of first message in the virtual rectangle horizontal direction is sent to first by the target NC redirects NC, institute It states first and redirects NC and side of first message in the virtual rectangle vertical direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message described in Side in virtual rectangle horizontal direction is sent to second and redirects NC, and described second redirects NC by second message along described virtual Side on rectangular vertical direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
4. according to the method described in claim 2, it is characterized in that, when the NC groups where the target NC and NC to be configured Including 4 NC, when four vertex that 4 NC form virtual rectangle and 4 NC are the virtual rectangle;
The dimension sequence routing algorithm is based on preset condition, and the preset condition is energy after the first message reaches the target NC Enough so that side of first message in the virtual rectangle vertical direction is sent to first by the target NC redirects NC, institute It states first and redirects NC and side of first message in the virtual rectangle horizontal direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message described in Side in virtual rectangle vertical direction is sent to second and redirects NC, and described second redirects NC by second message along described virtual Side on rectangular horizontal direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
5. method according to any one of claims 1 to 4, which is characterized in that as the target NC and the NC institutes to be configured NC groups include 4 NC when, if first routing iinformation instruction redirect NC be the first NC, the secondary route information The NC that redirects indicated is the 2nd NC, if the NC that redirects of first routing iinformation instruction is the 2nd NC, the secondary route letter The NC that redirects of breath instruction is the first NC;
First NC is any NC other than the target NC and the NC to be configured in 4 NC, described second NC is any NC other than the target NC and the NC to be configured in 4 NC.
6. method according to any one of claims 1 to 4, which is characterized in that the status information according to NC, described in judgement Whether each link where target NC breaks down, including:
For each association NC being connect with the target NC, if being used to indicate the target in the status information of the target NC NC is identified as first identifier with the working condition of the link of the association NC compositions, it is determined that the target NC and the association Link between NC can work normally, if being used to indicate the working condition of the target NC and the link of the association NC compositions Be identified as second identifier, it is determined that the target NC and it is described association NC between link failure.
7. method according to any one of claims 1 to 4, which is characterized in that first routing iinformation and second tunnel Include the enabled mark of routing by information, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is the NC to be configured, and the purpose NC of the secondary route information instruction is The target NC.
8. a kind of route management method of Node Controller, Node Controller NC groups include multiple NC, multiple NC two in NC groups Two connections, which is characterized in that including:
Target NC sends status information to first substrate controller BMC, and the status information is used to indicate where the target NC Each link working condition;
If the first link failure in multilink where the target NC, the first routing configuration that the BMC is sent is received Information, and according to first routing configuration information, obtain the of the target NC to the corresponding another NC of first link One routing iinformation;Wherein, the NC that redirects indicated in secondary route information redirects NC not phases with what first routing iinformation indicated Together, the secondary route information is that another NC is obtained according to the 2nd BMC secondary route configuration informations sent.
9. according to the method described in claim 8, it is characterized in that, first routing iinformation and the secondary route information are equal Including routeing enabled mark, the mark of the mark and purpose NC of NC is redirected;
The purpose NC of the first routing iinformation instruction is another NC, and the purpose NC of the secondary route information instruction is institute State target NC.
10. a kind of routing management device of Node Controller, Node Controller NC groups include multiple NC, the NC two in same NC groups Two connections, which is characterized in that including:
Judgment module judges the target NC institutes for the status information according to the target NC being connect with the baseboard controller Each link whether break down;The status information is used to indicate the working condition of each link where the target NC;
Sending module, if for there are the faulty link of a cisco unity malfunction, institutes corresponding for the faulty link Target NC and NC to be configured are stated, the first routing configuration information is sent to the target NC, sends secondary route configuration information to institute NC to be configured is stated, the first routing configuration information obtains the NC to the first via of the NC to be configured by believing for the target NC Breath, secondary route configuration information obtain the secondary route letter of the NC to be configured to the target NC for the NC to be configured The NC that redirects that breath, the first routing iinformation and secondary route information respectively indicate is differed.
11. device according to claim 10, which is characterized in that described device further includes determining module, is used for:
According to dimension sequence routing algorithm, first routing configuration information and secondary route configuration information are determined.
12. according to the devices described in claim 11, which is characterized in that the NC where the target NC and the NC to be configured Group includes 4 NC, 4 NC form virtual rectangle and 4 NC for the virtual rectangle four vertex when, the dimension Sequence routing algorithm is based on preset condition;
The preset condition is to enable to the target NC by first message after the first message reaches the target NC Side in the virtual rectangle horizontal direction is sent to first and redirects NC, and described first redirects NC by first message along institute It states the side in virtual rectangle vertical direction and is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message described in Side in virtual rectangle horizontal direction is sent to second and redirects NC, and described second redirects NC by second message along described virtual Side on rectangular vertical direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
13. according to the devices described in claim 11, which is characterized in that the NC where the target NC and the NC to be configured Group includes 4 NC, 4 NC form virtual rectangle and 4 NC for the virtual rectangle four vertex when, the dimension Sequence routing algorithm is based on preset condition;
The preset condition is to enable to the target NC by first message after the first message reaches the target NC Side in the virtual rectangle vertical direction is sent to first and redirects NC, and described first redirects NC can be by first message Side in the virtual rectangle horizontal direction is sent to the NC to be configured;
And after the second message reaches the NC to be configured, enable to the NC to be configured by second message described in Side in virtual rectangle vertical direction is sent to second and redirects NC, and described second redirects NC by second message along described virtual Side on rectangular horizontal direction is sent to the target NC;
Wherein, the line of the target NC and the NC to be configured are the diagonal line of the virtual rectangle.
14. according to any device of claim 10~13, which is characterized in that as the target NC and the NC to be configured When the NC groups at place include 4 NC, if the NC that redirects of first routing iinformation instruction is the first NC, the secondary route letter The NC that redirects of breath instruction is the 2nd NC, if the NC that redirects of first routing iinformation instruction is the 2nd NC, the secondary route The NC that redirects of information instruction is the first NC;
First NC is any NC other than the target NC and the NC to be configured in 4 NC, described second NC is any NC other than the target NC and the NC to be configured in 4 NC.
15. according to any device of claim 10~13, which is characterized in that the judgment module is specifically used for:
For each association NC being connect with the target NC, if being used to indicate the target in the status information of the target NC NC is identified as first identifier with the working condition of the link of the association NC compositions, it is determined that the target NC and the association Link between NC can work normally, if being used to indicate the working condition of the target NC and the link of the association NC compositions Be identified as second identifier, it is determined that the target NC and it is described association NC between link failure.
16. according to any device of claim 10~13, which is characterized in that first routing iinformation and described second Routing iinformation includes the enabled mark of routing, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is the NC to be configured, and the purpose NC of the secondary route information instruction is The target NC.
17. a kind of routing management device of Node Controller, Node Controller NC groups include multiple NC, multiple NC two in NC groups Two connections, which is characterized in that including:
Sending module, for sending status information to first substrate controller BMC, the status information is used to indicate target NC institutes Each link working condition;
Routing information acquisition module, if for the first link failure in the multilink where the target NC, described in reception The first routing configuration information that BMC is sent, and according to first routing configuration information, obtain the target NC to described first The first routing iinformation of the corresponding another NC of link;Wherein, indicated in secondary route information redirect NC and the first via by The NC that redirects of information instruction is differed, and the secondary route information is the secondary route that another NC is sent according to the 2nd BMC What configuration information obtained.
18. device according to claim 17, which is characterized in that first routing iinformation and the secondary route information Include the enabled mark of routing, redirects the mark of the mark and purpose NC of NC;
The purpose NC of the first routing iinformation instruction is another NC, and the purpose NC of the secondary route information instruction is institute State target NC.
19. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey Sequence, the computer program make processor perform claim require 1~7 any one of them method.
20. a kind of baseboard controller, which is characterized in that including:Memory and processor;
The memory, for storing program instruction;
The processor, for calling the described program stored in the memory instruction to realize as claim 1~7 is any Method described in.
21. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey Sequence, the computer program make processor perform claim require the method described in 8 or 9.
22. a kind of Node Controller, which is characterized in that including:Memory and processor;
The memory, for storing program instruction;
The processor, for calling the described program stored in the memory instruction to realize as described in claim 8 or 9 Method.
CN201810262549.6A 2018-03-28 2018-03-28 Routing management method and device of node controller Active CN108632142B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810262549.6A CN108632142B (en) 2018-03-28 2018-03-28 Routing management method and device of node controller

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810262549.6A CN108632142B (en) 2018-03-28 2018-03-28 Routing management method and device of node controller

Publications (2)

Publication Number Publication Date
CN108632142A true CN108632142A (en) 2018-10-09
CN108632142B CN108632142B (en) 2021-02-12

Family

ID=63696560

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810262549.6A Active CN108632142B (en) 2018-03-28 2018-03-28 Routing management method and device of node controller

Country Status (1)

Country Link
CN (1) CN108632142B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2001024007A2 (en) * 1999-09-30 2001-04-05 Silicon Graphics, Inc. Method and apparatus for processing errors in a computer system
CN101479704A (en) * 2006-03-27 2009-07-08 相干逻辑公司 Programming a multi-processor system
CN102232218A (en) * 2011-06-24 2011-11-02 华为技术有限公司 Computer subsystem and computer system
CN102449621A (en) * 2011-08-25 2012-05-09 华为技术有限公司 Node controller chain switching method, processor system, and nodes
CN105808499A (en) * 2016-04-01 2016-07-27 浪潮电子信息产业股份有限公司 CPU interconnection device and multichannel server CPU interconnection topological structure
CN107111572A (en) * 2014-10-31 2017-08-29 赛灵思公司 Method and circuit for avoiding deadlock

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2001024007A2 (en) * 1999-09-30 2001-04-05 Silicon Graphics, Inc. Method and apparatus for processing errors in a computer system
CN101479704A (en) * 2006-03-27 2009-07-08 相干逻辑公司 Programming a multi-processor system
CN102232218A (en) * 2011-06-24 2011-11-02 华为技术有限公司 Computer subsystem and computer system
CN102449621A (en) * 2011-08-25 2012-05-09 华为技术有限公司 Node controller chain switching method, processor system, and nodes
CN107111572A (en) * 2014-10-31 2017-08-29 赛灵思公司 Method and circuit for avoiding deadlock
CN105808499A (en) * 2016-04-01 2016-07-27 浪潮电子信息产业股份有限公司 CPU interconnection device and multichannel server CPU interconnection topological structure

Also Published As

Publication number Publication date
CN108632142B (en) 2021-02-12

Similar Documents

Publication Publication Date Title
CN1980230B (en) Method for managing VRRP group
KR100932585B1 (en) Method, device and system for sending messages
CN101427549B (en) Virtual inline configuration for a network device
US8139490B2 (en) Deadlock prevention in direct networks of arbitrary topology
CN100583819C (en) Integrated circuit and method for packet switching control
EP3226490B1 (en) Optical network-on-chip, optical router and signal transmission method
CN103188157B (en) A kind of router equipment
JP2004062535A (en) Method of dealing with failure for multiprocessor system, multiprocessor system and node
CN112422422B (en) Network redundancy method, device and equipment
US20190302742A1 (en) Method for Setting Up a Redundant Communication Connection, and Failsafe Control Unit
JP6809360B2 (en) Information processing equipment, information processing methods and programs
CN102611613A (en) Adaptive multi-redundant ring network system and method for selecting detour
CN103036701B (en) The N+1 alternative mean of a kind of cross-network segment and device
CN113992577B (en) Network intercommunication method, device, equipment and medium
CN102932250B (en) Non-deadlock self-adaptation routing method based on fault-tolerant computer network structure
CN102123080B (en) Data forwarding device and cable fastener plate and data forwarding method
CN105610555B (en) A kind of practical system-level redundancy communication network framework
US20190268235A1 (en) Method for managing network nodes and communication control method thereof
CN105763488A (en) Data center gathering core switcher and backboard thereof
WO2014075594A1 (en) Service transmission protection method and device based on intersecting ring of multi-ring structure network
CN108632142A (en) The route management method and device of Node Controller
US9524259B2 (en) Method for operating an automation device to reduce dead time on account of a physical interruption in a ring or a failed unit
CN104348737A (en) Multicast message transmission method and switches
RU2609074C2 (en) Noise-immune datagram transmission in communication network
CN114079630B (en) Service protection method, device, equipment and storage medium based on SPN (service provider network)

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right

Effective date of registration: 20211223

Address after: 450046 Floor 9, building 1, Zhengshang Boya Plaza, Longzihu wisdom Island, Zhengdong New Area, Zhengzhou City, Henan Province

Patentee after: xFusion Digital Technologies Co., Ltd.

Address before: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen

Patentee before: HUAWEI TECHNOLOGIES Co.,Ltd.

TR01 Transfer of patent right