System exception root cause positioning method, device, equipment and medium

Through the combination of large models and detection tool modules, the root causes of system abnormalities are automatically analyzed, and the problem of relying on artificially defining causal relationships in the existing technology is solved, and the degree of automation and accuracy of system abnormal positioning is improved.

CN118981389APending Publication Date: 2024-11-19BEIJING YOUTEJIE INFORMATION TECH
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411119417.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

The existing root cause positioning method relies on artificially defined causal relationships, has low degree of automation, and cannot be widely used in the positioning of system abnormalities.

Method used

By obtaining the system abnormal root cause analysis request, sending it to the large model for resampling mechanism analysis, obtaining the initial completion data, and combining the detection tool module to determine the system abnormal analysis result data, to realize automatic analysis of the root cause of the system abnormality.

Benefits of technology

It realizes automated analysis of the root causes of system abnormalities, does not rely on artificially defined cause and effect, improves the degree of automation and system universality, and can more accurately determine the root causes of system abnormalities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118981389A_ABST
    Figure CN118981389A_ABST
Patent Text Reader

Abstract

The invention discloses a root cause positioning method and device for system abnormity, equipment and a medium. The system exception root cause positioning method comprises the following steps: acquiring a system exception root cause analysis request, and sending the system exception root cause analysis request to a large model; analyzing an abnormal root cause analysis request of the system according to a resampling mechanism through the large model to obtain initial completion data; and determining system anomaly analysis result data based on the initial completion data, the large model and the detection tool module. According to the technical scheme provided by the embodiment of the invention, the root cause of the system abnormality can be better determined, a complete human-like cause determination process is provided, and the automation degree and the universality of the system are higher.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and medium for locating the root cause of a system anomaly. Background Art

[0002] In cloud-native technology, the increasing complexity of microservice architecture has brought huge challenges to the stability and efficiency maintenance of the system.

[0003] Currently, GROOT (root cause analysis method based on event graph) can summarize various indicators, logs and activities in system analysis, build a real-time causal relationship graph, and analyze the causal relationship between events through the real-time causal graph to locate the root cause, but the root cause location effect is not good. In order to combine the domain knowledge from network reliability engineers, GROOT can be customized through user-defined events and domain-specific rules, but it relies heavily on manually defined causality, has a low degree of automation, and is not suitable for promotion to a wider range of systems. Summary of the invention

[0004] The present invention provides a method, device, equipment and medium for locating the root cause of system anomalies to solve the problem that the existing root cause locating methods rely on artificially defined causes and conditions and cannot be automatically and widely applied to the system.

[0005] According to one aspect of the present invention, a method for locating a root cause of a system abnormality is provided, comprising:

[0006] Obtain a system abnormality root cause analysis request, and send the system abnormality root cause analysis request to the big model;

[0007] Through the large model, the root cause analysis request of the system anomaly is analyzed according to the resampling mechanism to obtain the initial completion data;

[0008] Based on the initial completion data, large model and detection tool module, determine the system anomaly analysis result data.

[0009] According to another aspect of the present invention, a device for locating the root cause of a system abnormality is provided, comprising:

[0010] The request acquisition and sending module is used to acquire the system abnormality root cause analysis request and send the system abnormality root cause analysis request to the big model;

[0011] The initial completion data determination module is used to analyze the root cause analysis request of the system abnormality through the large model according to the resampling mechanism to obtain the initial completion data;

[0012] The system anomaly analysis result data determination module is used to determine the system anomaly analysis result data based on the initial completion data, the large model and the detection tool module.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the root cause locating method of system abnormality described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the root cause locating method of a system abnormality described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention obtains a system abnormality root cause analysis request, and sends the system abnormality root cause analysis request to a large model, and then analyzes the system abnormality root cause analysis request according to a resampling mechanism through the large model to obtain initial completion data, and further determines the system abnormality analysis result data based on the initial completion data, the large model, and the detection tool module. In this solution, the large model analyzes the system abnormality root cause analysis request based on the resampling mechanism, so that the large model can be repeatedly executed multiple times to obtain as rich initial completion data as possible to support multi-directional reasoning, and based on the initial completion data, the large model, and the detection tool module, the system abnormality analysis result data is determined, which can realize automatic analysis of the root cause of the system abnormality, without relying on artificially defined causality, and solves the problem that the existing root cause location method relies on artificially defined causality and cannot be automatically and widely used in the system, and can better determine the root cause of the system abnormality, and has a complete human-like cause determination process, and the degree of automation and system universality are higher.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A flowchart of a method for locating the root cause of a system abnormality provided in Embodiment 1 of the present invention;

[0022] Figure 2 A flowchart of a method for locating the root cause of a system abnormality provided in Embodiment 2 of the present invention;

[0023] Figure 3 It is a flow chart of a method for locating the root cause of an abnormality of a multi-agent module execution system provided in Embodiment 2 of the present invention;

[0024] Figure 4 A schematic diagram of the structure of a device for locating the root cause of a system abnormality provided in Embodiment 3 of the present invention;

[0025] Figure 5 A schematic diagram of the structure of an electronic device that can be used to implement an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "initial", "target", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Embodiment 1

[0029] Figure 1 This is a flowchart of a method for locating the root cause of a system abnormality provided in the first embodiment of the present invention. This embodiment is applicable to the case of automatic and accurate root cause locating. The method can be executed by a root cause locating device for a system abnormality. The root cause locating device for the system abnormality can be implemented in the form of hardware and / or software. The root cause locating device for the system abnormality can be configured in an electronic device. Figure 1 As shown, the method includes:

[0030] Step 110: Obtain a system abnormality root cause analysis request, and send the system abnormality root cause analysis request to the big model.

[0031] Among them, the system anomaly root cause analysis request can be an analysis request for the root cause of the system anomaly. Exemplarily, the format of the system anomaly root cause analysis request can be "At XXX time, is there an anomaly in the system? If so, what is the root cause?" A large model refers to a deep learning model with a huge number of parameters and a complex structure, such as GPT-3. Large models require a lot of computing resources and data during training, but have better language understanding and generation capabilities. They can be applied to many fields such as natural language processing and computer vision to complete tasks such as text generation, translation, and summarization.

[0032] In an embodiment of the present invention, when a system operating abnormality occurs, a system abnormality root cause analysis request issued by the abnormal system can be obtained, and the system abnormality root cause analysis request can be sent to the big model to perform auxiliary analysis of the system abnormality through the big model.

[0033] Step 120: Analyze the root cause analysis request of the system exception through the large model according to the resampling mechanism to obtain initial completion data.

[0034] The initial completion data may be the data obtained by the large model for the initial action completion operation in response to the system abnormality root cause analysis request. The initial completion data may include optional actions and action inputs. Optional actions may be understood as actions to select an agent. Action inputs may be used to describe the detection tasks that the agent needs to control and complete.

[0035] In an embodiment of the present invention, the large model may utilize a resampling mechanism to resample and analyze the system abnormality root cause analysis request to output initial completion data.

[0036] Step 130: Determine system anomaly analysis result data based on the initial completion data, the large model, and the detection tool module.

[0037] The detection tool module may be a module composed of detection tools for detecting system abnormality problems. The system abnormality analysis result data may be an analysis result of the abnormal cause of the system corresponding to the system abnormality root cause analysis request.

[0038] In an embodiment of the present invention, the tests to be performed can be determined based on the initial completion data, and then the detection tool module can be called to perform the corresponding tests, and the large model and the detection tool module can be cyclically called based on the detection results until the root cause of the system abnormality is determined, that is, the system abnormality analysis result data.

[0039] The technical solution of the embodiment of the present invention obtains a system abnormality root cause analysis request, and sends the system abnormality root cause analysis request to a large model, and then analyzes the system abnormality root cause analysis request according to a resampling mechanism through the large model to obtain initial completion data, and further determines the system abnormality analysis result data based on the initial completion data, the large model, and the detection tool module. In this solution, the large model analyzes the system abnormality root cause analysis request based on the resampling mechanism, so that the large model can be repeatedly executed multiple times to obtain as rich initial completion data as possible to support multi-directional reasoning, and based on the initial completion data, the large model, and the detection tool module, the system abnormality analysis result data is determined, which can realize automatic analysis of the root cause of the system abnormality, without relying on artificially defined causality, and solves the problem that the existing root cause location method relies on artificially defined causality and cannot be automatically and widely used in the system, and can better determine the root cause of the system abnormality, and has a complete human-like cause determination process, and the degree of automation and system universality are higher.

[0040] Embodiment 2

[0041] Figure 2 This is a flowchart of a method for locating the root cause of a system anomaly provided in the second embodiment of the present invention. This embodiment is specific based on the above embodiment and provides a specific optional implementation method for determining the system anomaly analysis result data based on the initial completion data, the large model and the detection tool module. Figure 2 As shown, the method includes:

[0042] Step 210: Obtain a system abnormality root cause analysis request, and send the system abnormality root cause analysis request to the large model.

[0043] Step 220: Analyze the root cause analysis request of the system exception through the large model according to the resampling mechanism to obtain initial completion data.

[0044] In an optional embodiment of the present invention, before analyzing the system abnormality root cause analysis request according to the resampling mechanism through the big model to obtain the initial supplementary data, it can also include: obtaining historical root cause determination data; and inputting the historical root cause determination data into the big model.

[0045] The historical root cause determination data may be system anomaly analysis results recorded during historical system anomaly root cause analysis.

[0046] In an embodiment of the present invention, before analyzing the system abnormality root cause analysis request according to the resampling mechanism through the big model to obtain the initial completion data, the corresponding system abnormality analysis result data can be determined for each historical system abnormality root cause analysis request and then recorded as historical root cause determination data, and then the historical root cause determination data is input into the big model, so that the big model can perform more accurate inference based on the historical root cause determination data.

[0047] Step 230: Based on the initial completion data, the detection tool module and the large model, determine the current loop call completion data.

[0048] Among them, the current cycle call completion data can be the data obtained by the large model through the action completion operation when performing the current round call based on the initial completion data and the detection tool module.

[0049] In an embodiment of the present invention, after completing preliminary system anomaly detection based on the initial completion data and the detection tool module, a large model and a detection tool model are cyclically called based on the detection results, and the large model outputs the current cyclic call completion data.

[0050] In an optional embodiment of the present invention, based on the initial completion data, the detection tool module and the large model, determining the current loop call completion data can include: based on the root cause agent of the multi-agent module, parsing the initial completion data and determining the target sub-agent; based on the initial completion data, the detection tool module and the target sub-agent, determining the current loop call data to be completed; inputting the current loop call data to be completed into the large model through the root cause agent, and analyzing the current loop call data to be completed through the large model to determine the current loop call completion data.

[0051] Among them, the multi-agent module can be a module composed of multiple agents. The root cause agent is equivalent to the root node in the multi-agent module. The target subagent can be determined by the initial completion data, and the agent responsible for the anomaly detection subtask in the multi-agent module is equivalent to the leaf node in the multi-agent module. The target subagent may include at least one subagent. The current loop calls the data to be completed can be the data input into the big model by the root cause agent before executing the current round of calls to the big model.

[0052] In an embodiment of the present invention, the root cause agent of the multi-agent module can parse the initial completion data, determine the target sub-agent for calling the detection tool using the initial completion data, and then use the target sub-agent to call the detection tool module, and use the execution result of the detection tool module as the current loop call to-be-completed data, thereby sending the current loop call to-be-completed data to the big model based on the root cause agent, and executing the loop call of the big model of the current round, and further analyzing and completing the current loop call to-be-completed data through the big model to obtain the current loop call completion data.

[0053] In an optional embodiment of the present invention, the target subagent may include at least one of a transaction agent, a resource indicator agent, a system information agent, a log agent, and a topology relationship agent.

[0054] Among them, the transaction agent can be used to call the business anomaly detection tool to obtain information on whether the business and service have abnormalities in request volume, response time and success rate by calling the business anomaly detection tool. The resource indicator agent can be used to call the indicator anomaly detection tool to analyze whether the indicators of related resources such as pod (application instance) and host in the system are abnormal through the detection data returned by the indicator anomaly detection tool. The log agent can be used to call the information query tool to analyze whether the relevant logs in the system are abnormal or find the corresponding logs based on other anomalies based on the detection data returned by the information query tool. The topology relationship agent can be used to call the relationship query tool to query the topology data of the current system based on the data returned by the relationship query tool. Exemplarily, the topology data of the current system may include but is not limited to the host where the service is located, the pod where the service is located, the service call relationship, etc., and the query relationship requires a specified time.

[0055] Optionally, the root cause agent and the target subagent may perform data parsing based on a resampling mechanism.

[0056] In an optional embodiment of the present invention, based on the initial completion data, the detection tool module and the target sub-agent, determining the data to be completed for the current cycle call may include: determining the tool instructions to be executed based on the initial completion data and the target sub-agent; and obtaining the data to be completed for the current cycle call based on the tool instructions to be executed and the detection tool module.

[0057] The tool to-be-executed instruction may be an instruction sent by the target subagent to the detection tool module, and is used to call a corresponding detection module of the detection tool module.

[0058] In an embodiment of the present invention, the target subagent can parse the initial completion data and generate execution instructions for the detection that needs to be performed analyzed by the large model, namely, tool instructions to be executed, so as to call the corresponding detection tool in the detection tool module based on the tool instructions to be executed, and use the detection result of the detection tool as the data to be completed in the current cycle.

[0059] Exemplary, assuming that the indicator detection tool in the detection tool module needs to be called at present, the indicator detection tool can parse the tool's pending execution instructions, convert the action input generated by the large model into json format, and extract the indicator data of the corresponding time period (30 minutes before the start time point to 10 minutes after the start time point) from the non-relational database according to the hostname. The specific indicators to be extracted need to be defined in advance through the metric field. The indicator detection tool sets the indicator data resampling to one point every 30 seconds, and different indicators are detected using different detection methods (for example, the CPU usage rate is detected using a 90% threshold). After the detection is completed, it is determined whether an event is generated according to the preset rules (whether there are 6 abnormal points), and all events are output in a list format.

[0060] In an optional embodiment of the present invention, analyzing the data to be completed in the current loop call through a large model to determine the current loop call completion data may include: analyzing the data to be completed in the current loop call through a large model to obtain target action data; sampling and pruning the target action data through a semantic similarity threshold in the large model to obtain the current loop call completion data.

[0061] Among them, the target action data can be used to describe the optional actions that the large model uses to complete the action of the current loop calling the data to be completed.

[0062] In an embodiment of the present invention, the large model analyzes the data to be completed in the current loop call based on the resampling mechanism to obtain the target action data, and then calculates the similarity between the action corresponding to the target action data and the selected action, deletes the actions with similarity greater than the semantic similarity threshold, retains the actions with similarity less than the semantic similarity threshold, and adds them to the set of optional actions to complete the sampling and pruning of the target action data, thereby outputting the current loop call completion data.

[0063] Specifically, the number of optional actions in the set of optional actions may be limited as needed. For example, the maximum number of optional actions in the set of optional actions may be set to 3.

[0064] Step 240: Determine the system abnormality analysis result data according to the current loop call completion data and the loop call end condition of the large model.

[0065] Among them, the end condition of the loop call can be the abnormal problem detected by the big model based on the analysis and detection tool module, which can determine the condition of the root cause of the system abnormality, and represent the condition that there is no need to call the big model for action completion operation. The big model completes all reasoning about the root cause of the system abnormality.

[0066] In an embodiment of the present invention, after calling the detection tool module based on the current loop call completion data, if the large model determines that the detection result of the current round detection tool module is the cause of the system abnormality that needs to be identified in the end, then the loop call end condition of the large model is determined to be met, indicating that the large model has the ability to determine the root cause of the system abnormality, and there is no need to perform action completion operations. System abnormality analysis result data can be generated based on all analysis steps, and the system abnormality analysis result data is used as output. If the large model determines that the detection result based on the current round detection tool module requires further analysis of the cause of the system abnormality, it is determined that the loop call end condition is not met, and the large model is continued to be called in a loop until the system abnormality analysis result data is determined.

[0067] Figure 3 1 is a flowchart of a method for locating the root cause of an abnormality of a multi-agent module execution system provided in the second embodiment of the present invention. Figure 3 As shown, the system anomaly root cause analysis request is sent to the multi-agent module, which automatically analyzes the statement, executes task assignment and calls the detection tool, that is, cyclically calls the large model and the detection tool module to output the system anomaly analysis result data. Among them, the detection tool module also needs to extract the corresponding indicator data from the database when performing the detection task. The large model reasoning is divided into two parts: ideas and actions. Ideas generate analysis steps, and actions generate tool instructions to be executed, and the two are performed alternately until the final result is obtained. In ReAct (an agent mode), the tool call request comes after the reasoning step, which can better support other tools besides search.

[0068] Among them, the structure of the multi-agent module is a tree structure. The root node is the root agent, and there are multiple sub-agents (transaction agent, resource indicator agent, system information agent, log agent, topology relationship agent). Each sub-agent is responsible for a type of sub-task. Then the sub-node can use the detection tool and finally return the result to the root node.

[0069] In the original ReAct, all actions are linear, which is inappropriate in root cause analysis because root cause analysis is a multi-directional process. The multi-agent module ReAct in this solution supports multi-directional reasoning and provides comprehensive automated root cause analysis and solutions, surpassing the existing baseline level.

[0070] In a specific example, the input to the multi-agent module is: "Is there any abnormality in the frontend at 1722936555000? If there is an abnormality, what is the root cause of the system abnormality?"

[0071] Multi-agent module output:

[0072] start-rewoo_copilot

[0073] >Entering new AgentExecutor chain... [ ]

[0076] The idea of ​​the big model is: In order to determine whether the frontend service is abnormal at time 1722936555000, I first need to obtain the business indicators of the frontend service at that time from trade_agent, such as the number of requests, response time, and success rate, to determine whether there is an abnormality. If there is an abnormality, I will need to further analyze the possible root cause, which may involve checking the status of the underlying pod or host resources, and checking the call relationship between services to locate the source of the problem.

[0077] Optional action 1: trade_agent (trading agent)

[0078] Action Input 1: At timestamp 1722936555000, are there any abnormalities in the request volume, response time, and success rate of the frontend service?

[0079] Optional action 2: metric_agent (resource metric agent)

[0080] Action Input 2: At timestamp 1722936555000, are the pod and host resource indicators related to the frontend service normal?

[0081] start-trade_agent

[0082] >Entering new AgentExecutor chain

[0083] I need to use the service_analyzer tool to check if there are any anomalies in the request volume, response time, and success rate of the frontend service at a given timestamp.

[0084] Action: service_analyzer

[0085] Action input: {"start_timestamp":1722936555000,"service_names":["frontend"]}

[0086] start tool-service_analyzer

[0087] Service exception: ["object:('Service','frontend'),event_type:traffic_decline"]

[0088] end-service_analyzer

[0089] I know the final answer: at timestamp 1722936555000, the number of requests to the frontend service dropped abnormally.

[0090] >Finished chain.

[0091] finish-trade_agent

[0092] {'input':'At timestamp 1722936555000, are there any abnormalities in the request volume, response time and success rate of the frontend service? ','output':'At timestamp 1722936555000, the request volume of the frontend service decreased abnormally. '}

[0093] end-trade_agent

[0094] Big model idea: In order to determine whether the frontend service is abnormal at time 1722936555000, I first need to obtain the business indicators of the frontend service at that time from trade_agent, such as request volume, response time, and success rate, to determine whether there is an abnormality. If there is an abnormality, I will need to further analyze the possible root cause, which may involve checking the underlying pod or host resource status, and checking the call relationship between services to locate the source of the problem.

[0095] Optional Action 1: trade_agent

[0096] Action Input 1: At timestamp 1722936555000, are there any abnormalities in the request volume, response time, and success rate of the frontend service?

[0097] Optional action 2: metric_agent

[0098] Action Input 2: At timestamp 1722936555000, are the pod and host resource indicators related to the frontend service normal?

[0099] Then, based on metric_agent, the question of action input 2 is answered, and the root cause of the system abnormality is analyzed step by step. I will not repeat it here, and get the final idea: I have collected enough information to determine that at timestamp 1722936555000, the frontend service is indeed abnormal, and the root cause that may cause the abnormality is found. According to observations, not only the number of requests for the frontend service has dropped abnormally, but also the CPU load of the host k8s-node1 on which it runs has spikes and the memory utilization is too high. Further investigation found that all services running on k8s-node1 experienced abnormal traffic decline, indicating that the abnormality of host resources may be the direct cause of the service abnormality. My current confidence level is 95 points, because there is a clear correlation between the abnormality of host resources and the service abnormality. Final answer: At timestamp 1722936555000, the number of requests for the frontend service dropped abnormally. The root cause of this system abnormality is the CPU load spike and high memory utilization of the host k8s-node1 on which it runs, which affects the normal operation of all services on the host. The final answer is output by the root cause agent as system anomaly analysis result data.

[0100] It has been verified that the positioning accuracy can reach 70-80% by simulating using the hipster open source system.

[0101] The technical solution of the embodiment of the present invention obtains a system abnormality root cause analysis request and sends the system abnormality root cause analysis request to a large model, thereby analyzing the system abnormality root cause analysis request according to a resampling mechanism through the large model to obtain initial completion data, and then based on the initial completion data, the detection tool module and the large model, determine the current loop call completion data, and further determine the system abnormality analysis result data according to the current loop call completion data and the loop call end condition of the large model. In this solution, the large model analyzes the system abnormality root cause analysis request based on the resampling mechanism, so that the large model can be repeatedly executed multiple times to obtain as rich initial completion data as possible to support multi-directional reasoning, and based on the initial completion data, the large model and the detection tool module, the system abnormality analysis result data can realize automatic analysis of the root cause of the system abnormality, which does not rely on artificially defined causality, solves the problem that the existing root cause location method relies on artificially defined causality and cannot be widely used in the system, can better determine the root cause of the system abnormality, and has a complete human-like cause determination process, and the degree of automation and system universality are higher.

[0102] Embodiment 3

[0103] Figure 4 This is a schematic diagram of the structure of a device for locating the root cause of a system abnormality provided in Embodiment 3 of the present invention. Figure 4 As shown, the device comprises:

[0104] The request acquisition and sending module 310 is used to acquire a system abnormality root cause analysis request and send the system abnormality root cause analysis request to the large model;

[0105] The initial supplement data determination module 320 is used to analyze the system abnormality root cause analysis request through the large model according to the resampling mechanism to obtain the initial supplement data;

[0106] The system abnormality analysis result data determination module 330 is used to determine the system abnormality analysis result data based on the initial completion data, the large model and the detection tool module.

[0107] The technical solution of the embodiment of the present invention obtains a system abnormality root cause analysis request, and sends the system abnormality root cause analysis request to a large model, and then analyzes the system abnormality root cause analysis request according to a resampling mechanism through the large model to obtain initial completion data, and further determines the system abnormality analysis result data based on the initial completion data, the large model, and the detection tool module. In this solution, the large model analyzes the system abnormality root cause analysis request based on the resampling mechanism, so that the large model can be repeatedly executed multiple times to obtain as rich initial completion data as possible to support multi-directional reasoning, and based on the initial completion data, the large model, and the detection tool module, the system abnormality analysis result data can be determined to realize automatic analysis of the root cause of the system abnormality, which does not rely on artificially defined causality, solves the problem that the existing root cause location method relies on artificially defined causality and cannot be widely used in the system, can better determine the root cause of the system abnormality, and has a complete human-like cause determination process, and the degree of automation and system universality are higher.

[0108] Optionally, the system abnormality analysis result data determination module 330 includes a current loop call completion data determination unit and a system abnormality analysis result data determination unit. The current loop call completion data determination unit is used to determine the current loop call completion data based on the initial completion data, the detection tool module and the large model. The system abnormality analysis result data determination unit is used to determine the system abnormality analysis result data according to the current loop call completion data and the loop call end condition of the large model.

[0109] Optionally, the current loop call completion data determination unit is specifically used to parse the initial completion data based on the root cause agent of the multi-agent module to determine the target sub-agent; determine the current loop call data to be completed based on the initial completion data, the detection tool module and the target sub-agent; input the current loop call data to be completed into the large model through the root cause agent, and analyze the current loop call data to be completed through the large model to determine the current loop call completion data.

[0110] Optionally, the current cycle calls the completion data determination unit, specifically for determining the tool instructions to be executed according to the initial completion data and the target subagent; and obtaining the current cycle calls the data to be completed according to the tool instructions to be executed and the detection tool module.

[0111] Optionally, the root cause locating device for system abnormality further includes a history record module for acquiring historical root cause determination data; and inputting the historical root cause determination data into the large model.

[0112] Optionally, the target subagent includes at least one of a transaction agent, a resource indicator agent, a system information agent, a log agent, and a topology relationship agent.

[0113] Optionally, the current loop calls the completion data determination unit, which is specifically used to analyze the current loop calls the data to be completed through the large model to obtain target action data; sample and prune the target action data through the semantic similarity threshold in the large model to obtain the current loop calls the completion data.

[0114] The root cause locating device for system anomalies provided in the embodiments of the present invention can execute the root cause locating method for system anomalies provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0115] Embodiment 4

[0116] Figure 5 A schematic diagram of an electronic device that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0117] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0118] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0119] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a root cause location method for system abnormalities.

[0120] In some embodiments, the root cause location method of the system abnormality can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the root cause location method of the system abnormality described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the root cause location method of the system abnormality in any other appropriate manner (for example, by means of firmware).

[0121] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0122] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0123] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0124] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0125] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0126] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0127] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0128] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for locating the root cause of a system abnormality, characterized in that: include: Obtain a system abnormality root cause analysis request, and send the system abnormality root cause analysis request to the big model; By using the large model, the system abnormality root cause analysis request is analyzed according to the resampling mechanism to obtain initial completion data; Based on the initial completed data, the large model and the detection tool module, the system abnormality analysis result data is determined.

2. The method according to claim 1, characterized in that The determining of system abnormality analysis result data based on the initial completion data, the large model and the detection tool module includes: Determine the current loop call completion data based on the initial completion data, the detection tool module and the large model; The system abnormality analysis result data is determined according to the current loop call completion data and the loop call end condition of the large model.

3. The method according to claim 2, characterized in that The determining of the current loop calling completion data based on the initial completion data, the detection tool module and the large model includes: The root cause agent based on the multi-agent module parses the initial completion data and determines the target sub-agent; Based on the initial completion data, the detection tool module and the target subagent, determining the current loop calls the data to be completed; The current cycle call data to be completed is input into the large model through the root cause agent, and the current cycle call data to be completed is analyzed through the large model to determine the current cycle call completion data.

4. The method according to claim 3, characterized in that The determining of the current loop calling data to be completed based on the initial completion data, the detection tool module and the target subagent includes: Determine the tool instructions to be executed according to the initial completion data and the target subagent; According to the tool instructions to be executed and the detection tool module, the current cycle calls the data to be completed.

5. The method according to claim 1, characterized in that Before the step of analyzing the root cause analysis request of the system abnormality by the large model according to the resampling mechanism to obtain the initial supplementary data, the step further includes: Obtain historical root cause determination data; The historical root cause determination data is input into the macro model.

6. The method according to claim 3, characterized in that The target subagent includes at least one of a transaction agent, a resource indicator agent, a system information agent, a log agent, and a topology relationship agent.

7. The method according to claim 3, characterized in that The step of analyzing the current cycle calling the data to be completed by using the large model to determine the current cycle calling the completed data includes: Analyze the data to be completed in the current cycle call through the large model to obtain target action data; The target action data is sampled and pruned using the semantic similarity threshold in the large model to obtain the current cycle call completion data.

8. A root cause location device for system abnormality, characterized in that: include: The request acquisition and sending module is used to acquire the system abnormality root cause analysis request and send the system abnormality root cause analysis request to the big model; An initial supplement data determination module is used to analyze the system abnormality root cause analysis request according to the resampling mechanism through the large model to obtain initial supplement data; The system abnormality analysis result data determination module is used to determine the system abnormality analysis result data based on the initial completion data, the large model and the detection tool module.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for locating the root cause of a system abnormality according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the root cause locating method for a system abnormality according to any one of claims 1 to 7 when executed.

Citation Information

Cited By

  • Program error tracing method, electronic equipment, storage medium and program product

    CN121412024A

  • An operation and maintenance root cause analysis method based on a large model intelligent agent and a fault tree guide

    CN122985474A