Server test environment automatic monitoring method and system

Abstract

The invention relates to a server test environment automatic monitoring method and system. The method comprises the following steps: acquiring various data in a test environment of a plurality of servers in a specified period, and dividing the data into a training set, a test set and a verification set; the method comprises the following steps of: acquiring various monitored server test environments, inputting the monitored server test environments into a server test environment monitoring model for training, testing and verifying, acquiring various data in the monitored server test environments in real time, inputting the data into the server test environment monitoring model, and outputting monitoring results of the monitored server test environments in real time through the server test environment monitoring model. And the server obtains abnormal information according to the monitoring result, obtains an abnormal position and an abnormal severity degree of the monitoring test environment based on the abnormal information according to whether the abnormal information needs to be processed or not, analyzes an abnormal solution according to the abnormal position and abnormal data, and processes the abnormality based on the obtained abnormal solution. The server test environment monitoring efficiency is effectively improved, and the missed test rate is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of server test environment, and in particular relates to a method and system for automatically monitoring a server test environment. Background Art

[0002] Various professional tests in existing environments are mainly conducted by manual startup inspection and real-time monitoring of various functional points. It is best to monitor the test process 24 hours a day, which is quite time-consuming and laborious. If no inspection is performed in the middle, some functions may not be detected. At the same time, if the equipment is reset and restarted at night or during holidays, it is not monitored properly, or the network is abnormal when connected to the computer monitoring, resulting in the disconnection of the test tool, and the equipment operation status cannot be monitored properly. In the absence of personnel for 24-hour monitoring, it is difficult to determine the restart and reset time and abnormal conditions. At the same time, related services such as CPU, memory, disk, network card, etc. will stop after the reset and restart, resulting in the inability to fully load or carry business tests to meet the test requirements after the reset and restart, which can easily lead to inaccurate test results. In addition, if there are multiple restarts and resets during the test without records, there is a risk of missed tests.

[0003] The test monitoring scheme in the prior art is shown in the attached Figure 2 As shown, during the automatic monitoring test of the server test environment, there are cases where the temperature box or equipment fluctuations cause the relevant functions to fail to be detected in time. For example, in tests with extremely fast instantaneous changes such as mechanical shock, some functions may only appear during the change process. If the test is performed after the test, it is very likely that the functional abnormality has been restored; there have also been emergencies such as abnormal equipment reset or laboratory power outages, related services have stopped, and the services have not been loaded in time, resulting in the test being done in vain; there have also been cases where relevant function log records have not been collected in time before the equipment is abnormally reset during the test. At present, the operation of the equipment is monitored through the computer, and the functions are checked before, during and after the test. If there are enough personnel, personnel can be arranged to take turns on duty 24 hours a day to detect the equipment status, or the test equipment can be monitored through laboratory cameras. The test risk is minimized by increasing manpower and material resources. The monitoring test efficiency is extremely low, and there are missed tests.

[0004] Therefore, how to improve the existing automatic monitoring method of the server test environment to improve the monitoring efficiency of the server test environment and reduce the missed detection rate is a technical problem that urgently needs to be solved. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for automatically monitoring a server test environment, so as to improve the existing method for automatically monitoring a server test environment, so as to improve the efficiency of monitoring the server test environment and reduce the missed detection rate.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, a method for automatically monitoring a server test environment is provided, comprising the following steps:

[0008] S1: Acquire various data in the test environment of several servers in a specified period, mark the various data as normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set, and a validation set respectively;

[0009] S2: Create a server test environment monitoring model, input the training set, test set and validation set into the server test environment monitoring model in sequence, train, test and validate the server test environment monitoring model, and determine whether the server test environment monitoring model meets the preset requirements based on the validation results. If yes, execute step S3; if no, re-execute step S1 until the server test environment monitoring model meets the preset requirements.

[0010] S3: collecting various data in each monitored server test environment in real time, and inputting the data into the server test environment monitoring model, and outputting the monitoring results of each monitored server test environment in real time through the server test environment monitoring model;

[0011] S4: The server obtains the abnormal information according to the monitoring result, and determines whether the abnormal information needs to be processed. If so, step S5 is executed; if not, the server continues to automatically monitor the server test environment.

[0012] S5: Based on the abnormal information, the abnormal location and the severity of the abnormality in the monitoring test environment are obtained, and the abnormality solution is analyzed according to the abnormal location and abnormal data, and the abnormality is processed based on the obtained abnormality solution, and steps S3 to S5 are executed in a loop.

[0013] Preferably, the various data in the test environment of several servers obtained in step S1 include operating data of detection equipment, software operating data, log data, performance data of network equipment, network data, operating data of power supply equipment, and operating data of heat dissipation equipment. The detection equipment includes temperature and humidity detection equipment, video monitoring equipment, water leakage detection equipment, air conditioning detection equipment, and gas detection equipment.

[0014] Preferably, all data are first preprocessed before marking the data as normal data and abnormal data. The specific process of preprocessing is to delete duplicate data, noise data and unprocessable abnormal values, fill missing values, correct abnormal values, and then integrate the data from multiple data sources together to form a data set, and then perform data format conversion, data unification conversion, data normalization conversion and data discretization conversion on the original data to obtain data that is convenient for unified analysis, and finally normalize the data in the data set to a specified range.

[0015] Preferably, the server test environment monitoring model is a convolutional neural network model, and the specific process of training the server test environment monitoring model in step S2 is as follows:

[0016] S21: The data in the training set enters the model through the input layer of the server test environment monitoring model, and transmits the data to the hidden layer after weighted sum operation. The first layer of neurons in the hidden layer receives the weighted sum operation of the input layer plus the biased data, and transmits it to the next layer of neurons in the hidden layer after being processed by the activation function of the hidden layer, and finally transmits it to the output layer of the model, and the output layer outputs the processing result;

[0017] S22: The server test environment monitoring model compares the obtained processing result with the expected output processing result, determines the size of the error value between the two, and adjusts the weight and bias of the server test environment monitoring model according to the error value, and iteratively adjusts the weight and bias of each neuron in the model, step by step reducing the error value until the error value is within the preset error range.

[0018] Preferably, the process of adjusting the weights and biases of each layer of neurons in the server test environment monitoring model in step S22 is achieved by setting a loss function, and the delivered loss function is a mean square error loss function.

[0019] Preferably, during the training process of the server test environment monitoring model, the model is also optimized by the momentum stochastic gradient descent optimization algorithm. The specific process is: first, the momentum is updated by using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters. The momentum gradient descent method introduces a momentum variable v. At each iteration, the momentum is updated by using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters, wherein the iterative momentum variable v k+1 The formula is:

[0020]

[0021] x k+1 =x k +v k+1 ;

[0022] Where μ is the momentum factor, α k is the learning rate, is the gradient of the loss function with respect to the parameters, s k is a set of randomly selected samples.

[0023] In the second aspect, a server test environment automatic monitoring system is provided, which is used to implement any one of the server test environment automatic monitoring methods described above, including a data acquisition module, a data tag classification module, a model creation module, a server test environment monitoring model, a data collection module, and a control module; the data acquisition module, data tag classification module, model creation module, server test environment monitoring model, data collection module, and exception handling module are all connected to the control module.

[0024] The data acquisition module is used to acquire various data in the test environment of several servers in a specified period; the data labeling and classification module is used to mark the various data as normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set and a validation set respectively; the model creation module is used to create a server test environment monitoring model; the data collection module is used to collect various data in each monitored server test environment in real time; the server test environment monitoring model is used to receive the data in the training set, the test set and the validation set in sequence for training, testing and verification, and receive and collect various data in each monitored server test environment in real time, and output the monitoring results of each monitored server test environment in real time through the server test environment monitoring model; the control module is used to control the data acquisition module, the data labeling and classification module, the model creation module, the server test environment monitoring model and the data collection module; the exception handling module is used to handle exceptions in the server test environment.

[0025] The beneficial effects of the present invention include:

[0026] The server test environment automatic monitoring method and system provided by the present invention obtain various data in the test environment of several servers in a specified period and divide them into a training set, a test set and a verification set; and input the data into a server test environment monitoring model for training, testing and verification, collect various data in each monitored server test environment in real time, and input the data into the server test environment monitoring model, and output the monitoring results of each monitored server test environment in real time through the server test environment monitoring model. The server obtains abnormal information therein according to the monitoring result, and obtains the abnormal position and abnormal severity of the monitoring test environment based on the abnormal information according to whether the abnormal information needs to be processed, analyzes the abnormal solution according to the abnormal position and abnormal data, and processes the abnormality based on the obtained abnormal solution.

[0027] First, by obtaining various data in the server test environment of several servers within a specified period, the data includes normal data and abnormal data. Based on the normal data and abnormal data, the corresponding features of the normal data and abnormal data can be extracted through the server test environment monitoring model created later, so as to obtain the accuracy of the analysis and judgment of whether the subsequent real-time collected data is abnormal or not. Therefore, the various data in the server test environment of several servers within a specified period obtained in the early stage provide a data basis for the subsequent monitoring of the server test environment.

[0028] Secondly, by creating a server test environment monitoring model, and conducting detailed training tests and verification on it, and in the training process, the input and output data are weighted and calculated, and then transmitted to the hidden layer for weighted and calculated, and the biased data is output and processed; and according to the error value, the weight and bias of the server test environment monitoring model are adjusted, and the weight and bias of each neuron in the model are adjusted iteratively, and the error value is reduced step by step until the error value is within the preset error range. The entire training process is to adjust the network parameters through continuous iteration of forward propagation and backward propagation, so that the network can better fit the training data, and at the same time improve the prediction ability of unknown data. The entire training process is to adjust the network parameters through continuous iteration of forward propagation and backward propagation, so that the network can better fit the training data, and at the same time improve the prediction ability of unknown data.

[0029] Thirdly, the model is optimized by using the momentum stochastic gradient descent optimization algorithm, the momentum is updated using the gradient at the current point, and then the parameters are updated using the momentum as the increment of one-step iteration, so that the algorithm can more effectively use the historical gradient information and accelerate the convergence process when dealing with complex optimization problems. Especially when dealing with high-dimensional and non-convex optimization problems, the momentum gradient descent method shows good performance. It effectively improves the monitoring efficiency of the server test environment and reduces the missed detection rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The figure is a flow chart of the method for automatically monitoring the server test environment of the present invention.

[0031] Figure 2 The schematic diagram is a schematic diagram of the architecture of an automatic monitoring system for a server test environment in the prior art.

[0032] Figure 3 It is a schematic diagram of the architecture of the server test environment automatic monitoring system of the present invention. DETAILED DESCRIPTION

[0033] The following is combined with Figure 1 to Figure 3 The present invention is further described in detail:

[0034] Example 1

[0035] See attached Figure 1 As shown, a method for automatically monitoring a server test environment comprises the following steps:

[0036] S1: Obtain various data in the test environment of several servers in a specified period, mark the data as normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set, and a validation set, respectively. In this step, various data in the test environment of several servers in a specified period are obtained, and the data include normal data and abnormal data. Based on the normal data and abnormal data, the corresponding features of the normal data and abnormal data can be extracted through the server test environment monitoring model created later, so as to obtain the accuracy of the analysis and judgment of whether the data collected in real time later is abnormal or not. Therefore, the various data in the test environment of several servers in the specified period obtained in the early stage provide a data basis for the subsequent monitoring of the server test environment.

[0037] The data obtained in the test environment of several servers include the operation data of detection equipment, software operation data, log data, performance data of network equipment, network data, operation data of power supply equipment, and operation data of heat dissipation equipment. The detection equipment includes temperature and humidity detection equipment, video monitoring equipment, water leakage detection equipment, air conditioning detection equipment, and gas detection equipment.

[0038] S2: Create a server test environment monitoring model, input the training set, test set and verification set into the server test environment monitoring model in sequence, train, test and verify the server test environment monitoring model, and determine whether the server test environment monitoring model meets the preset requirements based on the verification results. If so, execute step S3; if not, re-execute step S1 until the server test environment monitoring model meets the preset requirements.

[0039] By creating a server test environment monitoring model, and conducting detailed training tests and verification on it, and in the training process, the input and output data are weighted and calculated, and then transmitted to the hidden layer for weighted and calculated, and the biased data is output and processed; and according to the error value, the weight and bias of the server test environment monitoring model are adjusted, and the weight and bias of each neuron in the model are adjusted iteratively, and the error value is reduced step by step until the error value is within the preset error range. The entire training process is to adjust the network parameters through continuous iteration of forward propagation and backward propagation, so that the network can better fit the training data, and at the same time improve the prediction ability of unknown data. The entire training process is to adjust the network parameters through continuous iteration of forward propagation and backward propagation, so that the network can better fit the training data, and at the same time improve the prediction ability of unknown data.

[0040] S3: collecting various data in each monitored server test environment in real time, and inputting the data into the server test environment monitoring model, and outputting the monitoring results of each monitored server test environment in real time through the server test environment monitoring model;

[0041] S4: The server obtains the abnormal information according to the monitoring result, and determines whether the abnormal information needs to be processed. If so, step S5 is executed; if not, the server continues to automatically monitor the server test environment.

[0042] S5: Based on the abnormal information, the abnormal location and the severity of the abnormality in the monitoring test environment are obtained, and the abnormality solution is analyzed according to the abnormal location and abnormal data, and the abnormality is processed based on the obtained abnormality solution, and steps S3 to S5 are executed in a loop.

[0043] Example 2

[0044] On the basis of Example 1, all data are first preprocessed before each data is marked as normal data and abnormal data. The specific process of preprocessing is to delete duplicate data, noise data and unprocessable abnormal values, fill missing values, correct abnormal values, and then integrate the data from multiple data sources together to form a data set, and then perform data format conversion, data unification conversion, data normalization conversion and data discretization conversion on the original data to obtain data that is convenient for unified analysis, and finally normalize the data in the data set to a specified range.

[0045] The server test environment monitoring model is a convolutional neural network model. The specific process of training the server test environment monitoring model in step S2 is as follows:

[0046] S21: The data in the training set enters the model through the input layer of the server test environment monitoring model, and transmits the data to the hidden layer after weighted sum operation. The first layer of neurons in the hidden layer receives the weighted sum operation of the input layer plus the biased data, and transmits it to the next layer of neurons in the hidden layer after being processed by the activation function of the hidden layer, and finally transmits it to the output layer of the model, and the output layer outputs the processing result;

[0047] S22: The server test environment monitoring model compares the obtained processing result with the expected output processing result, determines the size of the error value between the two, and adjusts the weight and bias of the server test environment monitoring model according to the error value, and iteratively adjusts the weight and bias of each neuron in the model, step by step reducing the error value until the error value is within the preset error range.

[0048] In this embodiment, the process of adjusting the weights and biases of each layer of neurons in the server test environment monitoring model in step S22 is achieved by setting a loss function, and the transmitted loss function is a mean square error loss function.

[0049] Example 3

[0050] On the basis of Example 1 or Example 2, during the training process of the server test environment monitoring model, the model is also optimized by the momentum stochastic gradient descent optimization algorithm. The specific process is: first, the momentum is updated by using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters. The momentum gradient descent method introduces a momentum variable v. At each iteration, the momentum is updated by using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters, wherein the iterative momentum variable v k+1 The formula is:

[0051]

[0052] x k+1 =x k +v k+1 ;

[0053] Where μ is the momentum factor, α k is the learning rate, is the gradient of the loss function with respect to the parameters, s k is a set of randomly selected samples.

[0054] The model is optimized through the momentum stochastic gradient descent optimization algorithm. The momentum is updated using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters. This allows the algorithm to more effectively use historical gradient information and accelerate the convergence process when dealing with complex optimization problems. In particular, the momentum gradient descent method shows good performance when dealing with high-dimensional and non-convex optimization problems. This further improves the monitoring efficiency of the server test environment and reduces the missed detection rate.

[0055] See also Figure 3 As shown, a server test environment automatic monitoring system is used to implement any one of the server test environment automatic monitoring methods described, including a data acquisition module, a data tag classification module, a model creation module, a server test environment monitoring model, a data collection module, and a control module; the data acquisition module, data tag classification module, model creation module, server test environment monitoring model, data collection module, and exception handling module are all connected to the control module.

[0056] The data acquisition module is used to acquire various data in the test environment of several servers in a specified period; the data labeling and classification module is used to mark the various data as normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set and a validation set respectively; the model creation module is used to create a server test environment monitoring model; the data collection module is used to collect various data in each monitored server test environment in real time; the server test environment monitoring model is used to receive the data in the training set, the test set and the validation set in sequence for training, testing and verification, and receive and collect various data in each monitored server test environment in real time, and output the monitoring results of each monitored server test environment in real time through the server test environment monitoring model; the control module is used to control the data acquisition module, the data labeling and classification module, the model creation module, the server test environment monitoring model and the data collection module; the exception handling module is used to handle exceptions in the server test environment.

[0057] In summary, the server test environment automatic monitoring method and system provided by the present invention obtains various data in the test environment of the server within the specified period of several servers, and the data includes normal data and abnormal data. Based on the normal data and abnormal data, the corresponding features of the normal data and abnormal data can be extracted through the server test environment monitoring model created later, so as to obtain the analysis and judgment accuracy of the abnormality of the subsequent real-time collected data. Therefore, the various data in the test environment of the server within the specified period of several servers obtained in the early stage provide a data basis for the monitoring of the subsequent server test environment. By creating a server test environment monitoring model, and performing detailed training tests and verification on it, and in the training process, the input and output data are weighted and calculated and then transmitted to the hidden layer for weighted and calculated and then the data output processing result after the bias is added; and according to the error value, the weight and bias of the server test environment monitoring model are adjusted, and the weight and bias of each neuron in the model are adjusted iteratively, and the error value is reduced step by step until the error value is within the preset error range. The whole training process is to adjust the network parameters by continuously iterating forward propagation and reverse propagation, so that the network can better fit the training data, and at the same time improve the prediction ability of unknown data.

[0058] The entire training process continuously iterates forward propagation and back propagation to adjust network parameters, so that the network can better fit the training data and improve the prediction ability of unknown data. The model is optimized by the momentum stochastic gradient descent optimization algorithm, and the momentum is updated using the gradient at the current point. Then, the momentum is used as the increment of one-step iteration to update the parameters. This allows the algorithm to more effectively use historical gradient information and accelerate the convergence process when dealing with complex optimization problems. In particular, the momentum gradient descent method shows good performance when dealing with high-dimensional, non-convex optimization problems. It effectively improves the monitoring efficiency of the server test environment and reduces the missed detection rate.

Claims

1. A method for automatically monitoring a server test environment, characterized in that: The following steps are involved: S1: Acquire various data in the test environment of several servers in a specified period, mark the various data as normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set, and a validation set respectively; S2: Create a server test environment monitoring model, input the training set, test set and validation set into the server test environment monitoring model in sequence, train, test and validate the server test environment monitoring model, and determine whether the server test environment monitoring model meets the preset requirements based on the validation results. If yes, execute step S3; if no, re-execute step S1 until the server test environment monitoring model meets the preset requirements. S3: collecting various data in each monitored server test environment in real time, and inputting the data into the server test environment monitoring model, and outputting the monitoring results of each monitored server test environment in real time through the server test environment monitoring model; S4: The server obtains the abnormal information according to the monitoring result, and determines whether the abnormal information needs to be processed. If so, step S5 is executed; if not, the server continues to automatically monitor the server test environment. S5: Based on the abnormal information, the abnormal location and the severity of the abnormality in the monitoring test environment are obtained, and the abnormality solution is analyzed according to the abnormal location and abnormal data, and the abnormality is processed based on the obtained abnormality solution, and steps S3 to S5 are executed in a loop.

2. A method for automatically monitoring a server test environment according to claim 1, characterized in that: The various data in the test environment of several servers obtained in step S1 include the operating data of the detection equipment, software operating data, log data, performance data of the network equipment, network data, operating data of the power supply equipment, and operating data of the heat dissipation equipment. The detection equipment includes temperature and humidity detection equipment, video monitoring equipment, water leakage detection equipment, air conditioning detection equipment, and gas detection equipment.

3. A method for automatically monitoring a server test environment according to claim 2, characterized in that: Before marking the data as normal data and abnormal data, all data are first preprocessed. The specific process of preprocessing is to delete duplicate data, noise data and unprocessable abnormal values, fill missing values, correct abnormal values, and then integrate the data from multiple data sources together to form a data set. The original data is then converted into data format, unified data, normalized data and discretized data to obtain data that is easy to analyze uniformly. Finally, the data in the data set is normalized to a specified range.

4. The method for automatically monitoring a server test environment according to claim 1, characterized in that: The server test environment monitoring model is a convolutional neural network model. The specific process of training the server test environment monitoring model in step S2 is as follows: S21: The data in the training set enters the model through the input layer of the server test environment monitoring model, and transmits the data to the hidden layer after weighted sum operation. The first layer of neurons in the hidden layer receives the weighted sum operation of the input layer plus the biased data, and transmits it to the next layer of neurons in the hidden layer after being processed by the activation function of the hidden layer, and finally transmits it to the output layer of the model, and the output layer outputs the processing result; S22: The server test environment monitoring model compares the obtained processing result with the expected output processing result, determines the size of the error value between the two, and adjusts the weight and bias of the server test environment monitoring model according to the error value, and iteratively adjusts the weight and bias of each neuron in the model, step by step reducing the error value until the error value is within the preset error range.

5. A method for automatically monitoring a server test environment according to claim 4, characterized in that: The process of adjusting the weights and biases of each layer of neurons in the server test environment monitoring model in step S22 is achieved by setting a loss function, and the delivered loss function is a mean square error loss function.

6. A method for automatically monitoring a server test environment according to claim 4, characterized in that: During the training process of the server test environment monitoring model, the model is also optimized by the momentum stochastic gradient descent optimization algorithm. The specific process is: first, the momentum is updated using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters. The momentum gradient descent method introduces a momentum variable v. At each iteration, the momentum is updated using the gradient at the current point, and then the momentum is used as the increment of one-step iteration to update the parameters. The iterative momentum variable v k+1 The formula is: x k+1 =x k +u k+1 ; Where μ is the momentum factor, α k is the learning rate, is the gradient of the loss function with respect to the parameters, s k is a set of randomly selected samples.

7. A server test environment automatic monitoring system, used to implement a server test environment automatic monitoring method according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a data tag division module, a model creation module, a server test environment monitoring model, a data acquisition module, and a control module; the data acquisition module, the data tag division module, the model creation module, the server test environment monitoring model, the data acquisition module, and the exception handling module are all connected to the control module; The data acquisition module is used to acquire various data in the test environment of several servers during a specified period; The data labeling and division module is used to label the data into normal data and abnormal data, and divide the normal data and abnormal data into a training set, a test set and a validation set respectively; The model creation module is used to create a server test environment monitoring model; The data acquisition module is used to collect various data in the test environment of each monitored server in real time; The server test environment monitoring model is used to sequentially receive the data in the training set, test set and validation set for training, testing and validation, and receive and collect various data in each monitored server test environment in real time, and output the monitoring results of each monitored server test environment in real time through the server test environment monitoring model; The control module is used to control the data acquisition module, the data tag division module, the model creation module, the server test environment monitoring model, and the data acquisition module; The exception handling module is used to handle exceptions in the server test environment.

Images