Network fault analysis method, device, equipment and storage medium

By acquiring and analyzing the abnormal event information of network equipment, determining the fault cause description information and considering the event identification and cause relationship, the problem of low accuracy of network failure analysis in the prior art is solved, and more refined fault propagation relationship determination and faulting are achieved. position.

CN115277357BActive Publication Date: 2024-11-15HUAWEI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110482854.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-04-30
Publication Date
2024-11-15
Estimated Expiration
2041-04-30

AI Technical Summary

Technical Problem

The existing network fault analysis methods are less accurate because the same abnormal event may be caused by different reasons, resulting in inaccurate analysis of the fault cause.

Method used

By obtaining information about abnormal events, including event identification and event description information, determining the fault cause description information, and determining the fault propagation direction based on this information, considering the event identification and cause relationship, improving the determination result of the fault propagation relationship.

Benefits of technology

It improves the accuracy of network fault analysis, making the determination of fault propagation relationship more precise and accurate, and can more accurately locate the cause of faults.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115277357B_ABST
    Figure CN115277357B_ABST
Patent Text Reader

Abstract

The present application discloses a network fault analysis method, device, equipment and storage medium, belonging to the field of network technology. The method includes: obtaining information of a first abnormal event and information of a second abnormal event; determining first fault cause description information and second fault cause description information according to the information of the first abnormal event and the information of the second abnormal event, respectively, the first and second fault cause description information are used to describe the cause of the corresponding abnormal event; based on the event identifier in the information of the first and second abnormal events, the first and second fault cause description information, determining that the first abnormal event corresponding to the first fault cause description information is the cause event that caused the second abnormal event corresponding to the second fault cause description information to occur. When performing network fault analysis, this method combines the identifier of the abnormal event and the corresponding fault cause description information to determine the cause event in the first and second abnormal events, which can improve the accuracy of the network fault analysis results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of network technology, and in particular to a network fault analysis method, device, equipment and storage medium. Background Art

[0002] A network failure refers to a state in which the network cannot provide normal services or the service quality is reduced due to hardware problems, software problems, network attacks, etc. When a network failure occurs, the network device will report abnormal event information, such as alarms and logs. The cause of the failure can be determined by analyzing the network failure based on the abnormal event information reported by the network device.

[0003] In the related art, abnormal event information at least includes an abnormal event identifier. Network fault analysis usually first classifies abnormal events with the same identifier into the same type of abnormal events, and then analyzes the association between abnormal events of different types to obtain the cause of the fault.

[0004] However, abnormal events with the same identifier may be caused by different reasons. Therefore, the accuracy of the network fault analysis results obtained in this way is low. Summary of the invention

[0005] The embodiments of the present application provide a network fault analysis method, apparatus, device and storage medium, which can improve the accuracy of network fault analysis results.

[0006] In a first aspect, the present application provides a network fault analysis method, the method comprising: first, obtaining information of a first abnormal event and information of a second abnormal event, wherein the information of the first abnormal event and the information of the second abnormal event both include event identifiers and event description information, etc. Then, determining first fault cause description information according to the first abnormal event and determining second fault description information according to the second abnormal event. The first fault cause description information is used to describe the cause of the first abnormal event; the second fault cause description information is used to describe the cause of the second abnormal event. Finally, based on the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, determining the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information. The fault propagation direction is used to indicate the cause event and the effect event in the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information.

[0007] In an embodiment of the present application, first fault cause description information for describing the cause of the occurrence of the first abnormal event is determined based on information about the first abnormal event, and second fault cause description information for describing the cause of the occurrence of the second abnormal event is determined based on information about the second abnormal event. When considering the fault propagation relationship between the first abnormal event and the second abnormal event, not only the identifier of the first abnormal event and the identifier of the second abnormal event are considered, but also the causes of the occurrence of the first abnormal event and the second abnormal event are considered, so as to determine that the first abnormal event corresponding to the first fault cause description information is the cause of the occurrence of the second abnormal event corresponding to the second fault cause description information, so that the determination result of the fault propagation relationship is more refined and accurate, thereby improving the accuracy of network fault analysis.

[0008] In a possible implementation, the first and second fault cause description information are both determined by using an extraction template corresponding to the network device that sends the corresponding abnormal event information. The extraction template corresponding to each network device includes event identifiers, fault code keywords, and other content. In some examples, the extraction template also includes candidate fault codes corresponding to the fault code keywords and fault code description information for each of the candidate fault codes. A candidate fault code refers to a fault code that a fault keyword may correspond to, and a fault keyword usually corresponds to multiple candidate fault codes.

[0009] In some examples, the process of determining the first fault cause description information is as follows. First, a first extraction template of a first network device is obtained, where the first network device is a network device that sends the information of the first abnormal event. Then, based on the same event identifier and fault code keywords in the first extraction template and the information of the first abnormal event, a fault code is extracted from the information of the first abnormal event to obtain a first extraction result. Finally, based on the first extraction result, the first fault cause description information is determined.

[0010] In some examples, the second fault cause description information is determined as follows. First, a second extraction template of a second network device is obtained, where the second network device is a network device that sends the second abnormal event information. Then, based on the same event identifier and fault code keyword in the second extraction template and the second abnormal event information, a fault code is extracted from the second abnormal event information to obtain a second extraction result. Finally, based on the second extraction result, the second fault cause description information is determined.

[0011] In this way, the first extraction template and the second extraction template can automatically extract the fault code in the information of the first abnormal event and the information of the second abnormal event, and determine the fault cause description information corresponding to the extracted fault code.

[0012] In a possible implementation manner, at least one of the first extraction result and the second extraction result includes at least one fault code.

[0013] In a possible implementation, the determining the first fault cause description information according to the first extraction result includes: when the first extraction result includes a fault code, combining the fault code included in the first extraction result with a corresponding fault code keyword, each combination is a first fault cause description information, thereby obtaining at least one first fault cause description information. When the first extraction result does not include a fault code, using the event identifier of the first abnormal event as the first fault cause description information, or determining that the first fault cause description information is empty.

[0014] The manner of determining the second fault cause description information according to the second extraction result is the same as the manner of determining the first fault cause description information according to the first extraction result.

[0015] In a possible implementation manner, the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information based on the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information includes: obtaining a first fault propagation relationship model, the first fault propagation relationship model includes multiple first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information; in response to determining that there is a first target fault propagation relationship in the first fault propagation relationship model, according to the first target fault propagation relationship, the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information, the first target fault propagation relationship being a first fault propagation relationship including the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0016] By comparing the identifier of the first abnormal event, the identifier of the second abnormal event, the first fault cause description information and the second fault cause description information with the first fault propagation relationships in the first fault propagation relationship model, the first target fault propagation relationship including the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event and the first and second fault cause description information is determined, thereby determining the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information, which is convenient and simple to implement.

[0017] In another possible implementation manner, if the first target fault propagation relationship does not exist in the first fault propagation relationship model, then determining the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information based on the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information includes: acquiring a second fault propagation relationship model, the second fault propagation relationship model includes multiple second fault propagation relationships, the second fault propagation relationship is used to record possible fault propagation directions of two abnormal events under the corresponding fault cause description information, and the second fault propagation relationship model is generated according to the first fault propagation relationship model; in response to determining that the second target fault propagation relationship exists in the second fault propagation relationship model, according to the second target fault propagation relationship, the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information, the second target fault propagation relationship being a second fault propagation relationship including the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0018] When the fault propagation relationship between the first abnormal event and the second abnormal event based on the corresponding fault cause description information cannot be determined according to the first fault propagation relationship model, the possible fault propagation relationship between the first abnormal event and the second abnormal event based on the corresponding fault cause description information is determined by the second fault propagation relationship model to provide a reference for locating the cause of the fault.

[0019] In a possible implementation, the method further includes: after determining the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship, verifying the second target fault propagation relationship; and updating the first fault propagation relationship model and the second fault propagation relationship model according to the verified second target fault propagation relationship.

[0020] After determining that the possible fault propagation relationship between the first abnormal event and the second abnormal event has passed verification, updating the first fault propagation relationship model and the second fault propagation relationship model can make the first fault propagation relationship model more perfect, while improving the accuracy of each second fault propagation relationship in the second fault propagation relationship model, so that the subsequent propagation relationship determination results are more accurate.

[0021] In some examples, the network management device sends the second target fault propagation relationship to the server, the cloud device verifies the second target fault propagation relationship, and updates the first fault propagation relationship model and the second fault propagation relationship model based on the verified second target fault propagation relationship, and then sends the updated first fault propagation relationship model and the second fault propagation relationship model to the network management device.

[0022] Optionally, the first fault propagation relationship in the first fault propagation relationship model is generated in the following manner: determine a third abnormal event on the first entity and third fault cause description information used to describe the cause of the occurrence of the third abnormal event, and a fourth abnormal event on the second entity and fourth fault cause description information used to describe the cause of the occurrence of the fourth abnormal event, the first entity and the second entity are associated with each other; according to expert annotation, determine the cause event and the effect event in the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information, and obtain the first fault propagation relationship.

[0023] Optionally, the second fault propagation relationship in the second fault propagation relationship model is generated in the following manner: determine a fifth abnormal event with unknown fault cause description information, wherein the unknown fault cause description information is the fault cause description information of the corresponding abnormal event other than the known fault cause description information, and the known fault cause description information is the fault cause description information in the first fault propagation relationship associated with the corresponding abnormal event; calculate the first similarity between the unknown fault cause description information of the fifth abnormal event and each of the known fault cause description information of the fifth abnormal event, respectively, to obtain a first similarity set; and determine, based on the first similarity set, the probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established.

[0024] Here, the fifth abnormal event is any abnormal event with unknown fault cause description information. In some examples, the fifth abnormal event has known fault description information present in the first fault propagation relationship model and unknown fault description information present in the corresponding extraction template but not in the first fault propagation relationship model.

[0025] Optionally, the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information includes two types. The first type is the second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event; the second type is the second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event.

[0026] For the first type, when there is a first similarity greater than a first similarity threshold in the first similarity set, the probability that the second fault propagation relationship is established is determined according to the first similarity greater than the first similarity threshold in the first similarity set and the second similarity greater than the second similarity threshold in the second similarity set, and the second similarity set is a set of second similarities between the unknown fault cause description information of the sixth abnormal event and each known fault cause description information of the sixth abnormal event. When the first similarities in the first similarity set are not greater than the first similarity threshold, the probability that the second fault propagation relationship is established is determined according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event.

[0027] For the second type, when there is a first similarity greater than a first similarity threshold in the first similarity set, the probability that the second fault propagation relationship is established is determined according to the first similarities greater than the first similarity threshold in the first similarity set. When all the first similarities in the first similarity set are not greater than the first similarity threshold, the probability that the second fault propagation relationship is established is determined according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event.

[0028] In the embodiment of the present application, when the fault code in the fault cause description information is in a string format, the similarity between the unknown fault cause description information and the known fault cause description information is the similarity between the fault codes in the string format. When the fault code in the fault cause description information is in a non-string format, such as a long integer, the similarity between the unknown fault cause description information and the known fault cause description information is the similarity between the fault code description information corresponding to the fault code.

[0029] In some examples, the probability that the second fault propagation relationship is established includes: at least one of a first probability and a second probability, wherein the first probability is the probability that the fifth abnormal event causes the sixth abnormal event, and the second probability is the probability that the sixth abnormal event causes the fifth abnormal event.

[0030] In some examples, the first probability is calculated based on the first similarity in the first similarity set that is greater than the first similarity threshold and the second similarity in the second similarity set that is greater than the second similarity threshold, including: determining a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to the first similarity greater than the first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to the second similarity greater than the second similarity threshold, to obtain a first relationship set; determining a first weight corresponding to each first fault propagation relationship in the first relationship set based on the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; determining a second weight corresponding to the first fault propagation relationship in the first relationship set that indicates that the fifth abnormal event causes the sixth abnormal event based on the first similarity and the second similarity corresponding to the first fault propagation relationship in the first relationship set that indicates that the fifth abnormal event causes the sixth abnormal event; and calculating the first probability based on the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the second weights.

[0031] In some examples, the second probability is calculated based on the first similarity in the first similarity set that is greater than the first similarity threshold and the second similarity in the second similarity set that is greater than the second similarity threshold, including: determining a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to the first similarity greater than the first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to the second similarity greater than the second similarity threshold, to obtain a first relationship set; determining a first weight corresponding to each first fault propagation relationship in the first relationship set based on the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; determining a fifth weight corresponding to the first fault propagation relationship in the first relationship set that indicates that the sixth abnormal event causes the fifth abnormal event based on the first similarity and the second similarity corresponding to the first fault propagation relationship in the first relationship set that indicates that the fifth abnormal event causes the sixth abnormal event; and calculating the second probability based on the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the fifth weights.

[0032] In some examples, the determining, based on the first similarity greater than the first similarity threshold in the first similarity set, the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event is established includes: determining the first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to the first similarity greater than the first similarity threshold to obtain a second relationship set; determining, based on the first similarity corresponding to each first fault propagation relationship in the second relationship set, a first weight corresponding to each first fault propagation relationship in the second relationship set; determining, based on the first similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the second relationship set, a second weight corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event; and calculating the probability that the fifth abnormal event causes the sixth abnormal event based on the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the second weights.

[0033] In some examples, the determining, based on the first similarity greater than the first similarity threshold in the first similarity set, the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event is established includes: determining the first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to the first similarity greater than the first similarity threshold to obtain a second relationship set; determining, based on the first similarity corresponding to each first fault propagation relationship in the second relationship set, a first weight corresponding to each first fault propagation relationship in the second relationship set; determining, based on the first similarity corresponding to the first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event in the second relationship set, a fifth weight corresponding to the first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event; and calculating, based on the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the fifth weights, the probability that the sixth abnormal event causes the fifth abnormal event.

[0034] Optionally, the method also includes: determining cause description keywords in the product manual of the target network device; extracting template content from the product manual of the target network device based on the cause description keywords; generating a target extraction template according to the template content; wherein, the target network device is the first network device, and the target extraction template is the first extraction template; or, the target network device is the second network device, and the target extraction template is the second extraction template.

[0035] In the second aspect, a network fault analysis method is provided. The method includes: determining entities associated with multiple abnormal events to obtain an entity set; using the method described in the first aspect, determining the fault propagation direction of the seventh abnormal event associated with the third entity and the eighth abnormal event associated with the fourth entity under the corresponding fault cause description information, wherein the third entity and the fourth entity are any two mutually associated entities in the entity set, the seventh abnormal event is any one of the at least one abnormal event associated with the third entity, and the eighth abnormal event is any one of the at least one abnormal event associated with the fourth entity; according to the determined fault propagation direction, determining the root cause event among the multiple abnormal events.

[0036] When a network failure occurs, multiple abnormal events will occur, and these multiple abnormal events are associated with multiple entities, and each entity is associated with at least one abnormal event. In this case, in addition to determining the causal relationship between any two abnormal events on two associated entities, it is also necessary to determine the root cause event in these multiple abnormal events to facilitate technical personnel to troubleshoot. The network failure method provided by the embodiment of the present application can accurately determine the root cause event in multiple abnormal events and realize network fault location.

[0037] In some examples, determining the root cause event among the multiple abnormal events based on the determined fault propagation direction includes: determining a reference value for each of the abnormal events based on the determined fault propagation direction, the reference value being the sum of the weights of the fault propagation directions of the result events to which the abnormal event belongs, the weights of the fault propagation directions being used to indicate the probability that the corresponding fault propagation direction is true; and determining the abnormal event with the smallest reference value as the root cause event among the multiple abnormal events.

[0038] Optionally, when the fault propagation direction is determined according to the first fault relationship model, the weight of the fault propagation direction is 1. When the fault propagation direction is determined according to the second fault propagation relationship model, the weight of the fault propagation direction is the probability corresponding to the corresponding second fault propagation relationship.

[0039] In a third aspect, a network fault analysis device is provided, the device comprising an acquisition module, a first determination module, and a second determination module. The acquisition module is used to acquire information about a first abnormal event and information about a second abnormal event. The first determination module is used to determine first fault cause description information based on information about the first abnormal event, and to determine second fault cause description information based on information about the second abnormal event, wherein the first fault cause description information is used to describe the cause of the occurrence of the first abnormal event, and the second fault cause description information is used to describe the cause of the occurrence of the second abnormal event. The second determination module is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information based on an event identifier in the information about the first abnormal event, an event identifier in the information about the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0040] In a possible implementation, the first determination module includes: a template acquisition submodule, an extraction submodule, and an information determination submodule. The template acquisition submodule is used to acquire a first extraction template of a first network device and a second extraction template of a second network device, wherein the first network device is a network device that sends information about the first abnormal event, and the second network device is a network device that sends information about the second abnormal event, and both the first extraction template and the second extraction template include an event identifier and a fault code keyword. The extraction submodule is used to extract a fault code from the information about the first abnormal event according to the same event identifier and fault code keyword in the first extraction template and the information about the first abnormal event, to obtain a first extraction result; and to extract a fault code from the information about the second abnormal event according to the same event identifier and fault code keyword in the second extraction template and the information about the second abnormal event, to obtain a second extraction result. The information determination submodule is used to determine the first fault cause description information according to the first extraction result, and to determine the second fault cause description information according to the second extraction result.

[0041] In a possible implementation, the second determination module includes: a first model acquisition submodule and a first relationship determination submodule. The first model acquisition submodule is used to acquire a first fault propagation relationship model, the first fault propagation relationship model includes multiple first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information. The first relationship determination submodule is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the first target fault propagation relationship in response to determining that there is a first target fault propagation relationship in the first fault propagation relationship model, the first target fault propagation relationship being a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0042] In another possible implementation, the second determination module includes: a first model acquisition submodule, a second model acquisition submodule, and a second relationship determination submodule. The first model acquisition submodule is used to acquire a first fault propagation relationship model, the first fault propagation relationship model includes multiple first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information. The second model acquisition submodule is used to, in response to determining that there is no first target fault propagation relationship in the first fault propagation relationship model, acquire a second fault propagation relationship model, the second fault propagation relationship model includes multiple second fault propagation relationships, the second fault propagation relationship is used to record the possible fault propagation direction of two abnormal events under the corresponding fault cause description information, and the second fault propagation relationship model is generated according to the first fault propagation relationship model. The second relationship determination submodule is used to, in response to determining that there is a second target fault propagation relationship in the second fault propagation relationship model, determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship, and the second target fault propagation relationship is a second fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0043] Optionally, the device further includes: a verification module and an update module. The verification module is used to verify the second target fault propagation relationship after the second relationship determination submodule determines the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship. The update module is used to update the first fault propagation relationship model and the second fault propagation relationship model according to the verified second target fault propagation relationship.

[0044] Optionally, the device further includes a first generation module. The first generation module is used to generate a first fault propagation relationship in the first fault propagation relationship model in the following manner: determining a third abnormal event on a first entity and third fault cause description information for describing the cause of the third abnormal event, and a fourth abnormal event on a second entity and fourth fault cause description information for describing the cause of the fourth abnormal event, wherein the first entity and the second entity are associated with each other; determining, based on expert annotations, the cause event and the effect event in the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information, to obtain the first fault propagation relationship.

[0045] Optionally, the device further includes a second generation module. The second generation module is used to generate the second fault propagation relationship in the second fault propagation relationship model in the following manner: determine a fifth abnormal event with unknown fault cause description information, wherein the unknown fault cause description information is the fault cause description information of the corresponding abnormal event other than the known fault cause description information, and the known fault cause description information is the fault cause description information in the first fault propagation relationship associated with the corresponding abnormal event; respectively calculate the first similarity between the unknown fault cause description information of the fifth abnormal event and each of the known fault cause description information of the fifth abnormal event to obtain a first similarity set; and determine, based on the first similarity set, the probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established.

[0046] In some examples, the second generating module is used to determine the probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established according to the first similarity set in accordance with at least one of the following implementations:

[0047] Determine, according to a first similarity greater than a first similarity threshold in the first similarity set and a second similarity greater than a second similarity threshold in the second similarity set, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event is established, wherein the second similarity set is a set of second similarities between the unknown fault cause description information of the sixth abnormal event and each known fault cause description information of the sixth abnormal event;

[0048] Determine, according to a first similarity in the first similarity set that is greater than a first similarity threshold, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established;

[0049] In response to determining that all first similarities in the first similarity set are not greater than a first similarity threshold, determining, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event, is established;

[0050] In response to determining that the first similarities in the first similarity set are not greater than a first similarity threshold, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, determine the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established.

[0051] In some examples, the second generation module is used to determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; determine a first weight corresponding to each first fault propagation relationship in the first relationship set according to the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; determine a second weight corresponding to the first fault propagation relationship in the first relationship set indicating that the fifth abnormal event causes the sixth abnormal event according to the first similarity and the second similarity corresponding to the first fault propagation relationship in the first relationship set indicating that the fifth abnormal event causes the sixth abnormal event; calculate the probability that the fifth abnormal event causes the sixth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the second weights.

[0052] In some examples, the second generation module is used to determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; determine a first weight corresponding to each first fault propagation relationship in the first relationship set according to the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; determine a fifth weight corresponding to the first fault propagation relationship in the first relationship set indicating that the sixth abnormal event causes the fifth abnormal event according to the first similarity and the second similarity corresponding to the first fault propagation relationship in the first relationship set indicating that the fifth abnormal event causes the sixth abnormal event; calculate the probability that the sixth abnormal event causes the fifth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the fifth weights.

[0053] In some examples, the second generation module is used to determine the first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to the first similarity greater than the first similarity threshold, to obtain a second relationship set; determine the first weight corresponding to each first fault propagation relationship in the second relationship set according to the first similarity corresponding to each first fault propagation relationship in the second relationship set; determine the second weight corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event; calculate the probability that the fifth abnormal event causes the sixth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the second weights; or,

[0054] In some examples, the second generation module is used to determine the first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to the first similarity greater than the first similarity threshold, to obtain a second relationship set; determine the first weight corresponding to each first fault propagation relationship in the second relationship set according to the first similarity corresponding to each first fault propagation relationship in the second relationship set; determine the fifth weight corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event; calculate the probability that the sixth abnormal event causes the fifth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the fifth weights.

[0055] Optionally, the device also includes: a third generation module, used to determine cause description keywords in the product manual of the target network device; based on the cause description keywords, extract template content from the product manual of the target network device; generate a target extraction template according to the template content; wherein, the target network device is the first network device, and the target extraction template is the first extraction template; or, the target network device is the second network device, and the target extraction template is the second extraction template.

[0056] In a fourth aspect, a network fault analysis device is provided, which includes an entity determination module, a direction determination module and a root cause event determination module. The entity determination module is used to determine entities associated with multiple abnormal events to obtain an entity set. The direction determination module is used to use the method described in the first aspect to determine the fault propagation direction of the seventh abnormal event associated with the third entity and the eighth abnormal event associated with the fourth entity under the corresponding fault cause description information, wherein the third entity and the fourth entity are any two mutually related entities in the entity set, the seventh abnormal event is any abnormal event of at least one abnormal event associated with the third entity, and the eighth abnormal event is any abnormal event of at least one abnormal event associated with the fourth entity. The root cause event determination module is used to determine the root cause event among the multiple abnormal events based on the determined fault propagation direction.

[0057] In some examples, the root cause event determination module includes a reference value determination submodule and a root cause event determination submodule. The reference value determination submodule is used to determine the reference value of each abnormal event according to the determined fault propagation direction, wherein the reference value is the sum of the weights of the fault propagation directions of the abnormal event to which it belongs as the result event, and the weight of the fault propagation direction is used to indicate the probability that the corresponding fault propagation direction is true. The root cause event determination submodule is used to determine the abnormal event with the smallest reference value as the root cause event among the multiple abnormal events.

[0058] In a fifth aspect, a computer device is provided, comprising a processor and a memory, wherein: the memory stores computer instructions, and the processor executes the computer instructions to implement the method of the first aspect and possible implementation methods thereof.

[0059] In a sixth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions. When the computer instructions in the computer-readable storage medium are executed by a computer device, the computer device executes the method of the first aspect and its possible implementation methods.

[0060] In a seventh aspect, a computer program product comprising instructions is provided, which, when executed on a computer device, enables the computer device to execute the method of the first aspect and its possible implementation methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a schematic diagram of a network architecture of an application scenario provided by an exemplary embodiment of the present application;

[0062] Figure 2 is a schematic diagram of the network structure of a campus network provided by an exemplary embodiment of the present application;

[0063] Figure 3 is a schematic diagram of the network structure of a data center network provided by an exemplary embodiment of the present application;

[0064] Figure 4 is a flow chart of a network fault analysis method provided by an exemplary embodiment of the present application;

[0065] Figure 5 is a schematic diagram of the generation process of the first fault propagation relationship model;

[0066] Figure 6 is a schematic diagram of the generation process of the second fault propagation relationship model;

[0067] Figure 7 is a flow chart of another network fault analysis method provided by an exemplary embodiment of the present application;

[0068] Figure 8 It is a structural schematic diagram of a network fault analysis device provided by an exemplary embodiment of the present application;

[0069] Fig. 9 It is a structural schematic diagram of another network fault analysis device provided by an exemplary embodiment of the present application;

[0070] Fig.10 It is a structural diagram of a computer device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0071] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0072] Figure 1 FIG. 1 is a schematic diagram of a network architecture of an application scenario provided by an exemplary embodiment of the present application. Figure 1 The network shown includes a plurality of network devices 10 and a network management device 20. The network devices 10 are communicatively connected with each other.

[0073] The network management device 20 is used to manage and control each network device 10, parse and present relevant data of the network device 10, analyze the traffic between the network devices 10, and perform intelligent maintenance on the network. In the embodiment of the present application, the network management device 20 is a server, or a server cluster composed of several servers, or a cloud computing service center.

[0074] The network device 10 includes but is not limited to a router, a switch, an access point (AP), a firewall (FW), a wireless access point control (WAC), etc. The type of the network device 10 is set according to actual needs.

[0075] In the embodiment of the present application, the network management device 20 is connected to at least part of the network devices 10 to interact with the connected network devices 10. When a network failure occurs, the network device 10 connected to the network management device 20 will report abnormal event information to the network management device, such as alarms, logs, etc. The network management device 20 performs network fault analysis based on the received abnormal event information to achieve functions such as fault location and troubleshooting.

[0076] like Figure 1 As shown, the network architecture also includes a cloud device 30. The cloud device 30 generally refers to the upper-level device of the network management device 20, and is used to provide the network management device 20 with a model for processing data, etc. Exemplarily, the model for processing data includes an extraction template, a first fault propagation relationship model, and a second fault propagation relationship model, etc. In other examples, the model for network processing data is manually configured in the network management device 20. In the embodiment of the present application, the cloud device 30 is a server, or a server cluster composed of several servers, or a cloud computing service center.

[0077] Optionally, the cloud device 30 and the network management device 20 may also be integrated into one device, which is not limited in the embodiments of the present application.

[0078] In the embodiments of the present application, the types of networks include but are not limited to campus networks, data center networks (DCNs), etc.

[0079] Figure 2 FIG. 1 is a schematic diagram of the network structure of a campus network provided by an exemplary embodiment of the present application. Figure 2 As shown, the network devices in the campus network include FW, WAC, LAN switch (LSW), AP, etc. LSW is connected to the external network (Internet, WAN) through FW.

[0080] Figure 3 FIG. 1 is a schematic diagram of a DCN network structure provided by an exemplary embodiment of the present application. Figure 3 As shown, the network devices in the DCN include FW, spine switches, leaf switches, etc.

[0081] Figure 4 A network fault analysis method provided by an exemplary embodiment of the present application is shown in FIG. 1 . The method is executed by a computer device, such as the aforementioned network management device. Figure 4 As shown, the method includes the following processes.

[0082] 201: Obtain information about a first abnormal event and information about a second abnormal event.

[0083] When a network device detects a network failure, it reports information about multiple abnormal events to a network management device. The information about multiple abnormal events can come from different network devices. After receiving the information about multiple abnormal events, the network management device saves it locally for subsequent network failure analysis. The information about the first abnormal event and the information about the second abnormal event are two of the information about the multiple abnormal events received by the network management device.

[0084] In an embodiment of the present application, the information of the abnormal event includes at least device information, event identifier and event description information. Among them, the device information is used to indicate the relevant information of the network device that sends the corresponding abnormal event information. Exemplarily, the device information includes the device model and system software version, etc. The event identifier, such as the event name, is used to identify different abnormal events. The event description information includes indication information of the entity associated with the abnormal event. The indication information of the entity is such as the name of the entity, etc. Exemplarily, the type of entity includes but is not limited to interfaces, devices, protocols and services, etc., and one network device can correspond to multiple entities.

[0085] In some examples, the event description information also includes fault cause description information, which is used to describe the cause of the corresponding abnormal event. The fault cause description information includes a fault code keyword and a fault code corresponding to the fault code keyword. Fault code keywords include ErrorCode and Reason, etc. The information of an abnormal event includes an event identifier and at least one fault code keyword, and each fault code keyword corresponds to a fault code. That is, the information of an abnormal event includes at least one fault code. When the information of an abnormal event includes multiple fault codes, the multiple fault codes are the same or different.

[0086] In other examples, the event description information does not include the fault code keyword and the fault code, or the abnormal event information includes the fault code keyword, but the fault code corresponding to the fault code keyword is empty. In this case, the abnormal event does not correspond to the fault cause description information.

[0087] In the embodiment of the present application, at least one of the information about the first abnormal event and the information about the second abnormal event includes fault cause description information.

[0088] Exemplarily, the information of the abnormal event is in the form of an alarm and / or a log.

[0089] In the embodiment of the present application, the first abnormal event and the second abnormal event are two abnormal events caused by the same network failure. Therefore, the first abnormal event and the second abnormal event are associated in space and / or time.

[0090] Spatial association means that there is an association relationship between the first entity corresponding to the first abnormal event and the second entity corresponding to the second abnormal event. The spatial association relationship includes but is not limited to the number of hops between the first entity and the second entity being within a set number of hops (for example, within 2 hops), the first entity and the second entity being peer entities, etc.

[0091] In a possible implementation, the first entity and the second entity belong to different network devices. For example, the first entity belongs to a first network device, and the second entity belongs to a second network device. Accordingly, the information of the first abnormal event is sent by the first network device, and the information of the second abnormal event is sent by the second network device.

[0092] In another possible implementation, the first entity and the second entity belong to the same network device, and the information of the first abnormal event and the information of the second abnormal event are sent by the same network device, that is, the first network device and the second network device are the same network device.

[0093] In some examples, the spatial association relationship is determined in the following manner: determining the corresponding entity based on the information of the abnormal event; determining whether there is an association relationship between the two entities corresponding to the two abnormal events based on the network map, wherein the network map is a relationship network composed of the association relationships between the entities in the network, and the network map also includes the identification of the abnormal events associated with each entity. If there is an association relationship between the two entities corresponding to the two abnormal events, it can be used as the information of the first abnormal event and the information of the second abnormal event, and the subsequent steps can be used to determine the association relationship between the first abnormal event and the second abnormal event.

[0094] In an embodiment of the present application, the network map is pre-formed based on the configuration information of each network device in the network, and the abnormal events reported by the network devices will be associated with the corresponding entities in the network map in real time.

[0095] The temporal association means that the time corresponding to the first abnormal event and the time corresponding to the second abnormal event are relatively close.

[0096] In some examples, a first interval between the occurrence time of the first abnormal event and the occurrence time of the second abnormal event is less than a first interval threshold. Since the occurrence times of two abnormal events with a propagation relationship are usually close, whether the interval between the occurrence time of the first abnormal event and the occurrence time of the second abnormal event is less than the first interval threshold can be used to indicate whether there may be a propagation relationship between the first abnormal event and the second abnormal event.

[0097] Here, the occurrence time of the first abnormal event and the second abnormal event are both determined by the corresponding network device, and are carried in the abnormal event information and sent to the network management device, for example, carried in the abnormal event information in the form of a timestamp.

[0098] In other examples, the second interval between the sending time of the information of the first abnormal event and the sending time of the information of the second abnormal event is less than the second interval threshold; or, the third interval between the receiving time of the information of the first abnormal event and the receiving time of the information of the second abnormal event is less than the third interval threshold; or, the second interval between the sending time of the information of the first abnormal event and the sending time of the information of the second abnormal event is less than the second interval threshold, and the third interval between the receiving time of the information of the first abnormal event and the receiving time of the information of the second abnormal event is less than the third interval threshold.

[0099] Here, the sending time of the information of the first abnormal event and the information of the second abnormal event refers to the time when the information is sent by the corresponding network device, and the sending time can be carried in the information of the abnormal event and sent to the network management device, for example, carried in the information of the abnormal event by a timestamp. The receiving time of the information of the first abnormal event and the information of the second abnormal event refers to the time when the network management device receives the corresponding information.

[0100] Since the two abnormal events with a propagation relationship occur very close in time, and the network equipment monitors the occurrence of abnormal events in real time and immediately reports the information of the abnormal events when the abnormal events are monitored, therefore, whether the second interval between the sending time of the information of the first abnormal event and the sending time of the information of the second abnormal event and the third interval between the receiving time of the information of the first abnormal event and the receiving time of the information of the second abnormal event are less than the corresponding interval threshold can also be used to indicate whether there may be a propagation relationship between the first abnormal event and the second abnormal event.

[0101] Among them, the first interval threshold, the second interval threshold and the third interval threshold are set as needed, and this application does not impose any restrictions on this.

[0102] In some examples, 201 includes: determining whether any two abnormal events have a correlation in time and space based on the received information of multiple abnormal events; treating the two abnormal events that have a correlation in time and space as the first abnormal event and the second abnormal event, and obtaining information about the first abnormal event and the second abnormal event.

[0103] In other examples, multiple abnormal events associated in time and space are screened out by clustering algorithms, such as K-means clustering method, density-based spatial clustering of applications with noise (DBSCAN) algorithm, etc. Any two of the multiple abnormal events associated in time and space are respectively the first abnormal event and the second abnormal event.

[0104] 202: Determine first fault cause description information according to information of the first abnormal event.

[0105] The first fault cause description information is used to describe the cause of the first abnormal event.

[0106] 203: Determine second fault cause description information according to the information of the second abnormal event.

[0107] The second fault cause description information is used to describe the cause of the second abnormal event.

[0108] 204: Based on the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information.

[0109] The fault propagation direction is used to indicate the cause event and the effect event in the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information.

[0110] In an embodiment of the present application, if the occurrence of a first abnormal event causes the occurrence of a second abnormal event, the fault propagation direction between the first abnormal event and the second abnormal event is from the first abnormal event to the second abnormal event. The first abnormal event can be referred to as the cause event of the second abnormal event, and the second abnormal event can be referred to as the result event of the first abnormal event. If the occurrence of the second abnormal event causes the occurrence of the first abnormal event, the fault propagation direction between the first abnormal event and the second abnormal event is from the second abnormal event to the first abnormal event. The second abnormal event can be referred to as the cause event of the first abnormal event, and the first abnormal event can be referred to as the result event of the second abnormal event.

[0111] In some examples, the identifier of the first abnormal event is the same as the identifier of the second abnormal event. Generally, when the event identifier of the first abnormal event is the same as the event identifier of the second abnormal event, at least one of the first fault cause description information and the second fault cause description information when the first abnormal event is a cause event and the first fault cause description information and the second fault cause description information when the first abnormal event is a result event is different.

[0112] For example, the first fault cause description information when the first abnormal event is a cause event is completely different or partially different from the first fault cause description information when the first abnormal event is a result event, and the second fault cause description information when the first abnormal event is a cause event is the same as the second fault cause description information when the first abnormal event is a result event. For another example, the first fault cause description information when the first abnormal event is a cause event is the same as the first fault cause description information when the first abnormal event is a result event, and the second fault cause description information when the first abnormal event is a cause event is completely different or partially different from the second fault cause description information when the first abnormal event is a result event. For another example, the first fault cause description information when the first abnormal event is a cause event is completely different or partially different from the first fault cause description information when the first abnormal event is a result event, and the second fault cause description information when the first abnormal event is a cause event is completely different or partially different from the second fault cause description information when the first abnormal event is a result event.

[0113] In this way, the propagation relationship between the first abnormal event and the second abnormal event can be distinguished by the first fault cause description information corresponding to the first abnormal event and the second fault cause description information corresponding to the second abnormal event.

[0114] The following example is used to illustrate the situation where the first fault cause description information when the first abnormal event is the cause event is completely different from the first fault cause description information when the first abnormal event is the result event, and the second fault cause description information when the first abnormal event is the cause event is the same as the second fault cause description information when the first abnormal event is the result event.

[0115] The information of the first abnormal event is an alarm caused by the switch port down in the campus network, and the event identifier of the first abnormal event is IF_LINKDOWN. The information of the second abnormal event is an alarm caused by the failure of the AP connected to the switch port, and the event identifier of the second abnormal event is AP_NORMAL_TO_FAULT. The switch port down will cause the AP connected to the switch port to fail, and the AP failure will also cause the switch port connected to the AP to be down. When the switch port down causes the AP connected to the switch port to fail, the first fault cause description information is "Heartbeat packet transmission for the CAPWAP control tunnel between the AC and AP times out". When the AP failure causes the switch port connected to the AP to be down, the first fault cause description information is "A command is delivered to reboot an AP".

[0116] If, according to the related art, the association relationship of abnormal events is determined based on the identification of abnormal events, it can only be determined that the first abnormal event and the second abnormal event are mutually causally related, and it is impossible to further distinguish the propagation relationship between the first abnormal event and the second abnormal event. In the embodiment of the present application, the cause event and the effect event in the first abnormal event and the second abnormal event can be distinguished based on the first fault cause description information, so that the fault location is more accurate and precise.

[0117] In other examples, the identifier of the first abnormal event is different from the identifier of the second abnormal event.

[0118] It should be noted that when the first fault cause description information and the second fault cause description information are both empty, the propagation relationship between the first abnormal event and the second abnormal event is determined based on the event identifier in the information of the first abnormal event and the event identifier in the information of the second abnormal event. This application does not limit the method of determining the propagation relationship between the first abnormal event and the second abnormal event based on the event identifier in the information of the first abnormal event and the event identifier in the information of the second abnormal event.

[0119] In some examples, 202 includes:

[0120] The first step is to obtain a first extraction template of the first network device and a second extraction template of the second network device, wherein the first extraction template and the second extraction template both include an event identifier, a fault code keyword, a candidate fault code corresponding to the fault code keyword, and fault code description information of each candidate fault code. Here, a candidate fault code refers to a fault code that a fault keyword may correspond to, and a fault keyword usually corresponds to multiple candidate fault codes.

[0121] Step 2: extract the fault code from the information of the first abnormal event according to the same event identifier and fault code keyword in the first extraction template and the information of the first abnormal event, and obtain a first extraction result.

[0122] Step 3: Extract the fault code from the information of the second abnormal event according to the second extraction template and the same event identifier and fault code keyword in the information of the second abnormal event to obtain a second extraction result.

[0123] Step 4: determine the first fault cause description information according to the first extraction result, and determine the second fault cause description information according to the second extraction result.

[0124] It should be noted that the second step and the third step in 202 have no sequence and can be executed sequentially or simultaneously.

[0125] In 202 and 203, the fault codes in the information of the first abnormal event and the information of the second abnormal event are extracted using corresponding extraction templates. In the embodiment of the present application, the extraction template is mainly used to quickly extract the fault code in the information of the abnormal event when the device reports the information of the abnormal event, and obtain the fault cause description information corresponding to the fault code for fine root cause location.

[0126] To facilitate understanding of the extraction process, the following briefly introduces the content and generation method of the extraction template.

[0127] Exemplarily, the template content in the extraction template includes event identifier, fault code keyword, fault code variable type, candidate fault codes corresponding to the fault code keyword, and fault code description information corresponding to each candidate fault code. Among them, the event identifier is used to identify different types of abnormal events, and the fault code keyword is used to represent the fault code name. The fault code variable type is used to represent the variable type of the fault code corresponding to the fault code keyword, including but not limited to unsigned long integer (ULONG), character string (STRING), etc. The candidate fault code is a possible variable value corresponding to the fault code keyword. The fault code description information corresponding to the candidate fault code is used to describe the meaning of the candidate fault code.

[0128] In addition to the template content, the extraction template also includes device information, including but not limited to the device model and the system software version of the device.

[0129] Table 1 Extraction template

[0130]

[0131] As shown in Table 1, the extraction template includes the device model "S12700" and the system software version "V200R011C10." The template content of the extraction template includes template content corresponding to multiple abnormal events.

[0132] In the extraction template, the template content corresponding to the first abnormal event includes: event identifier "rpc_failed", fault code keyword "ErrorCode", fault code variable type: "value": "ULONG", candidate fault codes and fault code description information corresponding to the candidate fault codes include "2: Failed to allocate message memory"; "3: Synchronous IPC timeout"; "4: Message waiting queue overflow".

[0133] In the extraction template, the template content corresponding to the second abnormal event includes: event identifier "ipallocfail", fault code keyword "ErrorCode", fault code variable type: "value": "STRING", candidate fault codes and fault code description information corresponding to the candidate fault codes include "No idle IP address: There is no IP address available for allocation in the address pool."; "Pool is locked: The address pool is locked"; "Exception occurred: System abnormal error".

[0134] In an embodiment of the present application, a method of generating an extraction template includes at least one of the following methods.

[0135] Method 1: Obtain a product manual of a target network device and generate an extraction template based on the product manual, wherein the target network device is any network device in the network.

[0136] In a possible implementation, generating a target extraction template based on a product manual of a target network device includes:

[0137] Step 1: Identify the cause description keywords in the product manual of the target network device;

[0138] Step 2: Based on the cause description keywords, extract the template content from the product manual of the target network device;

[0139] Step 3: Generate a target extraction template based on the template content.

[0140] In some examples, determining the cause description keywords in the product manual of the target network device can be done through semantic analysis or manual annotation. Optionally, the cause description keywords include at least one of a single word and a phrase consisting of at least two words. Exemplarily, the cause description keywords include but are not limited to: reason, error, because, due to, error code, because of, etc.

[0141] Exemplarily, in the second step, based on the cause description keywords, the template content is extracted from the product manual of the target network device, including: first, using a string matching method to search for each cause description keyword from the product manual; then, based on the regular expression corresponding to the found cause description keyword, the template content is extracted from the product manual. Here, the regular expression describes a string matching pattern (pattern) used to extract a substring that meets a certain condition from a certain string. One cause description keyword corresponds to at least one regular expression.

[0142] In the embodiment of the present application, there are at least two types of regular expressions. The first type of regular expressions is used to extract fault code keywords and fault code variable types. The second type of regular expressions is used to extract fault codes and fault code description information corresponding to the fault codes.

[0143] Exemplarily, the first type of regular expression contains the cause description keywords. The second type of regular expression contains the tags in the file format corresponding to the product document. For example, for a product document in the hypertext markup language (HTML) format, the second type of regular expression is used to match the line break tags in the product document.

[0144] In addition to regular expressions, other means that can extract the required template content can be used to extract template content from the product manual, such as crawler technology, etc., and this application does not impose any restrictions on this.

[0145] The following is an example to illustrate the process of extracting template content.

[0146] Taking the log RPC_FAILED as an example, this log indicates that the switch synchronous IPC (Inter-Process Communication) call failed. In the switch product documentation, the "ErrorCode = [ULONG]" field in the log information contains the specific reporting reason of the log, that is, the description of the fault cause. Based on the above errorcode keyword, regular expression matching can be performed to obtain the log fault code keyword ErrorCode, and the fault code variable is ULONG.

[0147] The fault code and the meaning of the fault code parameters (i.e. the fault code description information) are presented in the form of a table in the product manual. Taking the log RPC_FAILED as an example, in the source code of the HTML format product document, the tag and The content of the table is between each pair of labels and the content of a row of the table. The content of the table header is between the label and the content, that is, "Parameter Meaning", "Parameter Name", etc. The content of each cell of the table is between the label and the content of each row. The different meanings of the same fault code are usually or The html-specific tags are used to separate the fault codes. Based on this, regular expression matching can be performed to obtain information about the fault codes and the meaning of the fault code parameters.

[0148] Method 2: Generate an extraction template based on received input information, wherein the input information includes template content.

[0149] It should be noted that, here, the method of generating the extraction template includes at least one of the following methods, which means generating the extraction template using method one; or generating the extraction template using method two; or generating a part of the extraction template using method one and generating a part of the template using method two.

[0150] In some examples, the extraction template is generated by the cloud device and then sent to the network management device, and the network management device receives and stores the extraction template sent by the cloud device. In this case, the method further includes: the network management device sends a template acquisition request to the cloud device, and the template acquisition request includes device information of the network device; the cloud device sends the extraction template of the network device corresponding to the device information in the template acquisition request to the network management device according to the received template acquisition request.

[0151] In other examples, the extraction template may also be generated by the network management device and stored in the network management device. In this way, the network management device can subsequently extract the fault code from the abnormal event information according to the stored extraction template.

[0152] In an embodiment of the present application, at least one of the first extraction result and the second extraction result includes at least one fault code. In one possible implementation, the first extraction result is empty, that is, no fault code is extracted from the information of the first abnormal event, and the second extraction result includes at least one fault code. In another possible implementation, the first extraction result includes at least one fault code, and the second extraction result is empty. In yet another possible implementation, the first extraction result includes at least one fault code, and the second extraction result includes at least one fault code.

[0153] Extracting the fault code from the information of the first abnormal event according to the event identifier and the fault code keyword in the first extraction template to obtain a first extraction result includes:

[0154] First, determine the target event identifier and target fault code keyword in the first extraction template, that is, match the event identifier in the first extraction template with the event identifier in the information of the first abnormal event respectively, the event identifier in the first extraction template that is the same as the event identifier in the information of the first abnormal event is the target event identifier, and the fault code keyword in the first extraction template that corresponds to the target event identifier and is the same as the fault code keyword in the information of the first abnormal event is the target fault code keyword; then, based on the target fault code keyword, perform regular expression matching on the fault code keyword in the information of the first abnormal event to extract the fault code from the information of the first abnormal event.

[0155] For example, in the extraction template, the target fault code keyword and the corresponding variable type are "ErrorCode=[ULONG]", the corresponding regular expression is ErrorCode=[*], and the record format of the fault code keyword and the corresponding fault code in the abnormal event information is ErrorCode=[6]. According to the regular expression, the fault code corresponding to the fault code keyword is extracted as 6.

[0156] The method for generating the second extraction result is similar to the process for generating the first extraction result, and a detailed description is omitted in the embodiment of the present application.

[0157] In the fourth step, after determining the first extraction result, the fault code and the fault code keyword included in the first extraction result are combined to obtain the first fault cause description information. Similarly, after determining the second extraction result, the fault code and the fault code keyword included in the second extraction result are combined to obtain the second fault cause description information.

[0158] It should be noted that, when no fault code is extracted from the information of the first abnormal event, the first extraction result is empty, the first fault cause description information is empty, or the first fault cause description information is the event identifier of the first abnormal event. When no fault code is extracted from the information of the second abnormal event, the second extraction result is empty, the second fault cause description information is empty, or the second fault cause description information is the event identifier of the second abnormal event.

[0159] In some examples, 203 includes:

[0160] Obtain a first fault propagation relationship model, wherein the first fault propagation relationship model includes multiple first fault propagation relationships, each of which is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information; determine whether there is a first target fault propagation relationship in the first fault propagation relationship model. If there is a first target fault propagation relationship, determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the first target fault propagation relationship, and the first target fault propagation relationship is a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0161] In the embodiment of the present application, by comparing the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information with each first fault propagation relationship in the first fault propagation relationship model, the first fault propagation relationship including the event identifier in the information of the first abnormal event, the event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information is the first target fault propagation relationship. It is convenient and simple to determine whether the first target fault propagation relationship exists in the first fault propagation relationship model by comparing and searching.

[0162] Table 2

[0163]

[0164] As shown in Table 2, the first fault propagation relationship model includes multiple first fault propagation relationships.

[0165] In the first fault propagation relationship, the entity type "src_label" corresponding to the result event is "AP", and the entity type "dst_label" corresponding to the cause event is "AP". The event identifier "src_event" of the result event is "AP_FAULT", indicating an AP offline notification. The fault cause description information "src_err_code" corresponding to the result event is empty, indicating that the result event has no fault cause description information. The event identifier "dst_event" of the cause event: "AP_NORMAL_TO_FAULT", indicating that the AP changes from a normal state to an abnormal state. The fault cause description information "dst_err_code" corresponding to the cause event is the fault code keyword @fault code "Reason@a Command is delivered to reboot an ap", indicating that a command to reset the AP is issued.

[0166] In the second first fault propagation relationship, the entity type "src_label" corresponding to the result event is "AP", and the entity type "dst_label" corresponding to the cause event is "AP". The event identifier "src_event" of the result event is "AP_HOT_BOOT", indicating an AP hot start alarm. The fault cause description information "src_err_code" corresponding to the result event is empty, indicating that the result event has no fault cause description information. The event identifier "dst_event" of the cause event is: "AP_NORMAL_TO_FAULT", indicating that the AP changes from normal to abnormal status. The fault cause description information "dst_err_code" corresponding to the cause event is the fault code keyword @fault code Reason @a Command is delivered to reboot an ap", indicating that a command to reset the AP is issued.

[0167] Optionally, each first fault propagation relationship also includes association mode indication information, and the association mode indication information is used to indicate the number of hops between two entities. When the number of hops is equal to 0, the first abnormal event and the second abnormal event are on the same entity; when the number of hops is equal to 1, the two entities are directly connected; when the number of hops is equal to 2, the two entities are indirectly connected through one entity; and so on. For example, in the first and second first fault propagation relationships, "num_hops": 0 means that the number of hops is 0, that is, the two abnormal events are located on the same entity.

[0168] In some examples, the first fault propagation relationship model is obtained based on historical fault cases.

[0169] Exemplarily, any first fault propagation relationship in the first fault propagation relationship model is generated in the following manner:

[0170] The first step is to determine a first entity and a second entity that are mutually associated. Here, being mutually associated means that the number of hops between the two entities is within a set number of hops, such as within 1 hop or within 2 hops.

[0171] Step 2: determine the third abnormal event on the first entity and the third fault cause description information used to describe the cause of the third abnormal event, as well as the fourth abnormal event on the second entity and the fourth fault cause description information used to describe the cause of the fourth abnormal event; here, the number of the third abnormal events is one or more, and the number of the fourth abnormal events is one or more.

[0172] Step 3: According to the expert annotation, determine the cause event and the effect event in the third abnormal event corresponding to any third fault cause description information and the fourth abnormal event corresponding to any fourth fault cause description information, and obtain the first fault propagation relationship. The expert annotation is used to indicate the root cause in the third fault cause description information corresponding to the third abnormal event and the fourth fault cause description information corresponding to the fourth abnormal event.

[0173] Exemplarily, the mutually related first entity and second entity are determined by graph search or association learning algorithms. For example, a network map is first generated based on historical fault cases. The network map is a relationship network composed of associations between entities in the network, and the network map also contains identifiers of abnormal events associated with each entity. Based on the network map, the mutually related first entity and second entity, as well as the third abnormal event on the first entity and the fourth abnormal event on the second entity can be determined. Based on the actual situation of historical fault cases, combined with the third fault cause description information corresponding to the third abnormal event and the fourth fault cause description information corresponding to the fourth abnormal event, the expert determines the cause events in the third abnormal event and the fourth abnormal event and marks them. Based on the marking, the propagation relationship between the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information is learned to obtain the first fault propagation relationship, and added to the first fault propagation relationship model.

[0174] Combine the following Figure 5 The generation process of the first fault propagation relationship is exemplified. Figure 5 As shown in the figure, an open shortest path first (OSPF) protocol failure is taken as an example. An abnormality occurs in the OSPF protocol between two network devices. In the network graph, some abnormal events are associated with two OSPF protocol entities. The graph search algorithm is used to obtain the abnormal events on the two entities.

[0175] Among them, the abnormal events on the entity on the left include: event 1_1, the neighbor state changes to down, the event identifier is NBR_CHG_DOWN; event 1_2, the neighbor state changes to down, the event identifier is NBR_DOWN_REASON, and the fault cause description information is that the physical interface state changes (physical interface state change); event 1_3, the OSPF non-virtual link interface state changes, the event identifier is IFCHG, and the fault cause description information is that the physical interface state changes (physical interface state change); event 1_4, the OSPF neighbor state changes, the event identifier is NBRCHG, and the fault cause description information is that the physical interface state changes (physical interface statechange).

[0176] Among them, event 1_2, event 1_3, and event 1_4 have different event identifiers, but have the same fault cause description information, indicating that they are triggered by the same reason. For abnormal event information that does not contain specific fault cause description information, such as event 1_1, its fault cause description information can be the same as the event identifier.

[0177] The abnormal events on the right entity include: event 2_1, the neighbor state changes to down, the event identifier is NBR_DOWN_REASON, and the fault cause description information is received 1-Way Hello message (1-wayhello received); event 2_2, the OSPF neighbor state changes, the event identifier is NBRCHG, and the fault cause description information is the neighbor switch reason causes the alarm (peer router reason).

[0178] Based on the specific circumstances of the fault case, the expert marked the fault cause description information of the change in the physical interface status as the root cause of the fault. The algorithm combined with the root cause labeling learned the 0th-order propagation relationship of events 1_1, 1_2, 1_3, and 1_4 on the same entity (i.e., the 0th-order associated entity), that the change in the physical interface status on one entity caused the occurrence of events 1_1, 1_2, 1_3, and 1_4; learned the 1st-order propagation relationship between 1st-order associated entities, that the change in the physical interface status on one entity caused the other entity to receive a 1-Way Hello message, and then reported event 2_1; learned the 1st-order propagation relationship between 1st-order associated entities, that the change in the physical interface status on one entity caused the neighbor switch cause alarm on the other entity, and then reported event 2_2.

[0179] These fault code propagation relationships, which are combined with specific fault cause description information, are added to the fault propagation relationship library and can be used for real-time root cause tracing.

[0180] It should be noted that the associated entity in this fault case is a one-hop relationship. The fault code propagation relationship learning method for multi-hop relationships is similar and will not be described in detail here.

[0181] In a possible implementation, when an abnormal event corresponds to more than two fault cause description information, a main fault cause description information can be determined by expert annotation, etc. When performing network fault analysis, the propagation relationship between the two abnormal events is determined only based on the main fault cause description information.

[0182] Since historical fault cases are limited and cannot contain all fault cause description information of each abnormal event, a second fault propagation relationship model can also be generated based on the first fault propagation relationship model. The second fault propagation relationship model includes at least one second fault propagation relationship, and each second fault propagation relationship is used to record the possible fault propagation direction of two abnormal events under the corresponding fault cause description information.

[0183] Table 3

[0184]

[0185]

[0186] As shown in Table 3, the second fault propagation relationship model includes multiple second fault propagation relationships. The first and second second fault propagation relationships are used as examples for exemplary description.

[0187] In the first second fault propagation relationship, the entity type "src_label" corresponding to the result event is "interface", and the entity type "dst_label" corresponding to the cause event is "AP". The event identifier "src_event" of the result event is "AP_FAULT", indicating an AP offline notification. The fault cause description information "src_err_code" corresponding to the result event is empty "", indicating that the result event has no fault cause description information. The event identifier "dst_event" of the cause event is "AP_NORMAL_TO_FAULT", indicating that the AP changes from a normal state to an abnormal state. The fault cause description information "dst_err_code" corresponding to the cause event is the fault code keyword @fault code "Reason@The ap restarts due to a chip exception", indicating that the chip exception causes the AP to restart.

[0188] In the second fault propagation relationship of the second line, the entity type "src_label" corresponding to the result event is "interface", and the entity type "dst_label" corresponding to the cause event is "AP". The event identifier "src_event" of the result event is "AP_WORK_MODE_CHANGE", indicating that the AP working mode has changed. The fault cause description information "src_err_code" corresponding to the result event is empty, indicating that the result event has no fault cause description information. The event identifier "dst_event" of the cause event is "AP_NORMAL_TO_FAULT", indicating that the AP has changed from a normal state to an abnormal state. The fault cause description information "dst_err_code" corresponding to the cause event is the fault code keyword @fault code "Reason@The ap restarts due to a chipexception", indicating that the chip abnormality caused the AP to restart.

[0189] Optionally, the second fault propagation relationship also includes probability information for indicating the possibility of the propagation relationship between the two abnormal events being established, such as a "probability" field. For example, in the first second fault propagation relationship, the probability of the propagation relationship between the cause event and the effect event being established is 0.8; in the second second fault propagation relationship, the probability of the propagation relationship between the cause event and the effect event being established is 0.6.

[0190] When the propagation relationship between the first abnormal event and the second abnormal event cannot be determined according to the first fault propagation model, that is, the first target fault propagation relationship does not exist in the first fault propagation relationship model, S204 includes: obtaining a second fault propagation relationship model; in response to determining that there is a second target fault propagation relationship in the second fault propagation relationship model, determining the possible fault propagation direction between the first abnormal event and the second abnormal event according to the second target fault propagation relationship, the second target fault propagation relationship being a second fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0191] Exemplarily, the second fault propagation relationship in the second fault propagation relationship model is generated in the following manner:

[0192] For unknown fault cause description information that exists in the extraction template but does not exist in the first fault propagation relationship model, the possible fault propagation relationship between the two abnormal events under the corresponding fault cause description information is determined according to the first fault propagation relationship model, and the probability that the possible fault propagation relationship is established is calculated.

[0193] The second fault propagation relationship is divided into three cases. The first is the second fault propagation relationship based on the unknown fault cause description information corresponding to the fifth abnormal event and the unknown fault cause description information corresponding to the sixth abnormal event; the second is the second fault propagation relationship based on the unknown fault cause description information corresponding to the fifth abnormal event and the known fault cause description information corresponding to the sixth abnormal event; the third is the second fault propagation relationship based on the known fault cause description information corresponding to the fifth abnormal event and the unknown fault cause description information corresponding to the sixth abnormal event.

[0194] For the first case, the probability that the possible fault propagation relationship between the fifth abnormal event and the sixth abnormal event is established is calculated as follows:

[0195] First, the first similarities between the unknown fault cause description information corresponding to the fifth abnormal event and the various known fault cause description information of the fifth abnormal event are calculated respectively to obtain a first similarity set; and the second similarities between the unknown fault cause description information corresponding to the sixth abnormal event and the various known fault cause description information of the sixth abnormal event are calculated respectively to obtain a second similarity set; then, based on the first similarity set and the second similarity set, the probability that the second fault propagation relationship between the fifth abnormal event and the sixth abnormal event is established is determined, and this probability is also called the fault propagation probability.

[0196] The probability that the second fault propagation relationship is established includes at least one of a first probability and a second probability, the first probability being the probability that the fifth abnormal event causes the sixth abnormal event, and the second probability being the probability that the sixth abnormal event causes the fifth abnormal event.

[0197] In some examples, there is a first similarity greater than a first similarity threshold in the first similarity set, and there is a second similarity greater than a second similarity threshold in the second similarity set. The first probability is calculated based on the first similarity greater than the first similarity threshold in the first similarity set and the second similarity greater than the second similarity threshold in the second similarity set. The calculation process of the first probability is as follows:

[0198] Determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set;

[0199] Determine, according to the first similarity and the second similarity corresponding to the first fault propagation relationship in the first relationship set, the first weight corresponding to each first fault propagation relationship in the first relationship set, for example, taking the product of the first similarity and the second similarity as the first weight of the corresponding first fault propagation relationship;

[0200] According to the first similarity and the second similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event leads to the sixth abnormal event in the first relationship set, determine the second weight corresponding to the first fault propagation relationship indicating that the fifth abnormal event leads to the sixth abnormal event in the first relationship set, for example, taking the product of the first similarity and the second similarity as the second weight of the corresponding first fault propagation relationship;

[0201] The first probability is calculated according to the sum of the first weights and the sum of the fifth weights corresponding to each first fault propagation relationship in the first relationship set. For example, the ratio of the sum of the second weights to the sum of the first weights is taken as the first probability.

[0202] In some examples, the first similarity threshold and the second similarity threshold are equal. In other examples, the first similarity threshold and the second similarity threshold are not equal.

[0203] The calculation method of the second probability is the same as the calculation method of the first probability, and the detailed description is omitted here.

[0204] Exemplarily, the first probability is calculated according to the following formula (1):

[0205]

[0206] In formula (1), A represents the fifth abnormal event, a? represents the unknown fault cause description information corresponding to the fifth abnormal event, B represents the sixth abnormal event, b? represents the unknown fault cause description information corresponding to the sixth abnormal event, and P(A|a?->B|b?) is the fault propagation probability of the fifth abnormal event leading to the sixth abnormal event based on the unknown fault cause description information corresponding to the fifth abnormal event and the unknown fault cause description information corresponding to the sixth abnormal event. wi represents the first weight of the i-th first fault propagation relationship, m represents the total number of first fault propagation relationships, and m1 represents the number of first fault propagation relationships from the fifth abnormal event to the sixth abnormal event. A-Bj represents the j-th first fault propagation relationship representing the sixth abnormal event from the fifth abnormal event. W A-Bj represents the second weight corresponding to the j-th first fault propagation relationship indicating that the sixth abnormal event is caused by the fifth abnormal event.

[0207] The second probability is calculated according to the following formula (2):

[0208]

[0209] In formula (2), A represents the fifth abnormal event, a? represents the unknown fault cause description information corresponding to the fifth abnormal event, B represents the sixth abnormal event, b? represents the unknown fault cause description information corresponding to the sixth abnormal event, and P(B|b?->A|a?) is the fault propagation probability of the sixth abnormal event causing the fifth abnormal event based on the unknown fault cause description information corresponding to the fifth abnormal event and the unknown fault cause description information corresponding to the sixth abnormal event. wi represents the first weight of the i-th first fault propagation relationship, m represents the total number of first fault propagation relationships, and m2 represents the number of first fault propagation relationships from the sixth abnormal event to the fifth abnormal event. B-Aj represents the j-th first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event. W B-Aj represents a fifth weight corresponding to the j-th first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event.

[0210] For the second case, the probability that the possible fault propagation relationship between the fifth abnormal event and the sixth abnormal event is established is calculated as follows:

[0211] First, the first similarities between the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the fifth abnormal event are calculated respectively to obtain a first similarity set; then, according to the first similarity set, the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established is determined.

[0212] In some examples, based on the first similarity set, determining the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on unknown fault cause description information of the fifth abnormal event and known fault cause description information of the sixth abnormal event, includes: calculating at least one of the first probability and the second probability based on a first similarity in the first similarity set that is greater than a first similarity threshold.

[0213] Calculating a first probability according to a first similarity in the first similarity set that is greater than a first similarity threshold includes:

[0214] Determine a first fault propagation relationship associated with known fault cause description information of the fifth abnormal event and known fault cause description information of the sixth abnormal event corresponding to a first similarity greater than a first similarity threshold, to obtain a second relationship set;

[0215] Determine, according to the first similarity corresponding to the first fault propagation relationship in the second relationship set, a first weight corresponding to each first fault propagation relationship in the second relationship set, for example, using the first similarity as the first weight of the corresponding first fault propagation relationship;

[0216] According to the first similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event leads to the sixth abnormal event in the second relationship set, determine the second weight corresponding to the first fault propagation relationship indicating that the fifth abnormal event leads to the sixth abnormal event in the second relationship set, for example, use the first similarity as the corresponding second weight;

[0217] The first probability is calculated according to the sum of the first weights and the sum of the second weights corresponding to each first fault propagation relationship in the second relationship set. For example, the ratio of the sum of the second weights to the sum of the first weights is taken as the first probability.

[0218] Exemplarily, the first probability and the second probability are calculated according to the following formulas (3) and (4), respectively:

[0219] P(A->B)=w1 / (w1+w2+w3) (3)

[0220] Among them, P(A->B) represents the first probability, w1 represents the second weight corresponding to the first fault propagation relationship in the second relationship set, indicating that the fifth abnormal event leads to the sixth abnormal event, and w1+w2+w3 represents the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set.

[0221] P(B->A)=(w2+w3) / (w1+w2+w3) (4)

[0222] Among them, P(B->A) represents the second probability, w2+w3 represents the second weight corresponding to the first fault propagation relationship in the second relationship set, indicating that the sixth abnormal event causes the fifth abnormal event, and w1+w2+w3 represents the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set.

[0223] For the third case, the probability that the possible fault propagation relationship between the fifth abnormal event and the sixth abnormal event is established is calculated as follows:

[0224] First, the similarities between the unknown fault cause description information corresponding to the sixth abnormal event and the known fault cause description information of the sixth abnormal event are calculated respectively to obtain a second similarity set; then, the second similarity set is used to determine the fault propagation probability between the fifth abnormal event and the sixth abnormal event.

[0225] The relevant process refers to the second case mentioned above, and the detailed description is omitted here.

[0226] In some other examples, determining the fault propagation probability between the fifth abnormal event and the sixth abnormal event according to the first similarity set includes:

[0227] If all the first similarities in the first similarity set are not greater than the similarity threshold, then according to the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event, the fault propagation probability between the fifth abnormal event and the sixth abnormal event is calculated based on the unknown fault cause description information corresponding to the fifth abnormal event and the known fault cause description information (or unknown fault cause description information) of the sixth abnormal event.

[0228] Likewise, the fault propagation probability between the fifth abnormal event and the sixth abnormal event includes a first probability that the fifth abnormal event causes the sixth abnormal event, and a second probability that the sixth abnormal event causes the fifth abnormal event.

[0229] Exemplarily, the first probability is calculated by the following formula (5), and the second probability is calculated by the following formula (6).

[0230] P(A->B)=P1 / P0 (5)

[0231] Among them, P(A->B) represents the first probability, P0 represents the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event, and P1 represents the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event, indicating that the fifth abnormal event causes the sixth abnormal event to occur.

[0232] P(B->A)=P2 / P0 (6)

[0233] Among them, P(B->A) represents the second probability, P0 represents the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event, and P2 represents the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event, indicating that the sixth abnormal event causes the fifth abnormal event to occur.

[0234] In some other examples, determining the fault propagation probability between the fifth abnormal event and the sixth abnormal event according to the first similarity set includes: if there is no first similarity greater than the first similarity threshold in the first similarity set, and there is a second similarity greater than the second similarity threshold in the second similarity set, then according to the number of first fault propagation relationships between the fifth abnormal event and the sixth abnormal event associated with the second similarity greater than the second similarity threshold, calculating the fault propagation probability between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information corresponding to the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event. The calculation method refers to the relevant contents of formulas (5) and (6), and the detailed description is omitted here.

[0235] In an embodiment of the present application, when the fault code in the fault cause description information is in a string format, the similarity between the unknown fault cause description information and the known fault cause description information is the similarity between the fault codes in the string format. When the fault code in the fault cause description information is in a non-string format, such as a long integer, the similarity between the unknown fault cause description information and the known fault cause description information is the similarity between the fault code description information corresponding to the fault code. The fault code description information corresponding to the fault code can be determined by the aforementioned extraction template.

[0236] In the embodiments of the present application, there is no limitation on the method for calculating similarity, including but not limited to natural language processing methods such as sentence similarity calculation.

[0237] Combine the following Figure 6 The generation process of the second fault propagation relationship is exemplified. Figure 6 As shown, it is assumed that, in the first fault propagation relationship model, there are four first fault propagation relationships between abnormal event A and abnormal event B. In the first to third first fault propagation relationships, abnormal event A is the result event, and abnormal event B is the cause event. In the fourth first fault propagation relationship, abnormal event B is the result event, and abnormal event A is the cause event. In these four first fault propagation relationships, the fault cause description information corresponding to abnormal event A is a1, a2, and a3, respectively, and the fault cause description information corresponding to abnormal event B is b1, b2, and b3.

[0238] In the extraction template corresponding to abnormal event A, there is also fault cause description information a?, then the similarities between a? and a1, a2 and a3 are calculated respectively, and the first similarities calculated are wa1, wa2 and wa3 respectively. In the extraction template corresponding to abnormal event B, there is also fault cause description information b?, then the similarities between b? and b1, b2 and b3 are calculated respectively, and the second similarities calculated are wb1, wb2 and wb3 respectively.

[0239] Compare the first similarity and the second similarity with the similarity threshold respectively. If both the first similarity and the second similarity are not less than the similarity threshold. Then, the product of the first similarity and the second similarity corresponding to the first fault propagation relationship is used as the first weight of the first fault propagation relationship. Then, the first weight w1 corresponding to the first first fault propagation relationship is w1=wa1*wb1, the first weight w2 corresponding to the second first fault propagation relationship is w2=wa1*wb2, the first weight w3 corresponding to the third first fault propagation relationship is w2=wa2*wb2, and the first weight w4 corresponding to the fourth first fault propagation relationship is w3*wb2.

[0240] Abnormal event A is the result event, abnormal event B is the cause event, and the second weights corresponding to the first fault propagation relationship corresponding to the abnormal event A are w1, w2, and w3 respectively, and the first probability is equal to (w1+w2+w3) / (w1+w2+w3+w4). Abnormal event B is the result event, abnormal event A is the cause event, and the second weight corresponding to the first fault propagation relationship corresponding to the abnormal event A is w4, and the second probability is equal to w4 / (w1+w2+w3+w4).

[0241] If there is both a similarity not less than the similarity threshold and a similarity less than the similarity threshold in the first similarity and the second similarity, the weight corresponding to the similarity less than the similarity threshold needs to be removed from the formula.

[0242] If there is no similarity between the first similarity and the second similarity that is not less than the similarity threshold, the first probability and the second probability are calculated according to the number of first fault propagation relationships corresponding to abnormal event A as the result event and abnormal event B as the cause event and the number of first fault propagation relationships corresponding to abnormal event B as the result event and abnormal event A as the cause event.

[0243] Optionally, after determining the possible fault propagation relationship between the first abnormal event and the second abnormal event according to the second fault propagation relationship model, it is also necessary to verify the possible fault propagation relationship, that is, to verify the second target fault propagation relationship.

[0244] In some examples, verification is implemented through a cloud device. The network management device sends the second target fault propagation relationship to the cloud device, which provides it to an expert for verification to obtain a verified second target fault propagation relationship. In one possible implementation, after confirmation by an expert, the second target fault propagation relationship is correct, and the second target fault propagation relationship is the verified second target fault propagation relationship. In another possible implementation, the second target fault propagation relationship fails to pass the verification, and the second target fault propagation relationship is corrected according to the expert's correction instruction, and the corrected second target fault propagation relationship is the verified second target fault propagation relationship.

[0245] Table 4

[0246]

[0247] Optionally, the method further includes: updating the first fault propagation relationship model and the second fault propagation relationship model according to the verified second target fault propagation relationship.

[0248] Wherein, updating the first fault propagation relationship model includes adding the verified second target fault propagation relationship as a new first fault propagation relationship to the first fault propagation relationship model.

[0249] Updating the second fault propagation relationship model includes: recalculating the second fault propagation relationship in the second fault propagation relationship model according to the updated first fault propagation relationship model. The calculation method refers to the relevant content of generating the second fault propagation relationship in the second fault propagation relationship model mentioned above. The detailed description is omitted here. The more the number of first fault propagation relationships in the first fault propagation relationship model, the higher the accuracy of the calculated second fault propagation relationship, which is conducive to the accuracy of the possible fault propagation relationship determined by the subsequent use of the second fault propagation relationship model.

[0250] In an embodiment of the present application, updating of the first fault propagation relationship model and the second fault propagation relationship model is performed by a cloud device. After the cloud device updates the first fault propagation relationship model and the second fault propagation relationship model, the updated first fault propagation relationship model and the second fault propagation relationship model are sent to the network management device, so that the network management device can subsequently determine the propagation relationship between abnormal events according to the new first fault propagation relationship model and the second fault propagation relationship model.

[0251] When a fault occurs at a certain location in the network, it often causes abnormalities at other locations, which in turn causes multiple network devices to report information about multiple abnormal events. To this end, the present application also provides a network fault analysis method for determining the root cause events among multiple abnormal events, and determining the root cause entity corresponding to the root cause event, so that the network fault can be located. Figure 7 FIG. 1 is a flow chart of another network fault analysis method provided by an exemplary embodiment of the present application. Figure 7 As shown, the method includes the following processes.

[0252] 701: Determine entities associated with multiple abnormal events and obtain an entity set.

[0253] When multiple abnormal event information is received, entities associated with the multiple abnormal events are determined. The relevant contents of the multiple abnormal event information and the method for determining entities associated with the abnormal events are described in 201 above, and detailed description is omitted here.

[0254] The multiple abnormal events are associated with multiple entities, and each entity is associated with at least one abnormal event.

[0255] Any entity in the entity set has at least one associated entity belonging to the entity set. In the embodiment of the present application, the associated entity refers to an entity whose hop count with the corresponding entity is within the set hop count. For example, if the set hop count is 2 hops, the hop count between the associated entity and the corresponding entity is 0 hops, 1 hop, or 2 hops.

[0256] In the embodiment of the present application, the types of entities include but are not limited to interfaces, devices, protocols, and services, and one network device may correspond to multiple entities.

[0257] 702: Determine the fault propagation direction of the seventh abnormal event associated with the third entity and the eighth abnormal event associated with the fourth entity under the corresponding fault cause description information.

[0258] The third entity and the fourth entity are any two mutually related entities in the entity set. The seventh abnormal event is any abnormal event in at least one abnormal event associated with the third entity. The eighth abnormal event is any abnormal event in at least one abnormal event associated with the fourth entity.

[0259] The fault cause description information corresponding to the seventh abnormal event is the seventh fault cause description information. The fault cause description information corresponding to the eighth abnormal event is the eighth fault cause description information. The fault propagation direction is used to indicate the cause event and the effect event in the seventh abnormal event corresponding to the seventh fault cause description information and the eighth abnormal event corresponding to the eighth fault cause description information.

[0260] As before, each abnormal event corresponds to an event identifier and a description of the cause of the fault. In some examples, the fault cause description information corresponding to abnormal events with different event identifiers is the same, for example, the event identifiers of the seventh abnormal event and the eighth abnormal event are different, but the seventh fault cause description information and the eighth fault cause description information are the same. In other examples, the fault cause description information corresponding to abnormal events with different event identifiers is different, for example, the event identifiers of the seventh abnormal event and the eighth abnormal event are different, and the seventh fault cause description information and the eighth fault cause description information are also different. In some other examples, the fault cause description information corresponding to abnormal events with the same event identifier is different, for example, the event identifiers of the seventh abnormal event and the eighth abnormal event are the same, and the seventh fault cause description information and the eighth fault cause description information are also different.

[0261] The fault propagation direction of any two abnormal events in 702 under the corresponding fault cause description information is Figure 2 In a related embodiment, a method for determining a fault propagation direction between a first abnormal event corresponding to the first fault cause description information and a second abnormal event corresponding to the second fault cause description information.

[0262] In some examples, all fault propagation directions are determined using the first fault propagation relationship model. In other examples, a portion of the fault propagation directions are determined using the first fault propagation relationship model, and another portion of the fault propagation directions are determined using the second fault propagation relationship model.

[0263] In some examples, the fault propagation directions of two abnormal events associated with the associated entity pairs under the corresponding fault cause description information are determined in the order of increasing hop counts between the associated entity pairs. For example, the fault propagation directions of two abnormal events on the associated entity pair (i.e., the same entity) with a hop count of 0 under the corresponding fault cause description information are first determined. Then, the fault propagation directions of two abnormal events on the associated entity pair with a hop count of 1 under the corresponding fault cause description information are determined. And so on, until the fault propagation directions of abnormal events on all associated entity pairs under the corresponding fault description information are determined.

[0264] 703: Determine a root cause event among the multiple abnormal events according to the determined fault propagation direction.

[0265] Exemplarily, 703 includes the following two steps.

[0266] The first step is to determine the reference value of each abnormal event based on the determined fault propagation direction.

[0267] The reference value is the sum of the weights of the fault propagation directions of the abnormal event to which it belongs as the result event. The weight of the fault propagation direction is used to indicate the probability that the corresponding fault propagation direction is true. In an embodiment of the present application, when the fault propagation direction is determined according to the first fault relationship model, the weight of the fault propagation direction is 1. When the fault propagation direction is determined according to the second fault propagation relationship model, the weight of the fault propagation direction is the probability corresponding to the corresponding second fault propagation relationship.

[0268] The second step is to determine the root cause events among multiple abnormal events based on the reference values ​​of each abnormal event.

[0269] In the embodiment of the present application, the abnormal event with the smallest reference value is determined as the root cause event among the multiple abnormal events. The smallest reference value means that there are basically no other abnormal events that cause the abnormal event with the smallest reference value to occur.

[0270] In the embodiment of the present application, the number of root cause events determined is one or more. For example, the same cause triggers the network device to report multiple abnormal events associated with the same entity (carrying different fault cause description information), and these multiple abnormal events are all root cause events.

[0271] In the embodiment of the present application, the entity associated with the root cause event is the root cause entity. After the root cause event and the root cause entity are determined, the fault analysis result can be notified to the technician through prompt information to facilitate the technician to troubleshoot the fault.

[0272] Figure 8 The structure diagram of the network fault analysis device provided in the embodiment of the present application is shown in FIG. The device can be implemented as part or all of the device through software, hardware, or a combination of both. The device provided in the embodiment of the present application can implement the embodiment of the present application. Figure 2 The process, such as Figure 8 As shown, the device includes: an acquisition module 810, a first determination module 820 and a second determination module 830.

[0273] Among them, the acquisition module 810 is used to acquire information about the first abnormal event and information about the second abnormal event. The first determination module 820 is used to determine the first fault cause description information based on the information about the first abnormal event acquired by the acquisition module 810 and to determine the second fault cause description information based on the information about the second abnormal event acquired by the acquisition module 810, the first fault cause description information being used to describe the cause of the first abnormal event, and the second fault cause description information being used to describe the cause of the second abnormal event. The second determination module 830 is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information based on the event identifier in the information about the first abnormal event, the event identifier in the information about the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0274] In a possible implementation, the first determination module 820 includes: a template acquisition submodule 8201, an extraction submodule 8202, and an information determination submodule 8203. The template acquisition submodule 8201 is used to acquire a first extraction template of a first network device and a second extraction template of a second network device, the first network device is a network device that sends information about a first abnormal event, and the second network device is a network device that sends information about a second abnormal event, and the first extraction template and the second extraction template both include an event identifier and a fault code keyword. The extraction submodule 8202 is used to extract a fault code from the information about the first abnormal event according to the same event identifier and fault code keyword in the first extraction template and the information about the first abnormal event, to obtain a first extraction result; and to extract a fault code from the information about the second abnormal event according to the same event identifier and fault code keyword in the second extraction template and the information about the second abnormal event, to obtain a second extraction result. The information determination submodule 8203 is used to determine the first fault cause description information according to the first extraction result, and to determine the second fault cause description information according to the second extraction result.

[0275] In one possible embodiment, at least one of the first extraction result and the second extraction result includes at least one fault code, and the information determination submodule 8204 is used to combine the fault codes and corresponding fault code keywords contained in the first extraction result and the second extraction result, each combination into a fault cause description information, thereby obtaining at least one fault cause description information.

[0276] In a possible implementation, the second determination module 830 includes: a first model acquisition submodule 8301 and a first relationship determination submodule 8302. The first model acquisition submodule 8301 is used to acquire a first fault propagation relationship model, the first fault propagation relationship model includes multiple first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information. The first relationship determination submodule 8302 is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the first target fault propagation relationship in response to determining that there is a first target fault propagation relationship in the first fault propagation relationship model, the first target fault propagation relationship is a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0277] In another possible implementation, the second determination module 830 includes: a first model acquisition submodule 8301, a second model acquisition submodule 8303, and a second relationship determination submodule 8304. The first model acquisition submodule 8301 is used to acquire a first fault propagation relationship model, the first fault propagation relationship model includes multiple first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under the corresponding fault cause description information. The second model acquisition submodule 8303 is used to acquire a second fault propagation relationship model in response to determining that the first target fault propagation relationship does not exist in the first fault propagation relationship model, the first target fault propagation relationship is a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, the second fault propagation relationship model includes multiple second fault propagation relationships, the second fault propagation relationship is used to record the possible fault propagation direction of two abnormal events under the corresponding fault cause description information, and the second fault propagation relationship model is generated according to the first fault propagation relationship model. The second relationship determination submodule 8304 is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information in response to determining that there is a second target fault propagation relationship in the second fault propagation relationship model, according to the second target fault propagation relationship, the second target fault propagation relationship being a second fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

[0278] Optionally, the device further includes: a verification module 840 and an updating module 850. The verification module 840 is used to verify the second target fault propagation relationship after the second relationship determination submodule 8304 determines the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship. The updating module 850 is used to update the first fault propagation relationship model and the second fault propagation relationship model according to the verified second target fault propagation relationship.

[0279] Optionally, the device further includes a first generating module 860. The first generating module 860 is used to generate a first fault propagation relationship in the first fault propagation relationship model in the following manner: determining a third abnormal event on the first entity and third fault cause description information for describing the cause of the third abnormal event, and a fourth abnormal event on the second entity and fourth fault cause description information for describing the cause of the fourth abnormal event, the first entity and the second entity being associated with each other; determining, according to expert annotations, the cause event and the effect event in the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information, to obtain the first fault propagation relationship.

[0280] Optionally, the device also includes a second generation module 870. The second generation module 870 is used to generate a second fault propagation relationship in the second fault propagation relationship model according to the following method: determine a fifth abnormal event with unknown fault cause description information, the unknown fault cause description information is the fault cause description information of the corresponding abnormal event other than the known fault cause description information, and the known fault cause description information is the fault cause description information in the first fault propagation relationship associated with the corresponding abnormal event; calculate the first similarity between the unknown fault cause description information of the fifth abnormal event and each known fault cause description information of the fifth abnormal event, respectively, to obtain a first similarity set; according to the first similarity set, determine the probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established. For details, please refer to the previous text, and the detailed description is omitted here.

[0281] Optionally, the device also includes: a third generation module 880, used to determine cause description keywords in the product manual of the target network device; based on the cause description keywords, extract template content from the product manual of the target network device; generate a target extraction template according to the template content; wherein, the target network device is a first network device, and the target extraction template is a first extraction template; or, the target network device is a second network device, and the target extraction template is a second extraction template.

[0282] In an embodiment of the present application, first fault cause description information for describing the cause of the occurrence of the first abnormal event is determined based on information about the first abnormal event, and second fault cause description information for describing the cause of the occurrence of the second abnormal event is determined based on information about the second abnormal event. When considering the fault propagation relationship between the first abnormal event and the second abnormal event, not only the identifier of the first abnormal event and the identifier of the second abnormal event are considered, but also the causes of the occurrence of the first abnormal event and the second abnormal event are considered, so as to determine that the first abnormal event corresponding to the first fault cause description information is the cause of the occurrence of the second abnormal event corresponding to the second fault cause description information, so that the determination result of the fault propagation relationship is more refined and accurate, thereby improving the accuracy of network fault analysis.

[0283] Fig. 9 The structure diagram of the network fault analysis device provided in the embodiment of the present application is shown in FIG. The device can be implemented as part or all of the device through software, hardware, or a combination of both. The device provided in the embodiment of the present application can implement the embodiment of the present application. Figure 7 The process, such as Fig. 9 As shown, the device includes: an entity determination module 910, a direction determination module 920 and a root cause event determination module 930. The entity determination module 910 is used to determine entities associated with multiple abnormal events to obtain an entity set. The direction determination module 920 is used to adopt Figure 2 The method shown determines the fault propagation direction of the seventh abnormal event associated with the third entity and the eighth abnormal event associated with the fourth entity under the corresponding fault cause description information, wherein the third entity and the fourth entity are any two mutually associated entities in the entity set, the seventh abnormal event is any abnormal event of at least one abnormal event associated with the third entity, and the eighth abnormal event is any abnormal event of at least one abnormal event associated with the fourth entity. The root cause event determination module 930 is used to determine the root cause event among multiple abnormal events according to the determined fault propagation direction.

[0284] In some examples, the root cause event determination module 930 includes a reference value determination submodule 931 and a root cause event determination submodule 932. The reference value determination submodule 931 is used to determine the reference value of each abnormal event according to the determined fault propagation direction, and the reference value is the sum of the weights of the fault propagation directions of the abnormal event to which it belongs as the result event, and the weight of the fault propagation direction is used to indicate the probability that the corresponding fault propagation direction is true. The root cause event determination submodule 932 is used to determine the abnormal event with the smallest reference value as the root cause event among the multiple abnormal events.

[0285] The division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present application may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0286] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a terminal device (which can be a personal computer, mobile phone, or network device, etc.) or a processor (processor) to execute all or part of the steps of the method of each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0287] It should be noted that: the network fault analysis device provided in the above embodiment only uses the division of the above functional modules as an example when performing network fault analysis. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the network fault analysis device provided in the above embodiment and the network fault analysis method embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0288] A computer device is also provided in an embodiment of the present application. Fig.10 A possible architecture diagram of a computer device 1000 is provided by way of example.

[0289] The computer device 1000 includes a memory 1001, a processor 1002, a communication interface 1003 and a bus 1004. The memory 1001, the processor 1002 and the communication interface 1003 are connected to each other through the bus 1004.

[0290] The memory 1001 may be a read-only memory (ROM), a static storage device, a dynamic storage device or a random access memory (RAM). The memory 1001 may store a program. When the program stored in the memory 1001 is executed by the processor 1002, the processor 1002 and the communication interface 1003 are used to execute the network fault analysis method. The memory 1001 may also store a data set. For example, a portion of the storage resources in the memory 1001 is divided into a data set storage module for storing an extraction template, a first fault propagation relationship model and a second fault propagation relationship model.

[0291] The processor 1002 may be a general-purpose central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or one or more integrated circuits.

[0292] The processor 1002 may also be an integrated circuit chip with signal processing capability. In the implementation process, some or all of the functions of the network fault analysis device of the present application may be completed by the hardware integrated logic circuit or software instructions in the processor 1002. The above-mentioned processor 1002 may also be a general-purpose processor, a digital signal processor (Digital Signal Processing, DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods disclosed in the above-mentioned embodiments of the present application may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present application may be directly embodied as being executed by a hardware decoding processor, or may be executed by a combination of hardware and software modules in a decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 1001, and the processor 1002 reads the information in the memory 1001 and completes part of the functions of the network fault analysis device of the embodiment of the present application in combination with its hardware.

[0293] The communication interface 1003 uses a transceiver module such as, but not limited to, a transceiver to implement communication between the computer device 1000 and other devices or communication networks. For example, the communication interface 1003 can be used to obtain information about abnormal events, extract templates, a first fault propagation relationship model, and a second fault propagation relationship model.

[0294] The bus 1004 may include a path for transmitting information between various components of the computer device 1000 (eg, the memory 1001 , the processor 1002 , and the communication interface 1003 ).

[0295] The descriptions of the processes corresponding to the above-mentioned figures have different emphases. For parts that are not described in detail in a certain process, please refer to the relevant descriptions of other processes.

[0296] In an embodiment of the present application, a computer-readable storage medium is also provided, which stores computer instructions. When the computer instructions stored in the computer-readable storage medium are executed by a computer device, the computer device executes the network fault analysis method provided above.

[0297] In an embodiment of the present application, a computer program product including instructions is also provided, which, when executed on a computer device, enables the computer device to execute the network fault analysis method provided above.

[0298] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a server or terminal, the process or function described in the embodiment of the present application is generated in whole or in part. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a server or terminal or a data storage device such as a server or data center that includes one or more available media integration. The available medium can be a magnetic medium (such as a floppy disk, a hard disk and a tape, etc.), or an optical medium (such as a digital video disk (Digital Video Disk, DVD), etc.), or a semiconductor medium (such as a solid-state hard disk, etc.).

Claims

1. A network fault analysis method, characterized in that: The method comprises: Acquire information about the first abnormal event and information about the second abnormal event; Determine first fault cause description information according to the information of the first abnormal event, where the first fault cause description information is used to describe the cause of the first abnormal event; Determine second fault cause description information according to the information of the second abnormal event, where the second fault cause description information is used to describe the cause of the second abnormal event; Acquire a first fault propagation relationship model, where the first fault propagation relationship model includes a plurality of first fault propagation relationships, where the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under corresponding fault cause description information; In response to determining that the first target fault propagation relationship does not exist in the first fault propagation relationship model, a second fault propagation relationship model is acquired, the first target fault propagation relationship being a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, the second fault propagation relationship model including a plurality of second fault propagation relationships, the second fault propagation relationship being used to record possible fault propagation directions of two abnormal events under corresponding fault cause description information, and the second fault propagation relationship model being generated according to the first fault propagation relationship model; In response to determining that there is a second target fault propagation relationship in the second fault propagation relationship model, the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information is determined according to the second target fault propagation relationship, and the second target fault propagation relationship is a second fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

2. The method according to claim 1, characterized in that The determining the first fault cause description information according to the information of the first abnormal event includes: Acquire a first extraction template of a first network device, where the first network device is a network device that sends information about the first abnormal event, and the first extraction template includes an event identifier and a fault code keyword; Extracting the fault code from the information of the first abnormal event according to the same event identifier and fault code keyword in the first extraction template and the information of the first abnormal event to obtain a first extraction result; Determine the first fault cause description information according to the first extraction result; The determining the second fault cause description information according to the information of the second abnormal event includes: Acquire a second extraction template of a second network device, where the second network device is a network device that sends information about the second abnormal event, and the second extraction template includes an event identifier and a fault code keyword; extracting the fault code from the information of the second abnormal event according to the same event identifier and fault code keyword in the second extraction template and the information of the second abnormal event to obtain a second extraction result; The second fault cause description information is determined according to the second extraction result.

3. The method according to claim 1, characterized in that: The method further comprises: In response to determining that the first target fault propagation relationship exists in the first fault propagation relationship model, the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information is determined according to the first target fault propagation relationship.

4. The method according to claim 1, characterized in that: The method further comprises: After determining the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship, verifying the second target fault propagation relationship; The first fault propagation relationship model and the second fault propagation relationship model are updated according to the verified second target fault propagation relationship.

5. The method according to claim 1, characterized in that The first fault propagation relationship in the first fault propagation relationship model is generated in the following manner: Determine a third abnormal event on a first entity and third fault cause description information for describing a cause of the third abnormal event, and a fourth abnormal event on a second entity and fourth fault cause description information for describing a cause of the fourth abnormal event, wherein the first entity and the second entity are associated with each other; According to the expert annotation, the cause event and the effect event in the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information are determined to obtain the first fault propagation relationship.

6. The method according to any one of claims 1 to 5, characterized in that: The second fault propagation relationship in the second fault propagation relationship model is generated in the following manner: Determine a fifth abnormal event having unknown fault cause description information, where the unknown fault cause description information is fault cause description information of the corresponding abnormal event other than known fault cause description information, where the known fault cause description information is fault cause description information in a first fault propagation relationship associated with the corresponding abnormal event; respectively calculating first similarities between the unknown fault cause description information of the fifth abnormal event and each known fault cause description information of the fifth abnormal event to obtain a first similarity set; The probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established is determined according to the first similarity set.

7. The method according to claim 6, characterized in that The determining, based on the first similarity set, a probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established includes at least one of the following: Determine, according to a first similarity greater than a first similarity threshold in the first similarity set and a second similarity greater than a second similarity threshold in the second similarity set, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event is established, wherein the second similarity set is a set of second similarities between the unknown fault cause description information of the sixth abnormal event and each known fault cause description information of the sixth abnormal event; Determine, according to a first similarity in the first similarity set that is greater than a first similarity threshold, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established; In response to determining that all first similarities in the first similarity set are not greater than a first similarity threshold, determining, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event, is established; In response to determining that the first similarities in the first similarity set are not greater than a first similarity threshold, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, determine the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established.

8. The method according to claim 7, characterized in that The determining, according to a first similarity in the first similarity set that is greater than a first similarity threshold and a second similarity in the second similarity set that is greater than a second similarity threshold, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event is established, includes: Determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; Determining a first weight corresponding to each first fault propagation relationship in the first relationship set according to a first similarity and a second similarity corresponding to each first fault propagation relationship in the first relationship set; Determine, according to the first similarity and the second similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set, the second weight corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set; Calculate the probability that the sixth abnormal event is caused by the fifth abnormal event according to the sum of the first weights and the sum of the second weights corresponding to each first fault propagation relationship in the first relationship set; or, Determining, according to a first similarity in the first similarity set that is greater than a first similarity threshold and a second similarity in the second similarity set that is greater than a second similarity threshold, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event is established, includes: Determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; Determining a first weight corresponding to each first fault propagation relationship in the first relationship set according to a first similarity and a second similarity corresponding to each first fault propagation relationship in the first relationship set; Determine, according to the first similarity and the second similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set, a fifth weight corresponding to the first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event in the first relationship set; The probability that the sixth abnormal event causes the fifth abnormal event is calculated according to the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the fifth weights.

9. The method according to claim 7, characterized in that: The determining, according to the first similarity greater than the first similarity threshold in the first similarity set, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established, comprises: Determine a first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to a first similarity greater than a first similarity threshold, to obtain a second relationship set; determining, according to the first similarities corresponding to the first fault propagation relations in the second relationship set, a first weight corresponding to each first fault propagation relation in the second relationship set; Determine, according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event, a second weight corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event; Calculate the probability that the sixth abnormal event is caused by the fifth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the second weights; or, The determining, according to the first similarity greater than the first similarity threshold in the first similarity set, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established, comprises: Determine a first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to a first similarity greater than a first similarity threshold, to obtain a second relationship set; determining, according to the first similarities corresponding to the first fault propagation relations in the second relationship set, a first weight corresponding to each first fault propagation relation in the second relationship set; Determine, according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event, a fifth weight corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event; The probability that the sixth abnormal event causes the fifth abnormal event is calculated according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the fifth weights.

10. The method according to claim 2, characterized in that The method further comprises: Identify the cause description keywords in the product manual of the target network device; Based on the cause description keywords, extracting template content from the product manual of the target network device; Generate a target extraction template according to the template content; The target network device is the first network device, and the target extraction template is the first extraction template; or the target network device is the second network device, and the target extraction template is the second extraction template.

11. A network fault analysis device, characterized in that: The device comprises: An acquisition module, used to acquire information about the first abnormal event and information about the second abnormal event; a first determining module, configured to determine first fault cause description information according to information about the first abnormal event, and to determine second fault cause description information according to information about the second abnormal event, wherein the first fault cause description information is used to describe a cause of the first abnormal event, and the second fault cause description information is used to describe a cause of the second abnormal event; a second determination module, configured to determine, based on an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, a fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information; Wherein, the second determining module includes: A first model acquisition submodule is used to acquire a first fault propagation relationship model, wherein the first fault propagation relationship model includes a plurality of first fault propagation relationships, and the first fault propagation relationship is used to record the fault propagation direction of two abnormal events under corresponding fault cause description information; A second model acquisition submodule is used for acquiring a second fault propagation relationship model in response to determining that a first target fault propagation relationship does not exist in the first fault propagation relationship model, wherein the first target fault propagation relationship is a first fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information, the second fault propagation relationship model includes a plurality of second fault propagation relationships, the second fault propagation relationship is used to record possible fault propagation directions of two abnormal events under corresponding fault cause description information, and the second fault propagation relationship model is generated according to the first fault propagation relationship model; A second relationship determination submodule is used to determine, in response to determining that there is a second target fault propagation relationship in the second fault propagation relationship model, a fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship, wherein the second target fault propagation relationship is a second fault propagation relationship including an event identifier in the information of the first abnormal event, an event identifier in the information of the second abnormal event, the first fault cause description information, and the second fault cause description information.

12. The device according to claim 11, characterized in that The first determining module comprises: a template acquisition submodule, configured to acquire a first extraction template of a first network device and a second extraction template of a second network device, wherein the first network device is a network device that sends information about the first abnormal event, and the second network device is a network device that sends information about the second abnormal event, and both the first extraction template and the second extraction template include an event identifier and a fault code keyword; an extraction submodule, configured to extract a fault code from the information of the first abnormal event according to the same event identifier and fault code keyword in the first extraction template and the information of the first abnormal event, to obtain a first extraction result; and to extract a fault code from the information of the second abnormal event according to the same event identifier and fault code keyword in the second extraction template and the information of the second abnormal event, to obtain a second extraction result; An information determination submodule is used to determine the first fault cause description information according to the first extraction result, and to determine the second fault cause description information according to the second extraction result.

13. The device according to claim 11, characterized in that The second determining module further includes: The first relationship determination submodule is used to determine the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the first target fault propagation relationship in response to determining that the first target fault propagation relationship exists in the first fault propagation relationship model.

14. The device according to claim 11, characterized in that The device also includes: a verification module, configured to verify the second target fault propagation relationship after the second relationship determination submodule determines the fault propagation direction between the first abnormal event corresponding to the first fault cause description information and the second abnormal event corresponding to the second fault cause description information according to the second target fault propagation relationship; An updating module is used to update the first fault propagation relationship model and the second fault propagation relationship model according to the verified second target fault propagation relationship.

15. The device according to claim 11, characterized in that The device also includes: The first generating module is configured to generate a first fault propagation relationship in the first fault propagation relationship model in the following manner: Determine a third abnormal event on a first entity and third fault cause description information for describing a cause of the third abnormal event, and a fourth abnormal event on a second entity and fourth fault cause description information for describing a cause of the fourth abnormal event, wherein the first entity and the second entity are associated with each other; According to the expert annotation, the cause event and the effect event in the third abnormal event corresponding to the third fault cause description information and the fourth abnormal event corresponding to the fourth fault cause description information are determined to obtain the first fault propagation relationship.

16. The device according to any one of claims 11 to 15, characterized in that The device also includes: The second generating module is used to generate the second fault propagation relationship in the second fault propagation relationship model in the following manner: Determine a fifth abnormal event having unknown fault cause description information, where the unknown fault cause description information is fault cause description information of the corresponding abnormal event other than known fault cause description information, where the known fault cause description information is fault cause description information in a first fault propagation relationship associated with the corresponding abnormal event; respectively calculating first similarities between the unknown fault cause description information of the fifth abnormal event and each of the known fault cause description information of the fifth abnormal event to obtain a first similarity set; The probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established is determined according to the first similarity set.

17. The device according to claim 16, characterized in that The second generating module is configured to determine, according to the first similarity set, a probability that the second fault propagation relationship corresponding to the fifth abnormal event under the unknown fault cause description information is established in accordance with at least one of the following implementations: Determine, according to a first similarity greater than a first similarity threshold in the first similarity set and a second similarity greater than a second similarity threshold in the second similarity set, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event is established, wherein the second similarity set is a set of second similarities between the unknown fault cause description information of the sixth abnormal event and each known fault cause description information of the sixth abnormal event; Determine, according to a first similarity in the first similarity set that is greater than a first similarity threshold, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established; In response to determining that all first similarities in the first similarity set are not greater than a first similarity threshold, determining, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, a probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the unknown fault cause description information of the sixth abnormal event, is established; In response to determining that the first similarities in the first similarity set are not greater than a first similarity threshold, according to the number of first fault propagation relationships associated with the fifth abnormal event and the sixth abnormal event, determine the probability that a second fault propagation relationship between the fifth abnormal event and the sixth abnormal event, which is based on the unknown fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event, is established.

18. The device according to claim 17, characterized in that The second generating module is used to determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; determine a first weight corresponding to each first fault propagation relationship in the first relationship set according to the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; Determine the second weight corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set according to the first similarity and the second similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set; calculate the probability that the fifth abnormal event causes the sixth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the second weights; or The second generating module is used to determine a first fault propagation relationship between a fifth abnormal event associated with known fault cause description information corresponding to a first similarity greater than a first similarity threshold and a sixth abnormal event associated with known fault cause description information corresponding to a second similarity greater than a second similarity threshold, to obtain a first relationship set; determine a first weight corresponding to each first fault propagation relationship in the first relationship set according to the first similarity and the second similarity corresponding to each first fault propagation relationship in the first relationship set; Determine, according to the first similarity and the second similarity corresponding to the first fault propagation relationship indicating that the fifth abnormal event causes the sixth abnormal event in the first relationship set, a fifth weight corresponding to the first fault propagation relationship indicating that the sixth abnormal event causes the fifth abnormal event in the first relationship set; The probability that the sixth abnormal event causes the fifth abnormal event is calculated according to the sum of the first weights corresponding to each first fault propagation relationship in the first relationship set and the sum of the fifth weights.

19. The device according to claim 17, characterized in that The second generating module is used to determine a first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to a first similarity greater than a first similarity threshold, to obtain a second relationship set; determining, according to the first similarities corresponding to the first fault propagation relations in the second relationship set, a first weight corresponding to each first fault propagation relation in the second relationship set; Determine, according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event, a second weight corresponding to the first fault propagation relationship in the second relationship set indicating that the fifth abnormal event causes the sixth abnormal event; Calculate the probability that the sixth abnormal event is caused by the fifth abnormal event according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the second weights; or, The second generating module is used to determine a first fault propagation relationship associated with the known fault cause description information of the fifth abnormal event and the known fault cause description information of the sixth abnormal event corresponding to a first similarity greater than a first similarity threshold, to obtain a second relationship set; determining, according to the first similarities corresponding to the first fault propagation relations in the second relationship set, a first weight corresponding to each first fault propagation relation in the second relationship set; Determine, according to the first similarity corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event, a fifth weight corresponding to the first fault propagation relationship in the second relationship set indicating that the sixth abnormal event causes the fifth abnormal event; The probability that the sixth abnormal event causes the fifth abnormal event is calculated according to the sum of the first weights corresponding to each first fault propagation relationship in the second relationship set and the sum of the fifth weights.

20. The device according to claim 12, characterized in that The device also includes: A third generation module is used to determine the cause description keywords in the product manual of the target network device; based on the cause description keywords, extract template content from the product manual of the target network device; and generate a target extraction template according to the template content; The target network device is the first network device, and the target extraction template is the first extraction template; or the target network device is the second network device, and the target extraction template is the second extraction template.

21. A computer device, characterized in that: The computer device comprises a processor and a memory, wherein: The memory stores computer instructions; The processor executes the computer instructions to implement the method described in any one of claims 1 to 10.

22. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions. When the computer instructions in the computer-readable storage medium are executed by a computer device, the computer device executes the method according to any one of claims 1 to 10.