CN118093311A - Intelligent fault healing method, device, equipment, storage medium and product - Google Patents

Abstract

The invention discloses an intelligent fault cure method, a device, equipment, a storage medium and a computer program product, wherein the method is implemented by collecting application service log information of an application log file; configuring a log word frequency analysis strategy according to the application service log information; performing application log word frequency analysis according to the log word frequency analysis strategy to obtain word frequency monitoring indexes; training according to the word frequency monitoring index to obtain a word frequency detection model; and performing abnormality detection through the word frequency detection model, and performing fault cure according to the detection result. In this way, the log data summarizing and exposing all applications identifies potential problems and anomalies and enables targeted fault healing. Abnormality and fault can be found in time, and response and processing can be fast. The requirement of manual intervention is reduced, the operation and maintenance efficiency and accuracy are improved, and the performance and user experience of the system are improved.

Description

Intelligent fault healing method, device, equipment, storage medium and product

Technical Field

The present invention relates to the field of cloud platform technology, and in particular to an intelligent fault recovery method, device, equipment, storage medium and computer program product.

Background technique

When a cloud platform system fails or is abnormal, the operation and maintenance personnel need to spend a long time troubleshooting, repairing and restoring the system, and promptly notify relevant personnel, which usually requires downtime for maintenance or affects business operations, resulting in business continuity being affected. This not only consumes time and energy, but is also prone to operational errors or omissions. This approach has certain limitations in terms of efficiency, reliability and scalability, and cannot meet the needs of modern enterprises for efficiency, stability and flexibility.

Summary of the invention

The main purpose of the present invention is to provide an intelligent fault recovery method, device, equipment, storage medium and computer program product, aiming to solve the technical problem of low timeliness of fault processing in the prior art.

To achieve the above object, the present invention provides an intelligent fault healing method, which comprises the following steps:

Configure the log word frequency analysis strategy based on the application service log information collected from the application log file;

Perform application log word frequency analysis according to the log word frequency analysis strategy to obtain word frequency monitoring indicators;

A word frequency detection model is obtained by training according to the word frequency monitoring indicator;

Anomaly detection is performed using the word frequency detection model, and fault recovery is performed based on the detection result.

Optionally, performing application log word frequency analysis according to the log word frequency analysis strategy to obtain a word frequency monitoring indicator includes:

Input the application service log information into the Kafka cluster, and consume the cluster application log of the Kafka cluster according to the stream processing application library to obtain consumption data;

Determining log word rules according to the log word frequency analysis strategy;

A log word frequency monitoring indicator is generated according to the consumption data and the log word rule.

Optionally, generating a log word frequency monitoring indicator according to the consumption data and the log word rule includes:

comparing the consumption data with the log word rule;

determining compliance log data based on the comparison results;

The compliant log data is stored in a target cluster, and a log word frequency monitoring index is generated through a log word frequency statistics service and cluster data of the target cluster.

Optionally, the training of a word frequency detection model according to the word frequency monitoring indicator includes:

Obtaining historical fault records, and generating training samples according to the historical fault records and the word frequency monitoring index;

A word frequency detection model is obtained by training according to the training samples.

Optionally, generating a training sample according to the historical fault record and the word frequency monitoring indicator includes:

Determine the original indicator data time series according to the word frequency monitoring indicator;

Divide the original indicator data time series into multiple sample windows by using a sliding window method;

Determine an abnormal window according to the historical fault records;

Determine a normal window according to the sample window and the abnormal window;

A training sample is determined according to the normal window.

Optionally, performing anomaly detection by using the word frequency detection model and performing fault recovery according to the detection result includes:

Inputting the predicted sample into the word frequency detection model to obtain an output abnormal sample;

Determine a target fault healing action according to the abnormal sample, and perform fault healing by triggering the target fault healing action. In addition, to achieve the above purpose, the present invention also proposes an intelligent fault healing device, the intelligent fault healing device comprising:

A policy configuration module, used to configure a log word frequency analysis policy based on application service log information collected from application log files;

An indicator analysis module, used to perform application log word frequency analysis according to the log word frequency analysis strategy to obtain a word frequency monitoring indicator;

A model training module, used to train a word frequency detection model according to the word frequency monitoring indicator;

The fault recovery module is used to perform anomaly detection through the word frequency detection model and to recover the fault according to the detection result.

In addition, to achieve the above-mentioned objectives, the present invention also proposes an intelligent fault healing device, which includes: a memory, a processor, and an intelligent fault healing program stored in the memory and executable on the processor, wherein the intelligent fault healing program is configured to implement the steps of the intelligent fault healing method described above.

In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, on which an intelligent fault healing program is stored, and when the intelligent fault healing program is executed by a processor, the steps of the intelligent fault healing method described above are implemented.

In addition, to achieve the above objectives, the present invention also provides a computer program product, which includes an intelligent fault healing program, and when the intelligent fault healing program is executed by a processor, the steps of the intelligent fault healing method described above are implemented.

The present invention collects application service log information of application log files; configures a log word frequency analysis strategy according to the application service log information; performs application log word frequency analysis according to the log word frequency analysis strategy to obtain a word frequency monitoring index; trains a word frequency detection model according to the word frequency monitoring index; performs anomaly detection through the word frequency detection model, and performs fault healing according to the detection result. In this way, a centralized log management and monitoring function is provided, which can summarize and display the log data of all applications to identify potential problems and anomalies, and perform targeted fault healing. Anomalies and faults can be discovered in time and can be responded to and handled quickly. By applying log word frequency analysis and training of algorithm models, the automated process of fault healing is realized, reducing the need for manual intervention, improving operation and maintenance efficiency and accuracy, and enhancing system performance and user experience.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of the structure of an intelligent fault recovery device in a hardware operating environment according to an embodiment of the present invention;

FIG2 is a schematic diagram of a flow chart of a first embodiment of an intelligent fault recovery method according to the present invention;

FIG3 is a schematic diagram of a flow chart of a second embodiment of an intelligent fault recovery method according to the present invention;

FIG4 is a structural block diagram of the first embodiment of the intelligent fault recovery device of the present invention.

The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings.

Detailed ways

It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

Refer to FIG. 1 , which is a schematic diagram of the structure of an intelligent fault recovery device in a hardware operating environment according to an embodiment of the present invention.

As shown in FIG1 , the intelligent fault healing device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, WI-FI) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk storage. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

Those skilled in the art will appreciate that the structure shown in FIG. 1 does not constitute a limitation on the intelligent fault healing device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

As shown in FIG. 1 , the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and an intelligent fault recovery program.

In the intelligent fault healing device shown in Figure 1, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the intelligent fault healing device of the present invention can be set in the intelligent fault healing device, and the intelligent fault healing device calls the intelligent fault healing program stored in the memory 1005 through the processor 1001, and executes the intelligent fault healing method provided by the embodiment of the present invention.

An embodiment of the present invention provides an intelligent fault healing method. Referring to FIG. 2 , FIG. 2 is a schematic flow chart of a first embodiment of the intelligent fault healing method of the present invention.

In this embodiment, the intelligent fault recovery method includes the following steps:

Step S10: configuring a log word frequency analysis strategy according to the application service log information of the collected application log file.

It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, an intelligent fault healing device, etc. The following takes the intelligent fault healing device as an example to illustrate this embodiment and the following embodiments.

It should be understood that when a system fails or is abnormal, the operation and maintenance personnel need to spend a long time to troubleshoot, repair and restore the system, and notify relevant personnel in a timely manner. This usually requires downtime maintenance or affects business operations, resulting in business continuity being affected. This not only consumes time and energy, but is also prone to operational errors or omissions. This approach has certain limitations in terms of efficiency, reliability and scalability, and cannot meet the needs of modern enterprises for efficiency, stability and flexibility. In traditional operation and maintenance methods, fault handling usually relies on manual judgment and decision-making, involving complex tasks such as fault perception and traffic scheduling. However, manual processing has the problem of low timeliness, which may lead to slow service recovery, and human factors may also cause the expansion of problems. In order to solve this problem, fault healing has become an industry-leading solution, namely, automated fault handling. Through automated processing, the pre-set recovery process can ensure that the recovery process is more reliable. Using word frequency analysis technology and adaptive fault healing algorithms, faults can be located and restored more quickly, thereby improving the service availability of the enterprise and reducing dependence on human resources, and achieving unattended fault healing. Such a solution can effectively improve operation and maintenance efficiency and service quality, saving costs and manpower investment for enterprises. The log collector fluentd of the scheme of this embodiment collects the log data generated by the application. Through effective application log collection, the reliability, stability and security of the system can be improved. Application log word frequency analysis configuration management supports specifying phrases/regulars with spaces for word frequency analysis, supports specifying additional fields as statistical dimensions, and adds a configuration alias mechanism to facilitate word frequency statistical queries. The distributed word frequency analysis stream processing system uses KafkaStream as the data processing framework. The system analyzes the log data of the target application with the latest configuration, and stores the statistical dimension-related fields and aliases in Elasticsearch so that the statistical range can be delineated based on this information. Based on the analyzed data, word frequency monitoring indicator data is generated. Based on the application log word frequency data, combined with the application's fault point data, deep learning technology is introduced to train the algorithm, and an autoencoder and generative adversarial network cascade model is constructed. After the model is trained, both the autoencoder and the discriminator in the model can make abnormal judgments on the predicted samples. For normal samples, the reconstruction error calculated by the autoencoder is small, and a large one is an abnormal sample. For abnormal samples, the system automatically performs the corresponding fault healing action.

It should be noted that the log collector Fluentd is used to collect, transmit and forward application log files, realize centralized management and analysis of logs, support multiple input and output plug-ins, and can be integrated with various data sources and targets. Fluentd also has flexible data conversion and filtering functions, which can process, filter and format log data according to needs.

It should be understood that the log collector fluentd is deployed on each node of the Kubernetes (K8S) cluster in DaemonSet mode.

In a specific implementation, when the application image is running, the corresponding log directory needs to be mounted in the form of a volume, and the application log is written to the host disk in the form of a file.

It should be noted that the collector engine is responsible for collecting log files on the host disk and transmitting the collected log data to the Kafka cluster in the unified log cluster.

It should be understood that the word frequency analysis strategy for the application log is configured through the page, and the word frequency analysis strategy is persistently stored for the word frequency engine to analyze the application log.

In the specific implementation, it supports specifying phrases with spaces or regular expressions for word frequency analysis. You can specify specific phrases or regular expressions as needed to perform word frequency statistics of keywords in the log.

It should be noted that it supports specifying additional fields as statistical dimensions. In addition to the default word frequency statistics, you can specify additional fields as statistical dimensions as needed to analyze log data in more detail.

It should be understood that the configuration alias mechanism is added to facilitate word frequency statistics query. Aliases can be added to configuration items to facilitate users to use more intuitive and easy-to-understand names when querying, improving the readability and maintainability of the configuration.

Step S20: Performing application log word frequency analysis according to the log word frequency analysis strategy to obtain a word frequency monitoring index.

It should be noted that the distributed word frequency analysis stream processing system uses KafkaStream as the data processing framework. The system analyzes the log data of the target application with the latest configuration, and stores the relevant fields and aliases of the statistical dimensions in Elasticsearch so that the statistical scope can be defined based on this information. Based on the analyzed data, word frequency monitoring indicator data is generated.

Furthermore, in order to accurately obtain the word frequency monitoring index, step S20 includes: inputting the application service log information into the Kafka cluster, and consuming the cluster application log of the Kafka cluster according to the stream processing application library to obtain consumption data; determining the log word rule according to the log word frequency analysis strategy; generating the log word frequency monitoring index according to the consumption data and the log word rule.

It should be understood that log word frequency analysis uses Kafka Stream for processing. Kafka Stream is a library for building real-time stream processing applications that can transform and aggregate stream data. In log word frequency analysis, application logs are used as input streams, combined with the configuration of application log word frequency analysis strategies, and processed and analyzed through Kafka Stream to implement word frequency statistics. It can process large-scale log data in real time, extract keywords and calculate word frequencies.

In the specific implementation, the log word frequency analysis engine Kafka-Stream is deployed on the Kubernetes cluster in Deployment mode, and the volumeMounts mounting information is specified.

It should be noted that the configuration files are mounted in the form of ConfigMap, including db.properties and config.properties files, which specify the address information of the db, Kafka, and elasticsearch clusters.

It should be understood that the application logs are collected by the collection application service and then output to the Kafka cluster.

It should be noted that the log word frequency analysis engine reads the log word frequency strategy configuration through a scheduled task and stores it in memory.

In the specific implementation, the log word frequency analysis engine uses Kafka Stream technology to consume messages from the Kafka cluster in real time, compare them with the log word frequency strategy configuration, extract log data that meets the requirements, and store them in the Elasticsearch cluster.

It should be understood that the application log word frequency statistics service obtains the word frequency configuration of the application and generates monitoring indicators. The generated monitoring indicators are subsequently used for algorithm training and anomaly detection, ultimately achieving the effect of curing application failures.

It should be noted that the monitoring indicators are output through the following process: 1) The word frequency statistics service starts the coroutine, regularly queries the word frequency configuration of the component, and obtains application information. 2) The word frequency statistics service calls the interface to query the current total word frequency. 3) The word frequency statistics service calls the interface to query the word frequency increment in the last 5 minutes from Elasticsearch, and calculates the total word frequency through the following formula. After calculating the total word frequency, the metrics service is exposed and output as a monitoring indicator. Formula: Total word frequency = current word frequency + word frequency increment.

Furthermore, in order to accurately generate log word frequency monitoring indicators, the steps of generating log word frequency monitoring indicators based on the consumption data and the log word rules include: comparing the consumption data with the log word rules; determining compliant log data based on the comparison results; storing the compliant log data into a target cluster, and generating log word frequency monitoring indicators through a log word frequency statistics service and the cluster data of the target cluster.

It should be understood that the application log is first input into the Kafka cluster in real time, and then the log word frequency configuration data is periodically read by the log analysis engine and stored in memory. The log analysis engine consumes the application log of the Kafka cluster in real time based on Kafka-Stream, and then compares the consumed data with the log word rules, extracts the log data that meets the rules, and stores it in the ES cluster. Finally, based on the ES data, the log word frequency monitoring indicator is generated.

Step S30: training a word frequency detection model according to the word frequency monitoring indicator.

In the specific implementation, in order to obtain the word frequency detection model, it is also necessary to introduce historical fault records to construct training samples, and then train to obtain the word frequency detection model.

Step S40: performing anomaly detection using the word frequency detection model, and performing fault recovery according to the detection result.

It should be noted that after the word frequency detection model is trained, the predicted samples are imported, abnormal samples are found, and then the corresponding fault healing actions are automatically called to achieve automatic fault healing.

Furthermore, in order to achieve automatic detection and fault healing, step S40 includes: inputting the predicted sample into the word frequency detection model to obtain an output abnormal sample; determining a target fault healing action according to the abnormal sample, and performing fault healing by triggering the target fault healing action.

It should be understood that after completing the model training, both the autoencoder and the discriminator in the model can make anomaly judgments on the predicted samples, and based on the results of anomaly detection, perform intelligent fault healing of the application.

In the specific implementation, after the model training is completed, the autoencoder and discriminator in the model can both make abnormal judgments on the predicted samples. For normal samples, the reconstruction error calculated by the autoencoder is small, and a larger one is an abnormal sample. Similarly, normal samples can be encoded by the encoder into feature vectors that can confuse the discriminator, then the discriminator judges it to be true, and the abnormal samples will be judged to be false. Therefore, in the subsequent prediction process, the predicted sample q is input into the model, based on the output D(E(p)) of the autoencoder network, and the abnormal score s1 is calculated; in addition, based on the output G(E(p)) of the adversarial network, the abnormal score s2 is calculated at the same time. Finally, the two parts of the abnormal score are summarized by weighted averaging to obtain the final abnormal score s. When the abnormal score s is greater than the given threshold, it is judged that the predicted sample is abnormal and the fault recovery is triggered, otherwise it is not triggered.

s1＝mse(p,D(E(P)))

s2＝log(1-G(E(p)))

s＝α*s1+(1-α)*s2

It should be noted that the system presets a variety of fault recovery actions and associates them with corresponding algorithm models. The fault recovery actions are triggered based on the results of anomaly detection. For example, based on the keywords related to "access timeout" in the log word frequency, a corresponding algorithm model is built. The sample data is tested for anomalies to predict when "access timeout" will occur in the future of the application. The application's fault recovery actions are triggered in advance before the failure occurs.

This embodiment collects application service log information from application log files; configures a log word frequency analysis strategy based on the application service log information; performs application log word frequency analysis based on the log word frequency analysis strategy to obtain word frequency monitoring indicators; trains a word frequency detection model based on the word frequency monitoring indicators; performs anomaly detection through the word frequency detection model, and performs fault recovery based on the detection results. In this way, a centralized log management and monitoring function is provided, which can summarize and display the log data of all applications to identify potential problems and anomalies, and perform targeted fault recovery. Anomalies and faults can be discovered in a timely manner and can be responded to and handled quickly. By applying log word frequency analysis and training of algorithm models, the automated process of fault recovery is realized, reducing the need for manual intervention, improving operation and maintenance efficiency and accuracy, and enhancing system performance and user experience.

Refer to FIG3 , which is a flow chart of a second embodiment of the intelligent fault recovery method of the present invention.

Based on the above first embodiment, in this embodiment, step S30 includes:

Step S301: Obtain historical fault records, and generate training samples according to the historical fault records and the word frequency monitoring index.

It should be noted that based on the application log word frequency monitoring indicators, combined with the application failure data points, deep learning technology is introduced to train the algorithm and build a cascade model of autoencoder and generative adversarial network.

Furthermore, in order to accurately obtain training samples, step S301 includes: determining the original indicator data time series according to the word frequency monitoring indicator; dividing the original indicator data time series into multiple sample windows using a sliding window method; determining the abnormal window according to the historical fault records; determining the normal window according to the sample window and the abnormal window; and determining the training sample according to the normal window.

It should be understood that based on historical fault records, the real historical fault points are marked. Generally speaking, anomaly detection allows for delays within a controllable range. Therefore, we divide the original indicator data time series into multiple small window samples using a sliding window. The size of the sliding window can be determined within a controllable range based on actual conditions. Then, the window with the real historical fault point is marked as an abnormal window, and all normal windows are combined into a training sample X = { _X1 , _X2 , ..., _Xn }.

Step S302: obtaining a word frequency detection model by training according to the training samples.

In the specific implementation, in the first step, the training sample X is input into the model, and the encoder extracts the feature vector output E(x) from the training sample, which is used as the input of the decoder and discriminator. In the second step, the decoder restores the feature vector and outputs D(E(x)), compares it with the original input to calculate the reconstruction loss LOSS1, and updates the parameters of the encoder and decoder; the feature vector in the first step is input into the discriminator to generate the output G(E(x)), and the vector z is sampled from the Gaussian mixture distribution and input into the discriminator to generate the output G(z). The two calculate the loss LOSS2, which is used to update the discriminator parameters, and then calculate the loss LOSS3, which is used to update the encoder parameters.

in,

LOSS1＝mse(x,D(E(x))

LOSS2＝-log G(z)-log(1-G(E(x)))

LQSS3＝log(1-G(E(x)))

This embodiment obtains historical fault records, generates training samples based on the historical fault records and the word frequency monitoring index, and obtains a word frequency detection model through training based on the training samples. In this way, the automatic process of fault recovery is realized by applying log word frequency analysis and algorithm model training.

In addition, an embodiment of the present invention further provides a storage medium, on which an intelligent fault healing program is stored. When the intelligent fault healing program is executed by a processor, the steps of the intelligent fault healing method described above are implemented.

In addition, an embodiment of the present invention further provides a computer program product, including an intelligent fault healing program, which implements the steps of the intelligent fault healing method described above when executed by a processor.

The specific implementation methods of the computer program product of the present invention are basically the same as the above-mentioned embodiments of the intelligent fault healing method, and will not be repeated here.

Refer to FIG. 4 , which is a structural block diagram of a first embodiment of an intelligent fault healing device according to the present invention.

As shown in FIG4 , the intelligent fault recovery device proposed in the embodiment of the present invention includes:

The strategy configuration module 10 is used to configure the log word frequency analysis strategy according to the application service log information collected from the application log file.

The indicator analysis module 20 is used to perform word frequency analysis on the application log according to the log word frequency analysis strategy to obtain a word frequency monitoring indicator.

The model training module 30 is used to train a word frequency detection model according to the word frequency monitoring indicator.

The fault recovery module 40 is used to perform anomaly detection through the word frequency detection model and to recover the fault according to the detection result.

In this embodiment, application service log information of application log files is collected; a log word frequency analysis strategy is configured according to the application service log information; application log word frequency analysis is performed according to the log word frequency analysis strategy to obtain word frequency monitoring indicators; a word frequency detection model is trained according to the word frequency monitoring indicators; anomaly detection is performed through the word frequency detection model, and fault recovery is performed according to the detection results. In this way, a centralized log management and monitoring function is provided, which can summarize and display the log data of all applications to identify potential problems and anomalies, and perform targeted fault recovery. Anomalies and faults can be discovered in a timely manner and can be responded to and handled quickly. By applying log word frequency analysis and training of algorithm models, the automated process of fault recovery is realized, reducing the need for manual intervention, improving operation and maintenance efficiency and accuracy, and enhancing system performance and user experience.

In one embodiment, the indicator analysis module 20 is also used to input the application service log information into the Kafka cluster, and consume the cluster application log of the Kafka cluster according to the stream processing application library to obtain consumption data; determine the log word rule according to the log word frequency analysis strategy; and generate a log word frequency monitoring indicator according to the consumption data and the log word rule.

In one embodiment, the indicator analysis module 20 is also used to compare the consumption data with the log word rules; determine the compliant log data based on the comparison results; store the compliant log data in the target cluster, and generate a log word frequency monitoring indicator through the log word frequency statistics service and the cluster data of the target cluster.

In one embodiment, the model training module 30 is further used to obtain historical fault records, and generate training samples according to the historical fault records and the word frequency monitoring index; and obtain a word frequency detection model by training according to the training samples.

In one embodiment, the model training module 30 is also used to determine the original indicator data time series based on the word frequency monitoring indicator; divide the original indicator data time series into multiple sample windows using a sliding window; determine the abnormal window based on the historical fault record; determine the normal window based on the sample window and the abnormal window; and determine the training sample based on the normal window.

In one embodiment, the fault healing module 40 is further used to input the predicted sample into the word frequency detection model to obtain an output abnormal sample; determine a target fault healing action according to the abnormal sample, and perform fault healing by triggering the target fault healing action.

Other embodiments or specific implementations of the intelligent fault healing device of the present invention can refer to the above-mentioned method embodiments and will not be described in detail here.

It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory/random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims (10)

1. An intelligent fault cure method, characterized in that the intelligent fault cure method comprises:

configuring a log word frequency analysis strategy according to application service log information of an acquired application log file;

performing application log word frequency analysis according to the log word frequency analysis strategy to obtain word frequency monitoring indexes;

Training according to the word frequency monitoring index to obtain a word frequency detection model;

and performing abnormality detection through the word frequency detection model, and performing fault cure according to the detection result.

2. The intelligent fault healing method according to claim 1, wherein the applying log word frequency analysis according to the log word frequency analysis policy to obtain the word frequency monitoring index includes:

inputting the application service log information into a Kafka cluster, and consuming the cluster application log of the Kafka cluster according to a stream processing application program library to obtain consumption data;

Determining a log word rule according to the log word frequency analysis strategy;

And generating a log word frequency monitoring index according to the consumption data and the log word rule.

3. The intelligent fault-healing method of claim 2, wherein the generating a log word frequency monitor indicator from the consumption data and the log word rule comprises:

comparing the consumption data with the log word rule;

Determining compliance log data according to the comparison result;

And storing the compliance log data into a target cluster, and generating a log word frequency monitoring index through a log word frequency statistics service and cluster data of the target cluster.

4. The intelligent fault cure method of claim 1, wherein training according to the word frequency monitoring indicator to obtain a word frequency detection model comprises:

Acquiring a historical fault record, and generating a training sample according to the historical fault record and the word frequency monitoring index;

and training according to the training sample to obtain a word frequency detection model.

5. The intelligent fault-healing method of claim 4, wherein the generating training samples from the historical fault record and the word frequency monitoring indicator comprises:

Determining an original index data time sequence according to the word frequency monitoring index;

dividing the original index data time sequence into a plurality of sample windows in a sliding window mode;

Determining an abnormal window according to the historical fault record;

determining a normal window according to the sample window and the abnormal window;

And determining training samples according to the normal window.

6. The intelligent fault-healing method according to claim 1, wherein the abnormality detection by the word frequency detection model and the fault-healing according to the detection result comprise:

inputting the prediction sample into the word frequency detection model to obtain an output abnormal sample;

And determining a target fault healing action according to the abnormal sample, and performing fault healing by triggering the target fault healing action.

7. An intelligent fault cure device, characterized in that the intelligent fault cure device comprises:

the strategy configuration module is used for configuring a log word frequency analysis strategy according to the application service log information of the acquired application log file;

the index analysis module is used for carrying out application log word frequency analysis according to the log word frequency analysis strategy to obtain word frequency monitoring indexes;

the model training module is used for training according to the word frequency monitoring index to obtain a word frequency detection model;

and the fault cure module is used for carrying out abnormal detection through the word frequency detection model and carrying out fault cure according to the detection result.

8. An intelligent fault cure device, the device comprising: a memory, a processor, and a smart fault-healing program stored on the memory and executable on the processor, the smart fault-healing program configured to implement the steps of the smart fault-healing method of any one of claims 1 to 6.

9. A storage medium having stored thereon a smart fault cure program which when executed by a processor implements the steps of the smart fault cure method of any one of claims 1 to 6.

10. A computer program product comprising a smart fault cure program which when executed by a processor implements the steps of the smart fault cure method according to any one of claims 1 to 6.