Fault diagnosis method and device, computer equipment and storage medium
By automating the processing of log text, time-series monitoring, and network topology data from cloud computing clusters, and utilizing vector representation and multimodal features, the problem of reliance on human experience in fault diagnosis of traditional cloud computing platforms is solved, achieving efficient and accurate fault diagnosis.
Patent Information
- Application Number
- CN202511810801.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-02-17
AI Technical Summary
Traditional cloud computing platform fault diagnosis methods rely on human experience, which is not accurate enough and increases the complexity of operation and maintenance.
By acquiring log text data, time-series monitoring data, and network topology data from cloud computing clusters, and utilizing vector representation and multimodal feature representation, combined with classifiers and perceptron networks, the fault diagnosis results are automatically determined.
It requires no human intervention, significantly improving the accuracy and efficiency of fault diagnosis and saving manpower.
Smart Images
Figure CN121547348A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of cloud computing, and in particular to a fault diagnosis method and device, a computer device and a storage medium. BACKGROUND
[0002] With the rapid development of cloud computing technology, enterprises can enjoy more flexible and scalable IT infrastructure. However, with the rapid development of artificial intelligence technology, the types of AI applications are becoming more and more diverse, and the scale of cloud computing clusters is also rapidly expanding.
[0003] While artificial skill technology brings convenience to users, this convenience is accompanied by significantly increased operation and maintenance complexity. Traditional cloud computing platform fault diagnosis methods mostly rely on manual fault diagnosis based on collected abnormal data. However, this fault diagnosis method not only depends heavily on human experience, but also has accuracy problems with existing diagnosis methods. SUMMARY
[0004] Therefore, it is necessary to provide a fault diagnosis method, device, computer device and storage medium that can improve fault diagnosis accuracy and effectiveness to solve the above technical problems.
[0005] In a first aspect, the present application provides a fault diagnosis method. The method comprises:
[0006] obtaining log text data, time series monitoring data and network topology data of a cloud computing cluster;
[0007] determining a first vector representation corresponding to the log text data, a second vector representation corresponding to the time series monitoring data, and a third vector representation corresponding to the network topology data;
[0008] determining a fault diagnosis result of the cloud computing cluster according to the first vector representation, the second vector representation and the third vector representation.
[0009] In one embodiment, determining the first vector representation corresponding to the log text data comprises:
[0010] performing semantic analysis on the log text data to obtain a semantic analysis result and an initial semantic vector of the log text data;
[0011] determining a vector weight of the initial semantic vector according to the semantic analysis result;
[0012] performing weighted processing on the initial semantic vector according to the vector weight to obtain the first vector representation corresponding to the log text data.
[0013] In one embodiment, determining the vector weight of the initial semantic vector according to the semantic analysis result comprises:
[0014] In a case where the semantic analysis result contains the target keyword, the vector weight of the initial semantic vector is determined as a first numerical value;
[0015] In a case where the semantic analysis result does not contain the target keyword, the vector weight of the initial semantic vector is determined as a second numerical value; the first numerical value is greater than the second numerical value.
[0016] In one of the embodiments, the second vector representation corresponding to the timing monitoring data is determined, including:
[0017] The timing monitoring data is subjected to sliding window standardization processing to obtain a monitoring data sequence;
[0018] The monitoring data sequence is subjected to encoding processing to obtain the second vector representation corresponding to the timing monitoring data.
[0019] In one of the embodiments, the network topology data includes node running data of cloud computing nodes included in a cloud computing cluster and communication data of communication paths, and a network topology structure of the cloud computing cluster; the communication paths are communication paths between the cloud computing nodes;
[0020] The third vector representation corresponding to the network topology data is determined, including:
[0021] The graph vector of the cloud computing cluster is obtained according to the node running data of each cloud computing node, the communication data of each communication path, and the network topology structure;
[0022] The graph vector of the cloud computing cluster is taken as the third vector representation corresponding to the network topology data.
[0023] In one of the embodiments, the fault diagnosis result of the cloud computing cluster is determined according to the first vector representation, the second vector representation, and the third vector representation, including:
[0024] The multi-modal feature representation is determined according to the first vector representation, the second vector representation, and the third vector representation;
[0025] The fault diagnosis result of the cloud computing cluster is determined according to the first vector representation, the second vector representation, the third vector representation, and the multi-modal feature representation.
[0026] In one of the embodiments, the fault diagnosis result of the cloud computing cluster is determined according to the first vector representation, the second vector representation, the third vector representation, and the multi-modal feature representation, including:
[0027] The first diagnosis result and the first confidence corresponding to the first diagnosis result are determined based on the first vector representation and the first classifier;
[0028] The second diagnosis result and the second confidence corresponding to the second diagnosis result are determined based on the second vector representation and the second classifier.
[0029] Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result;
[0030] Based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the corresponding fourth confidence level;
[0031] Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
[0032] Secondly, this application also provides a fault diagnosis device. The device includes:
[0033] The acquisition module is used to acquire log text data, time-series monitoring data, and network topology data of the cloud computing cluster.
[0034] The first determining module is used to determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0035] The second determining module is used to determine the fault diagnosis results of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation.
[0036] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:
[0037] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0038] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0039] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0040] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:
[0041] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0042] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0043] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0044] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:
[0045] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0046] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0047] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0048] The aforementioned fault diagnosis method, apparatus, computer equipment, and storage medium acquire log text data, time-series monitoring data, and network topology data of a cloud computing cluster. A first vector representation corresponding to the log text data, a second vector representation corresponding to the time-series monitoring data, and a third vector representation corresponding to the network topology data are determined. Based on the first, second, and third vector representations, the fault diagnosis result of the cloud computing cluster is determined. The fault diagnosis method of this application requires no manual intervention. The server automatically determines the fault diagnosis result of the cloud computing cluster by acquiring the log text data, time-series monitoring data, and network topology data of the cloud computing cluster, and determining the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data. This not only saves manpower but also significantly improves the fault diagnosis results. Attached Figure Description
[0049] Figure 1 This is a diagram illustrating the application environment of the fault diagnosis method provided in this embodiment.
[0050] Figure 2 This is a flowchart illustrating the first fault diagnosis method provided in this embodiment;
[0051] Figure 3 This is a flowchart illustrating the process of determining a target fault in a cloud computing system, as provided in this embodiment.
[0052] Figure 4 This is a flowchart illustrating the process of diagnosing target faults in a cloud computing system, as provided in this embodiment.
[0053] Figure 5 This is a flowchart illustrating the second fault diagnosis method provided in this embodiment;
[0054] Figure 6 This is a structural block diagram of a fault diagnosis device provided in this embodiment;
[0055] Figure 7 This is an internal structural diagram of the computer device provided in this embodiment. Detailed Implementation
[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0057] With the rapid development of cloud computing technology, enterprises can enjoy more flexible and scalable IT infrastructure. However, with the rapid development of artificial intelligence technology, the types of AI applications are becoming increasingly diverse, and the scale of cloud computing clusters is also expanding rapidly.
[0058] While human skills and technologies bring convenience to users, this convenience also comes with a significant increase in operational complexity. Traditional cloud computing platform fault diagnosis methods mostly rely on manual fault diagnosis based on collected abnormal data. However, this fault diagnosis method not only heavily depends on human experience, but existing diagnostic methods also have accuracy issues.
[0059] To address the aforementioned technical problems, the fault diagnosis method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, server 104 acquires log text data, time-series monitoring data, and network topology data from cloud computing cluster 102. Server 104 determines the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data. Based on the first vector representation, the second vector representation, and the third vector representation, server 104 determines the fault diagnosis result of cloud computing cluster 102.
[0060] The server can be a standalone server or a server cluster. It should be noted that the server can be a standalone server or a server in a cloud computing cluster.
[0061] In one embodiment, such as Figure 2 As shown, a fault diagnosis method is provided, which can be applied to... Figure 1 Taking the server in the example, the following steps are included:
[0062] S201 retrieves log text data, time-series monitoring data, and network topology data from the cloud computing cluster.
[0063] Here, "cloud computing cluster" refers to the computing power cluster corresponding to cloud computing. A cloud computing cluster contains several cloud computing nodes, each of which can be an independent server or a logical computing node (e.g., a board). "Log text data" refers to the text data related to the log files generated by the cloud computing cluster. "Time-series monitoring data" refers to the monitoring data of the cloud computing cluster containing time information, including CPU utilization, GPU utilization, memory usage, network bandwidth, disk throughput, and latency information. "Network topology data" includes the node operation data and communication path data of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; where the communication path refers to the communication path between cloud computing nodes.
[0064] Optionally, in this embodiment, the log text data can be Kubernetes event logs and container logs.
[0065] Optionally, in this embodiment, time-series monitoring data of each cloud computing node in the cloud computing cluster can be obtained through monitoring tools (e.g., sensors or monitoring software).
[0066] Optionally, in this embodiment, the node operation data and communication path data of each cloud computing node in the cloud computing cluster, as well as the network topology, are obtained to obtain network topology data.
[0067] S202, determine the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data.
[0068] Among them, the first vector representation, the second vector representation, and the third vector representation are the vector features corresponding to log text data, time-series monitoring data, and network topology data, respectively.
[0069] Optionally, in this embodiment, a feature extraction network can be used to extract features from log text data, time-series monitoring data, and network topology data to obtain a first vector representation corresponding to the log text data, a second vector representation corresponding to the time-series monitoring data, and a third vector representation corresponding to the network topology data.
[0070] Optionally, in this embodiment, the time-series monitoring data is subjected to sliding window normalization to obtain a monitoring data sequence. The monitoring data sequence is then encoded to obtain a second vector representation corresponding to the time-series monitoring data. Specifically, the time-series monitoring data is normalized using a sliding window to obtain the monitoring data sequence. The monitoring data sequence is encoded using TCN (Temporal Convolutional Network) to obtain the second vector representation. .
[0071] Optionally, in this embodiment, a graph vector of the cloud computing cluster is obtained based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology. This graph vector of the cloud computing cluster is then used as the third vector representation corresponding to the network topology data. Specifically, the third vector representation can be characterized by the following formula: ;in, This is a high-order semantic representation obtained by aggregating communication data based on the node operation information and corresponding communication paths of each cloud computing node. The cloud computing node can be a virtual machine, microservice, server, or board, etc.
[0072] S203, Determine the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation.
[0073] An optional implementation of this application involves obtaining a first diagnostic result and its confidence level based on a first vector representation and a classifier. A second diagnostic result and its confidence level are then obtained based on a second vector representation and a classifier. A third diagnostic result and its confidence level are then obtained based on a third vector representation and a classifier. The diagnostic result with the highest confidence level is taken as the final fault diagnosis result for the cloud computing cluster.
[0074] In this embodiment, log text data, time-series monitoring data, and network topology data of the cloud computing cluster are acquired. A first vector representation corresponding to the log text data, a second vector representation corresponding to the time-series monitoring data, and a third vector representation corresponding to the network topology data are determined. Based on the first, second, and third vector representations, the fault diagnosis result of the cloud computing cluster is determined. The fault diagnosis method of this application requires no manual intervention. The server acquires the log text data, time-series monitoring data, and network topology data of the cloud computing cluster, and determines the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data, thereby automatically determining the fault diagnosis result of the cloud computing cluster. This not only saves manpower but also significantly improves the fault diagnosis result.
[0075] In one embodiment, to more accurately determine the representation of the first vector, such as Figure 3 As shown, one optional implementation of determining the first vector representation corresponding to the log text data in S202 includes:
[0076] S301, Perform semantic analysis on the log text data to obtain the semantic analysis results and initial semantic vector of the log text data.
[0077] Optionally, in this embodiment, the log text data is embedded using the SBERT model and transformed into semantic vectors. ;in, It is log text data sorted based on time series. This is the initial semantic vector.
[0078] S302, Based on the semantic analysis results, determine the vector weights of the initial semantic vector.
[0079] Optionally, in this embodiment, if the semantic analysis result contains the target keyword, the vector weight of the initial semantic vector is determined to be a first value. If the semantic analysis result does not contain the target keyword, the vector weight of the initial semantic vector is determined to be a second value; wherein the first value is greater than the second value. For example, the weight vector can be determined using the following formula: In the formula, "keywords" represents the target keywords, such as "error" or "timeout," which are fault-related keywords. The initial semantic vector for this type is assigned the first weight, which is the value in the formula. The initial semantic vector weight for keywords that did not hit the target keyword is 1.
[0080] S303, according to the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0081] Optionally, in this embodiment, the first vector representation corresponding to the log text data can be characterized by the following formula: ;in, This represents the first vector representation.
[0082] In this embodiment, semantic analysis is performed on the log text data to obtain the semantic analysis results and initial semantic vectors. Based on the semantic analysis results, the vector weights of the initial semantic vectors are determined. According to the vector weights, the initial semantic vectors are weighted to obtain the first vector representation corresponding to the log text data. In this embodiment, by combining the semantic analysis results and adding vector weights to the initial semantic vectors, the expression of keywords is enhanced, making the semantic expression of the obtained vector representation more accurate and focusing more on fault-related information.
[0083] In one embodiment, to more accurately determine the fault diagnosis result, such as Figure 4 As shown, one optional implementation method for determining the fault diagnosis result of a cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation includes:
[0084] S401, determine the multimodal feature representation based on the first vector representation, the second vector representation, and the third vector representation.
[0085] Optionally, in this embodiment, the first vector representation, the second vector representation, and the third vector representation are mapped to a unified semantic space to obtain the mapped feature vector: Then, the mapped feature vectors are input into the Siamese network structure of CLIP (Contrastive Language–Image Pre-training), spatially aligning the mapped first, second, and third vector representations. These aligned vector representations are then concatenated to obtain the multimodal feature representation. The modal feature representation can be characterized by the following formula: .
[0086] S402, based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation, determine the fault diagnosis result of the cloud computing cluster.
[0087] Optionally, in this embodiment, a first diagnostic result and its corresponding first confidence level are determined based on a first vector representation and a first classifier. A second diagnostic result and its corresponding second confidence level are determined based on a second vector representation and a second classifier. A third diagnostic result and its corresponding third confidence level are determined based on a third vector representation and a third classifier. A fourth diagnostic result and its corresponding fourth confidence level are determined based on a multimodal feature representation and a perceptron network. The fault diagnosis result of the cloud computing cluster is determined based on the first diagnostic result, first confidence level, second diagnostic result, second confidence level, third diagnostic result, third confidence level, fourth diagnostic result, and fourth confidence level. An optional implementation of determining the fourth diagnostic result and its corresponding fourth confidence level based on the multimodal feature representation and perceptron network in this embodiment is to generate weights for the multimodal feature representation through a gating network. Among them, w g This represents the weight tensor. The multimodal feature representations are weighted and fused, then input into an MLP (Multi-Layer Perceptron) to obtain the fourth diagnostic result and the fourth confidence score. The fourth diagnostic result can be represented by the following formula: In this embodiment, an optional implementation method for determining the fault diagnosis result of the cloud computing cluster based on the first diagnostic result, first confidence level, second diagnostic result, second confidence level, third diagnostic result, third confidence level, fourth diagnostic result, and fourth confidence level is as follows: Optionally, the first diagnostic result, second diagnostic result, and third diagnostic result each contain multiple fault types, and each fault type has a corresponding confidence level. In this embodiment, for each fault type, the confidence levels of the first diagnostic result, second diagnostic result, and third diagnostic result regarding that fault type are weighted and summed to obtain the final confidence level corresponding to that fault type. The expression for the final confidence level can be represented by the following formula: ;in, These represent the first confidence level of the fault type in the first diagnostic result, the second confidence level in the second diagnostic result, and the third confidence level in the third diagnostic result, respectively. The weight of this fault type in the first diagnostic result; The weight of this fault type in the second diagnostic result; The weight of the fault type in the third diagnostic result is determined; based on the final confidence level of each fault type calculated in the above embodiments and the fourth confidence level in the fourth diagnostic result, the fault type with the highest confidence level is selected as the target fault type, and a fault diagnosis result is generated based on the target fault type.
[0088] Optionally, in some embodiments, a fault diagnosis method may be implemented by determining a fault repair strategy based on the fault diagnosis results. Specifically, it first receives data from a knowledge graph. The topological structure information, combined with the abnormal indicators identified by the monitoring system, determines the root cause candidate set. Based on the causal propagation path information in the knowledge graph, calculate the set of all affected nodes:
[0089] in This is a set propagation probability threshold used to filter out non-critical impact paths. Subsequently, the multi-agent reinforcement learning model MADDPG is invoked to observe the state of each affected node and infer the optimal repair action: Finally, the nodes and their corresponding actions are combined into semantically interpretable repair strategy triples: in This indicates the execution order or priority, used for subsequent sorting and scheduling; combined with the reward function, the repair priority is calculated based on the scope of impact, cost, and SLA rules. To determine the execution order of repair actions, business impact and resource costs must be comprehensively considered. The priority scoring formula is defined as follows:
[0090] in: This indicates the degree of impact of the node on system stability (such as the number of services it depends on). This indicates the resource overhead required to perform the action; This indicates the compatibility score (0-1) of the operation in the current cloud environment, with a score closer to 1 indicating greater safety. , , These are dynamically adjusted weighting coefficients. SLA priority protection: For nodes that may lead to SLA violations (i.e., ... Unless resources are completely insufficient, repairs should be prioritized. All repair actions are scored and ranked using the formula above, determining which high-priority operations are executed first, thus achieving more effective fault recovery within a limited timeframe.
[0091] In this embodiment, a multimodal feature representation is determined based on the first vector representation, the second vector representation, and the third vector representation. The fault diagnosis result of the cloud computing cluster is then determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation. This embodiment makes the obtained fault diagnosis result more accurate.
[0092] In one embodiment, such as Figure 5 As shown, one optional implementation of a fault diagnosis method includes:
[0093] S501 retrieves log text data, time-series monitoring data, and network topology data from the cloud computing cluster.
[0094] S502, perform semantic analysis on the log text data to obtain the semantic analysis results and initial semantic vector of the log text data.
[0095] S503, if the semantic analysis results contain the target keyword, determine the vector weight of the initial semantic vector as the first value.
[0096] S504, if the semantic analysis results do not contain the target keyword, determine the vector weight of the initial semantic vector as the second value. The first value is greater than the second value.
[0097] S505, according to the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0098] S506 performs sliding window standardization on the time-series monitoring data to obtain the monitoring data sequence.
[0099] S507 encodes the monitoring data sequence to obtain the second vector representation corresponding to the time-series monitoring data.
[0100] S508 obtains the graph vector of the cloud computing cluster based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology.
[0101] S509 uses the graph vector of the cloud computing cluster as the third vector representation of the network topology data.
[0102] S510, determine the multimodal feature representation based on the first vector representation, the second vector representation, and the third vector representation.
[0103] S511, based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation, determine the fault diagnosis result of the cloud computing cluster.
[0104] S512, based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result.
[0105] S513, based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result.
[0106] S514, based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result.
[0107] S515, based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the fourth confidence level corresponding to the fourth diagnostic result.
[0108] S516. Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, determine the fault diagnosis result of the cloud computing cluster.
[0109] This application acquires log text data, time-series monitoring data, and network topology data from a cloud computing cluster. It determines a first vector representation for the log text data, a second vector representation for the time-series monitoring data, and a third vector representation for the network topology data. Based on these representations, it determines the fault diagnosis result for the cloud computing cluster. This fault diagnosis method requires no manual intervention. The server automatically determines the fault diagnosis result by acquiring the log text data, time-series monitoring data, and network topology data from the cloud computing cluster, and determining the corresponding vector representations. This not only saves manpower but also significantly improves the accuracy of fault diagnosis.
[0110] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0111] Based on the same inventive concept, this application also provides a fault diagnosis device for implementing the fault diagnosis method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more fault diagnosis device embodiments provided below can be found in the limitations of the fault diagnosis method described above, and will not be repeated here.
[0112] In one embodiment, such as Figure 6 As shown, a fault diagnosis device 1 is provided, comprising: an acquisition module 10, a first determination module 20, and a second determination module 30, wherein:
[0113] The acquisition module 10 is used to acquire log text data, time-series monitoring data, and network topology data of the cloud computing cluster.
[0114] The first determining module 20 is used to determine the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data;
[0115] The second determining module 30 is used to determine the fault diagnosis results of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation.
[0116] The fault diagnosis device of this application acquires log text data, time-series monitoring data, and network topology data of a cloud computing cluster. It determines a first vector representation corresponding to the log text data, a second vector representation corresponding to the time-series monitoring data, and a third vector representation corresponding to the network topology data. Based on these three vector representations, the fault diagnosis result of the cloud computing cluster is determined. This fault diagnosis method requires no manual intervention. The server acquires the log text data, time-series monitoring data, and network topology data of the cloud computing cluster, determines the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data, and then automatically determines the fault diagnosis result of the cloud computing cluster. This not only saves manpower but also significantly improves the fault diagnosis results.
[0117] In one embodiment, the first determining module is further specifically used for:
[0118] Semantic analysis is performed on the log text data to obtain the semantic analysis results and initial semantic vectors of the log text data;
[0119] Based on the semantic analysis results, determine the vector weights of the initial semantic vector;
[0120] Based on the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0121] In one embodiment, the first determining module is further specifically used for:
[0122] If the semantic analysis results contain the target keywords, the vector weight of the initial semantic vector is determined to be the first value;
[0123] If the semantic analysis results do not contain the target keyword, the vector weight of the initial semantic vector is determined to be the second value; where the first value is greater than the second value.
[0124] In one embodiment, the first determining module is further specifically used for:
[0125] The time-series monitoring data is standardized using a sliding window process to obtain the monitoring data sequence;
[0126] The monitoring data sequence is encoded to obtain the second vector representation corresponding to the time-series monitoring data.
[0127] In one embodiment, the network topology data includes node operation data and communication data of communication paths of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; wherein, the communication path is the communication path between cloud computing nodes.
[0128] The first determining module is also specifically used for:
[0129] Based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology, the graph vector of the cloud computing cluster is obtained.
[0130] The graph vector of the cloud computing cluster is used as the third vector representation of the network topology data.
[0131] In one embodiment, the second determining module is further specifically used for:
[0132] Based on the first vector representation, the second vector representation, and the third vector representation, determine the multimodal feature representation;
[0133] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation.
[0134] In one embodiment, the second determining module is further specifically used for:
[0135] Based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result;
[0136] Based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result;
[0137] Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result;
[0138] Based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the corresponding fourth confidence level;
[0139] Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
[0140] Each module in the aforementioned fault diagnosis device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.
[0141] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores information related to cloud computing clusters. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a fault diagnosis method.
[0142] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0143] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0144] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0145] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0146] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0147] In one embodiment, when the processor executes the computer program, it further performs the following steps: determining a first vector representation corresponding to the log text data, including:
[0148] Semantic analysis is performed on the log text data to obtain the semantic analysis results and initial semantic vectors of the log text data;
[0149] Based on the semantic analysis results, determine the vector weights of the initial semantic vector;
[0150] Based on the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0151] In one embodiment, when the processor executes the computer program, it further performs the following steps: determining the vector weights of the initial semantic vector based on the semantic analysis results, including:
[0152] If the semantic analysis results contain the target keywords, the vector weight of the initial semantic vector is determined to be the first value;
[0153] If the semantic analysis results do not contain the target keyword, the vector weight of the initial semantic vector is determined to be the second value; where the first value is greater than the second value.
[0154] In one embodiment, when the processor executes the computer program, it further performs the following steps: determining a second vector representation corresponding to the timing monitoring data, including:
[0155] The time-series monitoring data is standardized using a sliding window process to obtain the monitoring data sequence;
[0156] The monitoring data sequence is encoded to obtain the second vector representation corresponding to the time-series monitoring data.
[0157] In one embodiment, when the processor executes the computer program, it further implements the following steps: the network topology data includes node operation data and communication data of communication paths of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; wherein, the communication path is the communication path between cloud computing nodes.
[0158] Determine the third vector representation corresponding to the network topology data, including:
[0159] Based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology, the graph vector of the cloud computing cluster is obtained.
[0160] The graph vector of the cloud computing cluster is used as the third vector representation of the network topology data.
[0161] In one embodiment, when the processor executes the computer program, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation, including:
[0162] Based on the first vector representation, the second vector representation, and the third vector representation, determine the multimodal feature representation;
[0163] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation.
[0164] In one embodiment, when the processor executes the computer program, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation, including:
[0165] Based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result;
[0166] Based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result;
[0167] Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result;
[0168] Based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the corresponding fourth confidence level;
[0169] Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
[0170] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:
[0171] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0172] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0173] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0174] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining a first vector representation corresponding to the log text data, including:
[0175] Semantic analysis is performed on the log text data to obtain the semantic analysis results and initial semantic vectors of the log text data;
[0176] Based on the semantic analysis results, determine the vector weights of the initial semantic vector;
[0177] Based on the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0178] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the vector weights of the initial semantic vector based on the semantic analysis results, including:
[0179] If the semantic analysis results contain the target keywords, the vector weight of the initial semantic vector is determined to be the first value;
[0180] If the semantic analysis results do not contain the target keyword, the vector weight of the initial semantic vector is determined to be the second value; where the first value is greater than the second value.
[0181] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: determining a second vector representation corresponding to the timing monitoring data, including:
[0182] The time-series monitoring data is standardized using a sliding window process to obtain the monitoring data sequence;
[0183] The monitoring data sequence is encoded to obtain the second vector representation corresponding to the time-series monitoring data.
[0184] In one embodiment, when the computer program is executed by the processor, it further implements the following steps: the network topology data includes node operation data and communication data of communication paths of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; wherein, the communication path is the communication path between cloud computing nodes.
[0185] Determine the third vector representation corresponding to the network topology data, including:
[0186] Based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology, the graph vector of the cloud computing cluster is obtained.
[0187] The graph vector of the cloud computing cluster is used as the third vector representation of the network topology data.
[0188] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation, including:
[0189] Based on the first vector representation, the second vector representation, and the third vector representation, determine the multimodal feature representation;
[0190] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation.
[0191] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation, including:
[0192] Based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result;
[0193] Based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result;
[0194] Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result;
[0195] Based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the corresponding fourth confidence level;
[0196] Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
[0197] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:
[0198] Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster;
[0199] Determine the first vector representation corresponding to log text data, the second vector representation corresponding to time-series monitoring data, and the third vector representation corresponding to network topology data;
[0200] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, and the third vector representation.
[0201] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining a first vector representation corresponding to the log text data, including:
[0202] Semantic analysis is performed on the log text data to obtain the semantic analysis results and initial semantic vectors of the log text data;
[0203] Based on the semantic analysis results, determine the vector weights of the initial semantic vector;
[0204] Based on the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
[0205] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the vector weights of the initial semantic vector based on the semantic analysis results, including:
[0206] If the semantic analysis results contain the target keywords, the vector weight of the initial semantic vector is determined to be the first value;
[0207] If the semantic analysis results do not contain the target keyword, the vector weight of the initial semantic vector is determined to be the second value; where the first value is greater than the second value.
[0208] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: determining a second vector representation corresponding to the timing monitoring data, including:
[0209] The time-series monitoring data is standardized using a sliding window process to obtain the monitoring data sequence;
[0210] The monitoring data sequence is encoded to obtain the second vector representation corresponding to the time-series monitoring data.
[0211] In one embodiment, when the computer program is executed by the processor, it further implements the following steps: the network topology data includes node operation data and communication data of communication paths of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; wherein, the communication path is the communication path between cloud computing nodes.
[0212] Determine the third vector representation corresponding to the network topology data, including:
[0213] Based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology, the graph vector of the cloud computing cluster is obtained.
[0214] The graph vector of the cloud computing cluster is used as the third vector representation of the network topology data.
[0215] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation, including:
[0216] Based on the first vector representation, the second vector representation, and the third vector representation, determine the multimodal feature representation;
[0217] The fault diagnosis results of the cloud computing cluster are determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation.
[0218] In one embodiment, when the computer program is executed by a processor, it further performs the following steps: determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation, including:
[0219] Based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result;
[0220] Based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result;
[0221] Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result;
[0222] Based on multimodal feature representation and perceptron network, determine the fourth diagnostic result and the corresponding fourth confidence level;
[0223] Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
[0224] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchain. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., and are not limited to these.
[0225] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0226] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A fault diagnosis method, characterized in that, The method includes: Obtain log text data, time-series monitoring data, and network topology data from the cloud computing cluster; Determine the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data; The fault diagnosis result of the cloud computing cluster is determined based on the first vector representation, the second vector representation, and the third vector representation.
2. The method according to claim 1, characterized in that, Determining the first vector representation corresponding to the log text data includes: Perform semantic analysis on the log text data to obtain the semantic analysis results and initial semantic vector of the log text data; Based on the semantic analysis results, determine the vector weights of the initial semantic vector; Based on the vector weights, the initial semantic vector is weighted to obtain the first vector representation corresponding to the log text data.
3. The method according to claim 2, characterized in that, The step of determining the vector weights of the initial semantic vector based on the semantic analysis results includes: If the semantic analysis result contains the target keyword, the vector weight of the initial semantic vector is determined to be a first value; If the semantic analysis result does not contain the target keyword, the vector weight of the initial semantic vector is determined to be a second value; wherein the first value is greater than the second value.
4. The method according to claim 1, characterized in that, Determining the second vector representation corresponding to the time-series monitoring data includes: The time-series monitoring data is subjected to sliding window normalization to obtain a monitoring data sequence; The monitoring data sequence is encoded to obtain a second vector representation corresponding to the time-series monitoring data.
5. The method according to claim 1, characterized in that, The network topology data includes node operation data and communication data of communication paths of the cloud computing nodes contained in the cloud computing cluster, as well as the network topology structure of the cloud computing cluster; wherein, the communication path is the communication path between the cloud computing nodes. Determining the third vector representation corresponding to the network topology data includes: Based on the node operation data of each cloud computing node, the communication data of each communication path, and the network topology, the graph vector of the cloud computing cluster is obtained. The graph vector of the cloud computing cluster is used as the third vector representation of the network topology data.
6. The method according to claim 1, characterized in that, Determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation includes: Based on the first vector representation, the second vector representation, and the third vector representation, a multimodal feature representation is determined; The fault diagnosis result of the cloud computing cluster is determined based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation.
7. The method according to claim 6, characterized in that, Determining the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, the third vector representation, and the multimodal feature representation includes: Based on the first vector representation and the first classifier, determine the first diagnostic result and the first confidence level corresponding to the first diagnostic result; Based on the second vector representation and the second classifier, determine the second diagnostic result and the second confidence level corresponding to the second diagnostic result; Based on the third vector representation and the third classifier, determine the third diagnostic result and the third confidence level corresponding to the third diagnostic result; Based on the multimodal feature representation and the perceptron network, a fourth diagnostic result and a fourth confidence level corresponding to the fourth diagnostic result are determined; Based on the first diagnostic result, the first confidence level, the second diagnostic result, the second confidence level, the third diagnostic result, the third confidence level, the fourth diagnostic result, and the fourth confidence level, the fault diagnosis result of the cloud computing cluster is determined.
8. A fault diagnosis device, characterized in that, include: The acquisition module is used to acquire log text data, time-series monitoring data, and network topology data of the cloud computing cluster. The first determining module is used to determine the first vector representation corresponding to the log text data, the second vector representation corresponding to the time-series monitoring data, and the third vector representation corresponding to the network topology data; The second determining module is used to determine the fault diagnosis result of the cloud computing cluster based on the first vector representation, the second vector representation, and the third vector representation.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the fault diagnosis method according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the fault diagnosis method according to any one of claims 1 to 7.