Intelligent problem troubleshooting method, device, computer equipment and storage medium

Abstract

The embodiments of the present application disclose a method, apparatus, computer equipment and storage medium for intelligent problem troubleshooting. The current problem for the target program is obtained; if the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model, the target diagnostic path including at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis and data anomaly diagnosis and an adapter interface corresponding to each diagnostic dimension; for each diagnostic dimension, the corresponding current adapter is loaded according to the corresponding adapter interface, and the corresponding current diagnostic content is determined according to the corresponding current adapter; the diagnostic result corresponding to the diagnostic dimension is determined according to the current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; the diagnostic result is cross-validated to obtain the problem troubleshooting result of the current problem. By implementing the method of the embodiments of the present application, the troubleshooting efficiency of technical problems can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to intelligent problem troubleshooting methods, devices, computer equipment, and storage media. Background Art

[0002] After providing service support to merchants, in order to ensure service quality, it is necessary to respond to customer inquiries in a timely manner and repair problems in a timely manner.

[0003] In the current business environment, technical support and customer service teams are facing an ever-increasing workload. Merchant inquiries cover a wide range of issues in different fields, including business consultation, technical issues, product function issues, system failures, and configuration issues. Especially for technical issues, customer service, technical support, and even R&D need to spend a lot of time on problem troubleshooting, and the problem troubleshooting efficiency is low. Summary of the Invention

[0004] The embodiments of the present application provide a method, apparatus, computer equipment, and storage medium for intelligent problem troubleshooting, which can improve the efficiency of troubleshooting technical problems.

[0005] In a first aspect, an embodiment of the present application provides a method for intelligently troubleshooting problems, which includes:

[0006] Get the current issues for the target program;

[0007] If the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model, where the target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis, and data anomaly diagnosis, and an adapter interface corresponding to each of the diagnostic dimensions;

[0008] For each of the diagnostic dimensions, loading a current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining a current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter;

[0009] For each of the diagnostic dimensions, determining a diagnostic result corresponding to the diagnostic dimension according to a current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content;

[0010] The diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path are cross-validated to obtain a troubleshooting result for the current problem.

[0011] In a second aspect, an embodiment of the present application further provides an intelligent problem troubleshooting device, the intelligent problem troubleshooting device comprising a transceiver unit and a processing unit, wherein:

[0012] The transceiver unit is used to obtain current issues for the target program;

[0013] The processing unit is used to generate a target diagnostic path for the current problem according to the problem type of the current problem and a preset reinforcement learning model when the current problem is an unresolved technical problem, the target diagnostic path including at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis and data anomaly diagnosis and an adapter interface corresponding to each diagnostic dimension; for each diagnostic dimension, load the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determine the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter; for each diagnostic dimension, determine the diagnostic result corresponding to the diagnostic dimension according to the current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; cross-validate the diagnostic results corresponding to each diagnostic dimension in the target diagnostic path to obtain the problem troubleshooting result of the current problem.

[0014] In a third aspect, an embodiment of the present application further provides a computer device, which includes a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the computer program.

[0015] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the above method can be implemented.

[0016] The embodiment of the present application provides a method, apparatus, computer equipment and storage medium for intelligent problem troubleshooting. The method includes: obtaining a current problem for a target program; if the current problem is an unresolved technical problem, generating a target diagnostic path for the current problem based on the problem type of the current problem and a preset reinforcement learning model, wherein the target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis and data anomaly diagnosis and an adapter interface corresponding to each diagnostic dimension; for each diagnostic dimension, loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter; for each diagnostic dimension, determining the diagnostic result corresponding to the diagnostic dimension according to the current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; cross-validating the diagnostic results corresponding to each diagnostic dimension in the target diagnostic path to obtain a problem troubleshooting result for the current problem. On the one hand, the embodiment of the present application dynamically determines the optimal diagnostic path for the current problem based on the reinforcement learning model, which can avoid the solidification of the diagnostic process. On the other hand, the embodiment of the present application can perform multi-dimensional diagnosis of the current problem and automatically determine the problem troubleshooting results based on the multi-dimensional diagnosis results, which not only ensures the accuracy of problem troubleshooting, but also eliminates the need for manual troubleshooting, thereby improving the efficiency of problem troubleshooting. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 A flowchart of the intelligent problem troubleshooting method provided in an embodiment of the present application;

[0019] Figure 2 A schematic block diagram of an intelligent problem troubleshooting device provided in an embodiment of the present application;

[0020] Figure 3 A schematic block diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0023] It should also be understood that the terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] It should be further understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0025] The embodiments of the present application provide a method, apparatus, computer equipment, and storage medium for intelligent problem troubleshooting.

[0026] The executor of the intelligent problem troubleshooting method may be the intelligent problem troubleshooting device provided in the embodiment of the present application, or a computer device (intelligent problem troubleshooting system) integrating the intelligent problem troubleshooting device, wherein the intelligent problem troubleshooting device may be implemented in hardware or software, and the computer device may be a terminal or a server.

[0027] Figure 1 This is a flow chart of the intelligent problem troubleshooting method provided by the embodiment of this application. Figure 1 As shown, the method includes the following steps S110-S150.

[0028] S110: Obtain the current problem for the target program.

[0029] In this embodiment, the current issue can be a consulting question (business consulting question) reported by a customer regarding the target program, or it can be a Jira issue obtained from Jira (a system for tracking tasks, faults, and requirements). The target program can be a program that provides services to customers, such as a mall mini program.

[0030] First, the intelligent intent classifier based on deep learning is used to perform intent recognition on the input current problem to determine whether the current problem is a Jira problem or a business consulting problem. If it is a Jira problem, it will further trigger Jira semantic analysis and then input the problem status prediction engine. If it is a business consulting problem, it will be directly input into the problem status prediction engine. The problem status prediction engine obtains a structured parameter package, which includes the problem description, status, and cause.

[0031] Jira semantic parsing includes the following steps:

[0032] First, JiraKey (the unique issue number of a Jira issue) is extracted. For example, entity recognition technology based on the Large Language Model (LLM) model is used to extract the issue number (such as ONLINE-123). Then, the Jira interface is called to obtain multi-source data. The multi-source data includes issue descriptions, attachments, comments, and group chat summary messages associated with the issue obtained by calling the group message interface. After performing a triple cleansing mechanism on the multi-source data, the basic features corresponding to the current issue are obtained. Multi-modal fusion is performed on the multi-source data to obtain fused data. Finally, the fused data is input into the status prediction engine to generate a structured parameter package.

[0033] The triple cleaning mechanism is a hierarchical cleaning rule, including:

[0034] Regular expression filtering layer: removes special characters and irrelevant unstructured content;

[0035] Dynamic rule base layer: generates context-aware cleaning strategies based on historical data training LLM;

[0036] Semantic verification layer: ensures that the cleaned data retains the key semantics of the original problem.

[0037] After obtaining the structured parameter package, the problem category of the current problem will be determined based on the problem status in the structured parameter package. The problem categories include product problems, technical problems or consulting problems. The technical problems include solved technical problems and unsolved technical problems. The consulting problems include solved consulting problems and unsolved consulting problems.

[0038] Finally, a three-loop routing decision is made based on the problem category, where the three-loop routing decision includes:

[0039] 1. Technical problem routing decision:

[0040] For resolved issues: trigger the knowledge archiving process;

[0041] For unresolved issues: trigger multi-dimensional diagnosis (proceed to step S120);

[0042] 2. Consulting question routing decision:

[0043] For resolved issues: trigger the knowledge archiving process;

[0044] For unresolved issues: trigger knowledge base search;

[0045] 3. Product problem routing decision:

[0046] Solved problem: triggers the knowledge archiving process;

[0047] Generate a Product Requirements Document (PRD) and generate automatic tasks to synchronize to the product. That is, generate a standardized PRD document (including background description, functional description, and priority assessment) through the intelligent requirements generator, and synchronize it to the product management system through the requirements verification gateway.

[0048] S120. If the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model. The target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis, and data anomaly diagnosis, and an adapter interface corresponding to each of the diagnostic dimensions.

[0049] In this embodiment, if the current problem is determined to be an unresolved technical problem, a corresponding target diagnostic path is dynamically generated based on the problem type of the current problem and the reinforcement learning model. The reinforcement learning model can dynamically generate a diagnostic path and intelligently allocate computing resources based on historical experience of the corresponding problem type.

[0050] Among them, the target diagnostic path not only includes diagnostic dimensions, but also includes adapter interfaces corresponding to each diagnostic dimension. The adapter interface is used to dynamically load the corresponding platform adapter according to the execution requirements of the corresponding diagnostic dimension, thereby reducing the size of the intelligent problem troubleshooting system.

[0051] S130. For each diagnostic dimension, load the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determine the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter.

[0052] In this embodiment, after the target diagnostic path is determined, the diagnostic steps of each diagnostic dimension in the target diagnostic path may be executed synchronously or sequentially according to the settings.

[0053] S140 . For each of the diagnostic dimensions, determine a diagnostic result corresponding to the diagnostic dimension according to a current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content.

[0054] In this implementation, different dimensions correspond to different diagnostic rules. After determining the current diagnostic content, the problem diagnosis is performed on the current diagnostic content based on the current diagnostic rules to obtain the corresponding diagnostic results.

[0055] In some embodiments, for code exception diagnosis in a target diagnostic path (i.e., when the currently executed diagnostic dimension is the code exception diagnosis), the current adapter corresponding to the code exception diagnosis is a code platform adapter, and the code platform adapter stores multiple versions of program code corresponding to the target program. The current adapter corresponding to the diagnostic dimension is loaded according to the corresponding adapter interface, and the current diagnostic content corresponding to the diagnostic dimension is determined according to the corresponding current adapter, including:

[0056] Loading the code platform adapter according to the corresponding adapter interface; obtaining a log timestamp of a problem exception log corresponding to the current problem; obtaining a target program code of a version corresponding to the log timestamp from multiple program codes of the code platform adapter; determining the current diagnostic content from the target program code according to the exception stack information in the problem exception log;

[0057] Specifically, the code version (Git commit record) is matched according to the log timestamp, the code commit version closest to the log timestamp is located as the target program code, and the class name, method name, and line number in the exception stack are extracted based on regular expressions, and the current diagnostic content is determined from the target program code based on the extracted information.

[0058] At this time, determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes:

[0059] Performing syntax defect analysis, logic defect analysis, performance defect analysis, and security defect analysis on the current diagnosis content to obtain a diagnosis result corresponding to the code anomaly diagnosis.

[0060] Specifically, the current diagnostic content indicating abnormality in the current code is automatically identified, and multiple aspects of code abnormality diagnosis are performed on the current code.

[0061] Among them, in some embodiments, if the current diagnostic content contains configuration dependencies, at this time, it automatically jumps to the corresponding configuration item, and then performs abnormal diagnosis on the corresponding configuration item to obtain the corresponding configuration abnormality diagnosis result. When performing cross-validation later, the configuration abnormality diagnosis result and the diagnostic results corresponding to each diagnostic dimension can be combined to determine the problem troubleshooting result of the current problem.

[0062] In some embodiments, when the diagnostic dimension is the configuration abnormality diagnosis, the current adapter corresponding to the configuration abnormality diagnosis is a configuration adapter, and loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, includes:

[0063] Loading the configuration adapter according to the corresponding adapter interface; obtaining the abnormal time window corresponding to the current problem; obtaining configuration change information within the abnormal time window in the configuration adapter, and determining the configuration change information as the current diagnosis content, the configuration change information including at least one configuration item;

[0064] The configuration adapter stores configuration information of the target program.

[0065] At this time, determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes:

[0066] A service dependency graph is constructed based on the correspondence between preset configuration items and services; a sandbox environment is constructed based on the service dependency graph, and the configuration change information is injected into the sandbox environment; multiple verification use cases associated with the configuration items are obtained, and multiple verification use cases are executed in the sandbox environment to obtain diagnostic results corresponding to the configuration anomaly diagnosis.

[0067] Specifically, verification use cases can be obtained from the corresponding logs. When building a sandbox environment, the production environment image is quickly cloned based on Docker (container), and then the configuration items to be verified are injected; then an automated smoke test is performed: verification use cases (such as creating order use cases and payment callback use cases) are executed in the sandbox environment, and the corresponding diagnostic results are output. If the configuration is abnormal, the rollback strategy can be used to repair the configuration in the sandbox environment. If the test pass rate is ≥90%, the rollback is determined to be safe. At this time, the diagnosis result is rollback safety. Otherwise, it prompts that the configuration is wrong and needs to be modified, that is, the diagnosis result corresponding to the configuration abnormality diagnosis is obtained.

[0068] In some embodiments, when the diagnostic dimension is the data anomaly diagnosis, the current adapter corresponding to the data anomaly diagnosis is a database adapter, and loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, includes:

[0069] Load the database adapter according to the corresponding adapter interface; obtain abnormal data corresponding to the current problem, and use the abnormal data as the current diagnosis content;

[0070] The database adapter stores propagation links (propagation trees) corresponding to various data.

[0071] The determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes:

[0072] Determine the data propagation tree corresponding to the abnormal data according to the database adapter; determine at least one affected node associated with the abnormal data according to the preset gradient descent path tracing algorithm and the data propagation tree; determine the node weight of each affected node according to the preset breadth-first contamination tracing algorithm; perform abnormal diagnosis on multiple affected nodes according to the node weights of each affected node, and obtain a diagnosis result corresponding to the data abnormality diagnosis.

[0073] Specifically, anomaly propagation analysis is performed through the following steps: first, gradient descent path tracing is performed. A loss function is defined based on the data propagation tree to quantify the impact of numerical deviations between fields. Backpropagation is then used to locate key contamination nodes. For example, if the anomalous data is an order amount anomaly, analysis will determine that the affected node is the coupon calculation service. Then, using the breadth-first contamination tracing algorithm, a null value propagation tree diagram is constructed, and the contamination diffusion path is marked hierarchically. Nodes with high frequency of modification are prioritized (i.e., nodes with higher frequency of modification receive higher weights). For example, if a user address is abnormal, the geocoding service node is traced back to the node with high frequency of modification (subsequently diagnosed as missing data at the geocoding service node).

[0074] S150 , cross-validating the diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path to obtain a troubleshooting result for the current problem.

[0075] In this embodiment, after cross-validation of the diagnostic results, a troubleshooting result is generated, recording the causal relationship between the multi-dimensional diagnostic results. For example, the troubleshooting result is that the configuration item A times out, causing the code module B to report an error.

[0076] After cross-validating the diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path to obtain the troubleshooting result of the current problem, the method further includes:

[0077] If the problem troubleshooting result indicates that there is an abnormal problem, a technical repair plan is generated based on the problem troubleshooting result, and the target program is repaired according to the technical repair plan; specifically, a comprehensive repair plan is generated, and the comprehensive repair plan includes at least one repair plan among code patches, configuration changes and data repair scripts.

[0078] If the problem troubleshooting result indicates that there is no abnormal problem, the target diagnostic path is updated, and the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface for each diagnostic dimension, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter are returned to execute, until the problem troubleshooting result indicates that there is an abnormal problem or the number of updates of the target diagnostic path is greater than the preset update threshold.

[0079] Specifically, if a problem is found, a corresponding technical repair plan is generated, and the target program is repaired according to the technical repair plan. For example, the problem troubleshooting result is that configuration item A times out, causing code module B to report an error. At this time, the generated technical repair plan includes both code patch instructions and configuration change instructions, and automatically repairs the target program based on the code patch instructions and configuration change instructions.

[0080] For data repair strategies in the area of ​​data anomaly diagnosis, differential snapshot comparisons can be performed. Data fingerprints (MD5 (Message Digest Algorithm Version 5) hash aggregate values) are generated at hourly granularity. When an anomaly occurs, the three most recent healthy snapshots are loaded for line-by-line comparison. Based on the differential analysis results, data compensation actions are automatically generated, creating intelligent repair scripts.

[0081] Specifically, if no problem is found, the result of the failure of problem diagnosis according to the current target diagnostic path will be fed back to the reinforcement learning model (similarly, when the problem diagnosis is successful according to the current target diagnostic path, the response result will also be fed back to the reinforcement learning model), so that the reinforcement learning model is updated in real time. In addition, when the diagnosis fails through the target diagnostic path, the target diagnostic path will be updated, such as the secondary generation of the diagnostic path according to the reinforcement learning model. For example, the target diagnostic path generated for the first time is code anomaly diagnosis and configuration anomaly diagnosis, and the target diagnostic path generated for the second time is adjusted to data anomaly diagnosis.

[0082] Furthermore, if the diagnosis fails through the target diagnostic path, the path can be regenerated without resetting it. Instead, the parameter range corresponding to the diagnostic path (such as the query range of the log) or the retrieval dimension (such as the field dimension of the log, code or configuration item) can be adjusted first. If the diagnosis fails for a preset number of consecutive times (such as 3 times) after adjusting the parameters, the path is regenerated, and the problem diagnosis is performed based on the regenerated path.

[0083] Furthermore, in order to discover problems in a timely manner, the present embodiment also provides real-time diagnosis of anomalies based on logs. At this time, the method also includes: real-time querying of the work diary of the target program; if there is a work anomaly log in the work diary, determining the anomaly code from the code platform adapter according to the work anomaly log; performing code anomaly diagnosis on the anomaly code to obtain a code anomaly diagnosis result; if it is detected that there is a configuration dependency in the anomaly code, obtaining the dependency configuration item corresponding to the configuration dependency; performing configuration anomaly diagnosis on the dependency configuration item to obtain a configuration anomaly diagnosis result; generating a real-time problem troubleshooting result based on the code anomaly diagnosis result and the configuration anomaly diagnosis result.

[0084] After generating the technical repair plan, the disposal results need to be output. This embodiment supports dual-path output of the disposal results. First, it is output to the corresponding user end (for example, returning the technical repair plan to the user terminal via the Internet), and second, it is output to the knowledge end of the system, triggering the technical closed-loop mechanism to automatically precipitate the solution.

[0085] Furthermore, this embodiment also provides an offline supplementary knowledge base function: using intelligent event-driven and timed tasks, it automatically triggers compensation tasks based on Jira issue status change events, and records analysis conclusions and synchronizes Jira notes.

[0086] Specifically, the knowledge acquisition pipeline includes offline knowledge acquisition and online real-time synchronization. For offline knowledge acquisition, historical Jira data is batch-pulled through scheduled tasks (for example, at midnight daily), filtered by the intelligent quality inspection gateway, and stored in the knowledge graph. Closed issues are annotated with business categories (such as "Customer Management" and "Member Level") using the LLM model. For online real-time synchronization, when the Jira issue status changes to "Resolved," the solution synchronization process is automatically triggered. After filtering through the intelligent quality inspection gateway, data is stored in the knowledge graph to ensure knowledge validity.

[0087] This embodiment also provides a conflict detection algorithm and a version resolution strategy in closed-loop knowledge management. Specifically, for the conflict detection algorithm, semantic similarity is calculated based on the knowledge graph, the new knowledge entries are compared with the historical versions, and the conflict resolution methods are automatically marked; for the version resolution strategy, a timestamp priority strategy and an expert voting mechanism are included. For the timestamp priority strategy: the knowledge entries extracted offline and obtained synchronously online are uniformly timestamped. When a conflict is detected, the latest timestamp entry is used by default to overwrite the old version. For the expert voting mechanism: the manual review process is forcibly triggered for high-conflict risk entries (such as configuration change plans involving system security).

[0088] To sum up, on the one hand, the embodiment of the present application dynamically determines the optimal diagnostic path for the current problem based on the reinforcement learning model, which can avoid the solidification of the diagnostic process. On the other hand, the embodiment of the present application can perform multi-dimensional diagnosis of the current problem and automatically determine the problem troubleshooting results based on the multi-dimensional diagnosis results, which not only ensures the accuracy of problem troubleshooting, but also eliminates the need for manual troubleshooting, thereby improving the efficiency of problem troubleshooting.

[0089] Figure 2 This is a schematic block diagram of an intelligent problem troubleshooting device provided by an embodiment of the present application. Figure 2 As shown, corresponding to the above problem intelligent troubleshooting method, this application also provides a problem intelligent troubleshooting device 200. The problem intelligent troubleshooting device 200 includes a unit for executing the above problem intelligent troubleshooting method, and the problem intelligent troubleshooting device 200 can be configured in a terminal or a server. Figure 2 The intelligent problem troubleshooting device 200 includes a transceiver unit 201 and a processing unit 202, wherein:

[0090] The transceiver unit 201 is used to obtain the current problem of the target program;

[0091] The processing unit 202 is used to generate a target diagnostic path for the current problem according to the problem type of the current problem and a preset reinforcement learning model when the current problem is an unresolved technical problem, the target diagnostic path including at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis and data anomaly diagnosis and an adapter interface corresponding to each diagnostic dimension; for each diagnostic dimension, load the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determine the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter; for each diagnostic dimension, determine the diagnostic result corresponding to the diagnostic dimension according to the current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; cross-validate the diagnostic results corresponding to each diagnostic dimension in the target diagnostic path to obtain the problem troubleshooting result of the current problem.

[0092] In some embodiments, when the diagnostic dimension is the code exception diagnosis, the current adapter corresponding to the code exception diagnosis is a code platform adapter, and the code platform adapter stores multiple versions of program codes corresponding to the target program. When the processing unit 202 executes the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, it is specifically used to:

[0093] Loading the code platform adapter according to the corresponding adapter interface;

[0094] Obtain the log timestamp of the problem exception log corresponding to the current problem;

[0095] Acquire a target program code of a version corresponding to the log timestamp from a plurality of program codes of the code platform adapter;

[0096] determining the current diagnosis content from the target program code according to the exception stack information in the problem exception log;

[0097] At this time, when the processing unit 202 executes the step of determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content, it is specifically configured to:

[0098] Performing syntax defect analysis, logic defect analysis, performance defect analysis, and security defect analysis on the current diagnosis content to obtain a diagnosis result corresponding to the code anomaly diagnosis.

[0099] In some embodiments, when the diagnostic dimension is the configuration abnormality diagnosis, the current adapter corresponding to the configuration abnormality diagnosis is the configuration adapter. When the processing unit 202 executes the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, it is specifically configured to:

[0100] Loading the configuration adapter according to the corresponding adapter interface;

[0101] Obtain the abnormal time window corresponding to the current problem;

[0102] Acquire configuration change information within the abnormal time window in the configuration adapter, and determine the configuration change information as the current diagnosis content, wherein the configuration change information includes at least one configuration item;

[0103] At this time, when the processing unit 202 executes the step of determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content, it is specifically configured to:

[0104] Build a service dependency graph based on the correspondence between preset configuration items and services;

[0105] Building a sandbox environment according to the service dependency graph, and injecting the configuration change information into the sandbox environment;

[0106] Acquire multiple verification use cases associated with the configuration item, and execute the multiple verification use cases in the sandbox environment to obtain a diagnosis result corresponding to the configuration anomaly diagnosis.

[0107] In some embodiments, when the diagnostic dimension is the data anomaly diagnosis, the current adapter corresponding to the data anomaly diagnosis is a database adapter. When the processing unit 202 executes the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, it is specifically configured to:

[0108] Loading the database adapter according to the corresponding adapter interface;

[0109] Acquire abnormal data corresponding to the current problem, and use the abnormal data as the current diagnosis content;

[0110] At this time, when the processing unit 202 executes the step of determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content, it is specifically configured to:

[0111] Determining a data propagation tree corresponding to the abnormal data according to the database adapter;

[0112] Determining at least one affected node associated with the abnormal data according to a preset gradient descent path tracing algorithm and the data propagation tree;

[0113] Determine the node weight of each affected node according to a preset breadth-first pollution source tracing algorithm;

[0114] Anomaly diagnosis is performed on the multiple affected nodes according to the node weights of the affected nodes respectively, to obtain a diagnosis result corresponding to the data anomaly diagnosis.

[0115] In some embodiments, after performing the step of cross-validating the diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path to obtain the troubleshooting result of the current problem, the processing unit 202 is specifically configured to:

[0116] If the problem troubleshooting result indicates that an abnormal problem exists, generating a technical repair plan based on the problem troubleshooting result, and repairing the target program according to the technical repair plan;

[0117] If the problem troubleshooting result indicates that there is no abnormal problem, the target diagnostic path is updated, and the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface for each diagnostic dimension, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter are returned to execute, until the problem troubleshooting result indicates that there is an abnormal problem or the number of updates of the target diagnostic path is greater than the preset update threshold.

[0118] In some embodiments, the processing unit 202 is further configured to:

[0119] Query the work diary of the target program in real time;

[0120] If there is a work abnormality log in the work diary, determining the abnormality code from the code platform adapter according to the work abnormality log;

[0121] Performing code abnormality diagnosis on the abnormal code to obtain a code abnormality diagnosis result;

[0122] If configuration dependency is detected in the abnormal code, the corresponding dependency configuration item is obtained;

[0123] Performing configuration anomaly diagnosis on the dependent configuration item to obtain a configuration anomaly diagnosis result;

[0124] Generate real-time problem troubleshooting results based on the code anomaly diagnosis results and the configuration anomaly diagnosis results.

[0125] In some embodiments, after obtaining the current problem for the target program, and before generating a target diagnostic path for the current problem based on the problem type of the current problem and a preset reinforcement learning model if the current problem is an unresolved technical problem, the processing unit 202 is further configured to:

[0126] Determine the problem category of the current problem, where the problem category includes product problems, technical problems, or consulting problems. The technical problems include resolved technical problems and unresolved technical problems, and the consulting problems include resolved consulting problems and unresolved consulting problems.

[0127] To sum up, on the one hand, the embodiment of the present application dynamically determines the optimal diagnostic path for the current problem based on the reinforcement learning model, which can avoid the solidification of the diagnostic process. On the other hand, the embodiment of the present application can perform multi-dimensional diagnosis of the current problem and automatically determine the problem troubleshooting results based on the multi-dimensional diagnosis results, which not only ensures the accuracy of problem troubleshooting, but also eliminates the need for manual troubleshooting, thereby improving the efficiency of problem troubleshooting.

[0128] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned intelligent problem troubleshooting device and each unit can refer to the corresponding description in the aforementioned method embodiment. For the convenience and brevity of the description, it will not be repeated here.

[0129] The above-mentioned intelligent troubleshooting device can be implemented in the form of a computer program. The computer program can be used in Figure 3 Runs on the computer equipment shown.

[0130] See also Figure 3 , Figure 3 This is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 300 can be a terminal or a server. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, personal digital assistant, wearable device, or other electronic device with communication capabilities. The server can be a standalone server or a server cluster consisting of multiple servers.

[0131] See Figure 3 The computer device 300 includes a processor 302 , a memory, and a network interface 305 connected via a system bus 301 , wherein the memory may include a non-volatile storage medium 303 and an internal memory 304 .

[0132] The non-volatile storage medium 303 can store an operating system 3031 and a computer program 3032. The computer program 3032 includes program instructions, which, when executed, can enable the processor 302 to execute a method for intelligent problem troubleshooting.

[0133] The processor 302 is used to provide computing and control capabilities to support the operation of the entire computer device 300.

[0134] The internal memory 304 provides an environment for the operation of the computer program 3032 in the non-volatile storage medium 303. When the computer program 3032 is executed by the processor 302, the processor 302 can execute an intelligent problem troubleshooting method.

[0135] The network interface 305 is used to communicate with other devices through the network. Figure 3The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 300 to which the solution of the present application is applied. The specific computer device 300 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0136] The processor 302 is configured to execute a computer program 3032 stored in the memory to implement the following steps:

[0137] Get the current issues for the target program;

[0138] If the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model, where the target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis, and data anomaly diagnosis, and an adapter interface corresponding to each of the diagnostic dimensions;

[0139] For each of the diagnostic dimensions, loading a current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining a current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter;

[0140] For each of the diagnostic dimensions, determining a diagnostic result corresponding to the diagnostic dimension according to a current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content;

[0141] The diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path are cross-validated to obtain a troubleshooting result for the current problem.

[0142] It should be understood that in the embodiment of the present application, the processor 302 may be a central processing unit (CPU), and the processor 302 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0143] Those skilled in the art will appreciate that all or part of the steps in the method of the above-described embodiment can be implemented by instructing the relevant hardware through a computer program. The computer program includes program instructions, which can be stored in a storage medium that is computer-readable. The program instructions are executed by at least one processor in the computer system to implement the steps in the method of the above-described embodiment.

[0144] Therefore, the present application also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor performs the following steps:

[0145] Get the current issues for the target program;

[0146] If the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model, where the target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis, and data anomaly diagnosis, and an adapter interface corresponding to each of the diagnostic dimensions;

[0147] For each of the diagnostic dimensions, loading a current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining a current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter;

[0148] For each of the diagnostic dimensions, determining a diagnostic result corresponding to the diagnostic dimension according to a current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content;

[0149] The diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path are cross-validated to obtain a troubleshooting result for the current problem.

[0150] The storage medium may be any computer-readable storage medium that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk.

[0151] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and other division methods may be used in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not implemented.

[0153] The steps in the method of the embodiment of the present application can be adjusted in order, combined, and deleted according to actual needs. The units in the device of the embodiment of the present application can be combined, divided, and deleted according to actual needs. In addition, the functional units in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into a single unit.

[0154] If this integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, terminal, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of this application.

[0155] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present application, and such modifications or substitutions should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for intelligent problem troubleshooting, characterized in that: include: Obtaining a current question for a target program, where the target program is a mall mini program that provides services to customers; If the current problem is an unresolved technical problem, a target diagnostic path for the current problem is generated based on the problem type of the current problem and a preset reinforcement learning model, where the target diagnostic path includes at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis, and data anomaly diagnosis, and an adapter interface corresponding to each of the diagnostic dimensions; For each of the diagnostic dimensions, loading a current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining a current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter; For each of the diagnostic dimensions, determining a diagnostic result corresponding to the diagnostic dimension according to a current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; Cross-validate the diagnostic results corresponding to each of the diagnostic dimensions in the target diagnostic path to obtain a troubleshooting result for the current problem; If the problem troubleshooting result indicates that an abnormal problem exists, generating a technical repair plan based on the problem troubleshooting result, and repairing the target program according to the technical repair plan; If the problem troubleshooting result indicates that there is no abnormal problem, the current target diagnostic path is fed back to the reinforcement learning model as a result of problem diagnosis failure to update the reinforcement learning model, and the target diagnostic path is updated by the updated reinforcement learning model, and the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface for each diagnostic dimension, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter are returned to execute, until the problem troubleshooting result indicates that there is an abnormal problem or the number of updates of the target diagnostic path is greater than the preset update threshold.

2. The method according to claim 1, characterized in that When the diagnostic dimension is the code exception diagnosis, the current adapter corresponding to the code exception diagnosis is a code platform adapter, and the code platform adapter stores multiple versions of program codes corresponding to the target program. The current adapter corresponding to the diagnostic dimension is loaded according to the corresponding adapter interface, and the current diagnostic content corresponding to the diagnostic dimension is determined according to the corresponding current adapter, including: Loading the code platform adapter according to the corresponding adapter interface; Obtain the log timestamp of the problem exception log corresponding to the current problem; Acquire a target program code of a version corresponding to the log timestamp from a plurality of program codes of the code platform adapter; determining the current diagnosis content from the target program code according to the exception stack information in the problem exception log; The determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes: Performing syntax defect analysis, logic defect analysis, performance defect analysis, and security defect analysis on the current diagnosis content to obtain a diagnosis result corresponding to the code anomaly diagnosis.

3. The method according to claim 1, characterized in that When the diagnostic dimension is the configuration abnormality diagnosis, the current adapter corresponding to the configuration abnormality diagnosis is the configuration adapter, and loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, includes: Loading the configuration adapter according to the corresponding adapter interface; Obtain the abnormal time window corresponding to the current problem; Acquire configuration change information within the abnormal time window in the configuration adapter, and determine the configuration change information as the current diagnosis content, wherein the configuration change information includes at least one configuration item; The determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes: Build a service dependency graph based on the correspondence between preset configuration items and services; Building a sandbox environment according to the service dependency graph, and injecting the configuration change information into the sandbox environment; Acquire multiple verification use cases associated with the configuration item, and execute the multiple verification use cases in the sandbox environment to obtain a diagnosis result corresponding to the configuration anomaly diagnosis.

4. The method according to claim 1, wherein When the diagnostic dimension is the data anomaly diagnosis, the current adapter corresponding to the data anomaly diagnosis is a database adapter, and loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, includes: Loading the database adapter according to the corresponding adapter interface; Acquire abnormal data corresponding to the current problem, and use the abnormal data as the current diagnosis content; The determining the diagnosis result corresponding to the diagnosis dimension according to the current diagnosis rule corresponding to the diagnosis dimension and the current diagnosis content includes: Determining a data propagation tree corresponding to the abnormal data according to the database adapter; Determining at least one affected node associated with the abnormal data according to a preset gradient descent path tracing algorithm and the data propagation tree; Determine the node weight of each affected node according to a preset breadth-first pollution source tracing algorithm; Anomaly diagnosis is performed on the multiple affected nodes according to the node weights of the affected nodes respectively, to obtain a diagnosis result corresponding to the data anomaly diagnosis.

5. The method according to claim 1, wherein The method further comprises: Query the work diary of the target program in real time; If there is a work abnormality log in the work diary, determining the abnormality code from the code platform adapter according to the work abnormality log; Performing code abnormality diagnosis on the abnormal code to obtain a code abnormality diagnosis result; If configuration dependency is detected in the abnormal code, the corresponding dependency configuration item is obtained; Performing configuration anomaly diagnosis on the dependent configuration item to obtain a configuration anomaly diagnosis result; Generate real-time problem troubleshooting results based on the code anomaly diagnosis results and the configuration anomaly diagnosis results.

6. The method according to claim 1, characterized in that After obtaining the current problem for the target program, if the current problem is an unresolved technical problem, before generating a target diagnostic path for the current problem based on the problem type of the current problem and a preset reinforcement learning model, the method further includes: Determine the problem category of the current problem, where the problem category includes product problems, technical problems, or consulting problems. The technical problems include resolved technical problems and unresolved technical problems, and the consulting problems include resolved consulting problems and unresolved consulting problems.

7. An intelligent problem troubleshooting device, characterized in that: The intelligent problem troubleshooting device includes a transceiver unit and a processing unit, wherein: The transceiver unit is used to obtain the current problem for the target program, where the target program is a mall mini program that provides services to customers; The processing unit is used to generate a target diagnostic path for the current problem according to the problem type of the current problem and a preset reinforcement learning model when the current problem is an unresolved technical problem, the target diagnostic path including at least one diagnostic dimension of code anomaly diagnosis, configuration anomaly diagnosis and data anomaly diagnosis and an adapter interface corresponding to each diagnostic dimension; for each diagnostic dimension, load the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface, and determine the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter; for each diagnostic dimension, determine the diagnostic result corresponding to the diagnostic dimension according to the current diagnostic rule corresponding to the diagnostic dimension and the current diagnostic content; cross-validate the diagnostic results corresponding to each diagnostic dimension in the target diagnostic path , obtain the problem troubleshooting result of the current problem; if the problem troubleshooting result indicates that there is an abnormal problem, generate a technical repair plan according to the problem troubleshooting result, and repair the target program according to the technical repair plan; if the problem troubleshooting result indicates that there is no abnormal problem, feed back the current target diagnostic path as a result of problem diagnosis failure to the reinforcement learning model to update the reinforcement learning model, update the target diagnostic path through the updated reinforcement learning model, and return to execute the steps of loading the current adapter corresponding to the diagnostic dimension according to the corresponding adapter interface for each diagnostic dimension, and determining the current diagnostic content corresponding to the diagnostic dimension according to the corresponding current adapter, until the problem troubleshooting result indicates that there is an abnormal problem or the number of updates of the target diagnostic path is greater than the preset update threshold.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the problem intelligent troubleshooting method according to any one of claims 1 to 6 is implemented.

9. A storage medium, characterized in that: The storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor, the processor executes the intelligent problem troubleshooting method according to any one of claims 1 to 6.

Images