Multi-node server testing method and apparatus

Abstract

The application provides a kind of multi-node server test method and device, it is related to test and computing device technical field, wherein, the multi-node server test method, comprising: according to the application scene of the multi-node server to be tested, constructs test set;Each node in the multi-node server is tested based on the test set, and the running data of each node is recorded;In the case where there is an abnormal node in each node, the running data of the abnormal node is used for node anomaly detection, and a test result is obtained.The application not only can automatically analyze abnormal node, obtain fault reason, assist test personnel to find out problem, so as to solve the problem subsequently, but also can simulate actual application scene, and then improve test effect.

Description

Technical Field

[0001] This invention relates to the field of testing and computing equipment technology, and in particular to a multi-node server testing method and apparatus. Background Technology

[0002] With the rapid development of the economy and society, the multi-node server industry has risen rapidly. Massive amounts of data are impacting various industries at an unprecedented growth rate, placing higher demands on the maintainability of multi-node servers. Automated testing is required in the research and development, design, production, and use of multi-node servers to verify their actual working performance.

[0003] Currently, testing multi-node servers relies on limited testing content, which fails to effectively simulate real-world application scenarios, resulting in unsatisfactory testing outcomes. Therefore, an effective solution is urgently needed to address these issues. Summary of the Invention

[0004] To address the problems existing in related technologies, embodiments of the present invention provide a multi-node server testing method and apparatus.

[0005] This invention provides a multi-node server testing method, comprising:

[0006] Based on the application scenario of the multi-node server to be tested, construct a test suite;

[0007] The test set is used to test each node in the multi-node server, and the running data of each node is recorded.

[0008] If there are abnormal nodes among the nodes, node anomaly detection is performed based on the operational data of the abnormal nodes to obtain test results.

[0009] According to a multi-node server testing method provided by the present invention, the step of performing node anomaly detection based on the operational data of the abnormal node to obtain test results includes:

[0010] The operational data of the abnormal node is input into a pre-trained fault cause analysis model to detect node anomalies and obtain the fault cause analysis results of the abnormal node.

[0011] Test results are generated based on the identifier of the abnormal node, its abnormal state, and the analysis results of the cause of the failure.

[0012] According to a multi-node server testing method provided by the present invention, the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component location unit, a feature location unit, and a fault cause matching unit;

[0013] Accordingly, the step of inputting the operational data of the abnormal node into a pre-trained fault cause analysis model for node anomaly detection, and obtaining the fault cause analysis result of the abnormal node, includes:

[0014] The operational data of the abnormal node is input to the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; the operational data of the abnormal node is input to the feature extraction unit for feature extraction to obtain initial abnormal features, which characterize the fault phenomenon occurring in the abnormal node.

[0015] The fault category is input into the component positioning unit for positioning processing to identify the abnormal component;

[0016] The at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal feature are input to the feature localization unit for feature localization to obtain the target abnormal feature, wherein the target abnormal feature is the preset abnormal feature that matches the initial abnormal feature;

[0017] The target anomaly features are input into the fault cause matching unit for fault cause matching to obtain the fault cause analysis results of the anomaly node.

[0018] According to a multi-node server testing method provided by the present invention, before inputting the running data of the abnormal node into a pre-trained fault cause analysis model for node anomaly detection, the method further includes:

[0019] Collect historical operational data and causes of historical failures from historical abnormal nodes;

[0020] The historical operational data is input into the initial analysis model to detect node anomalies and obtain the predicted causes of failures.

[0021] Calculate the loss value based on the predicted causes of failure and the historical causes of failure;

[0022] Adjust the model parameters of the initial analysis model based on the loss value, and continue to execute the steps of collecting historical operating data and historical fault causes of historical abnormal nodes until the preset training stop condition is reached to obtain the trained fault cause analysis model.

[0023] According to a multi-node server testing method provided by the present invention, before constructing the test set based on the application scenario of the multi-node server to be tested, the method further includes:

[0024] Obtain the model number of the multi-node server to be tested;

[0025] Retrieve the application data corresponding to the model number;

[0026] Based on the application data, the multi-node server is analyzed to determine the working application scope corresponding to the multi-node server;

[0027] The application scope of the work is divided into scenarios to obtain the application scenarios of the multi-node server;

[0028] Accordingly, the step of constructing a test set based on the application scenario of the multi-node server to be tested includes:

[0029] Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified;

[0030] Collect test cases corresponding to each of the tasks to be tested;

[0031] Construct a test suite based on each of the described test cases;

[0032] Accordingly, the testing of each node in the multi-node server based on the test set includes:

[0033] Based on the application scenario, determine the execution order of each of the tasks to be tested;

[0034] According to the execution order, each test case in the test set is loaded to test each node in the multi-node server.

[0035] According to a multi-node server testing method provided by the present invention, before obtaining the test result by performing node anomaly detection based on the operational data of the abnormal nodes when abnormal nodes exist among the nodes, the method further includes:

[0036] Obtain the test scripts generated during the testing process;

[0037] The return values ​​in the test script are parsed to determine whether there are any abnormal nodes among the nodes.

[0038] The multi-node server testing method provided by the present invention further includes:

[0039] During the test, the node status of each node is displayed in real time, including normal status and abnormal status.

[0040] According to a multi-node server testing method provided by the present invention, after recording the operating data of each node, the method further includes:

[0041] Test logs are generated based on the operational data of each node.

[0042] The present invention also provides a multi-node server testing device, comprising:

[0043] The build module is configured to build a test suite based on the application scenario of the multi-node server to be tested;

[0044] The testing module is configured to test each node in the multi-node server based on the test set and record the running data of each node.

[0045] The detection module is configured to perform node anomaly detection based on the operational data of the abnormal nodes when there are abnormal nodes among the nodes, and obtain test results.

[0046] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-node server testing method described above.

[0047] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-node server testing method as described above.

[0048] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the multi-node server testing method as described above.

[0049] The multi-node server testing method and apparatus provided by this invention constructs a test set based on the application scenario of the multi-node server to be tested. This not only improves the correlation between the test set and the application scenario, thus ensuring that the actual application scenario can be effectively simulated during testing, thereby improving the testing effect and reliability of the multi-node server; but also tests each node in the multi-node server through the test set, records the operating data of each node, and performs node anomaly detection based on the operating data of the abnormal nodes when abnormal nodes are present, obtaining test results. It can automatically analyze the abnormal nodes, determine the cause of the failure, and assist testers in discovering problems so that they can be resolved subsequently. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in this invention or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is one of the flowcharts of the multi-node server testing method provided by the present invention;

[0052] Figure 2This is a schematic diagram of the node anomaly detection process provided by the present invention;

[0053] Figure 3 This is a flowchart illustrating the display of node status provided by the present invention;

[0054] Figure 4 This is the second flowchart of the multi-node server testing method provided by the present invention;

[0055] Figure 5 This is a schematic diagram of the structure of the multi-node server testing device provided by the present invention;

[0056] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0058] To facilitate a clearer understanding of the various embodiments of the present invention, some relevant background information is first introduced as follows.

[0059] With the rapid development of the economy and society, the server industry has risen rapidly. Massive amounts of data are impacting various industries at an unprecedented growth rate, placing higher demands on the maintainability of servers. Servers require automated testing during research and development, design, production, and use to verify their actual working performance.

[0060] A server is a type of computer that runs faster, handles higher loads, and is more expensive than a regular computer. Servers provide computing or application services to other client machines on a network, such as personal computers (PCs), smartphones, automatic teller machines (ATMs), and even large equipment like train systems. Servers feature high-speed core processors (CPUs) with powerful processing capabilities, long-term reliable operation, robust input / output (I / O) external data throughput, and better scalability.

[0061] Servers possess characteristics such as scalability, ease of use, and availability. The following is a brief explanation of server scalability, ease of use, and availability.

[0062] Servers must possess a certain degree of scalability because enterprise networks cannot remain unchanged indefinitely, especially in today's information age. If a server lacks scalability, it will become unmanageable as the number of users increases, rendering a server worth tens or even hundreds of thousands of dollars obsolete in a short period—a cost no enterprise can afford. To maintain server scalability, servers typically have expansion capabilities and redundant components, such as disk array bays, peripheral component interconnects (PCI), and memory slots. Scalability is specifically reflected in the following aspects: whether the hard drive can be expanded, whether the CPU can be upgraded or expanded, and whether the system supports multiple mainstream operating systems such as Windows NT, Linux, or UNIX. Only in this way can the initial investment in the server be fully utilized in the long run.

[0063] Servers are far more complex than PCs, not only in terms of hardware configuration but also, and perhaps more importantly, software system configuration. It's unimaginable for a server to perform so many functions without comprehensive software support. However, too much software can lead to decreased server performance and make it difficult for administrators to operate effectively. Therefore, many server manufacturers, when designing servers, not only fully consider availability and stability but also invest heavily in ease of use. Server ease of use is mainly reflected in the following aspects: how easy is the server to operate? Is the user navigation system comprehensive? Is the chassis design user-friendly? Does it have critical recovery functions? Does it have operating system backups? And is there sufficient training support?

[0064] For servers, a very important aspect is their availability, that is, the selected server can meet the requirements of long-term stable operation and cannot frequently experience problems.

[0065] With the rapid development of servers, high-end server models have emerged, such as multi-node servers. A multi-node server contains multiple nodes, each of which is an independent sub-server providing services. Currently, the main testing items for multi-node servers include server health function checks, CPU performance tests, memory performance tests, hard drive performance tests, and network performance tests. This includes monitoring whether the temperature of each input / output port is within the required range, and whether the input / output voltages of each key component are within the standard range. If any test value is abnormal, an error will occur, indicating the corresponding error node and error code. For example, an abnormal voltage value for a component in a node, a discrepancy between the required memory quantity and the actual memory quantity in a node, a malfunction in a device within a node, or an error in a high-speed serial computer expansion bus standard (PCIE, Peripheral Component Interconnect Express) link will all trigger an error message with the corresponding error node and error code.

[0066] Currently, testing multi-node servers often involves limited testing content, failing to effectively simulate real-world application scenarios and resulting in unsatisfactory testing outcomes. Furthermore, current technologies typically involve identifying initial fault locations and outputting error messages when test scripts encounter errors. However, this method suffers from several shortcomings: it cannot accurately pinpoint the cause of the error, such as identifying which specific component or value caused the error; some error messages are poorly analyzed, requiring extensive verification by relevant personnel, such as engineers and / or production staff, to gradually pinpoint the root cause, consuming significant time; and it cannot distinguish between software and hardware issues, necessitating time-consuming and labor-intensive investigation by relevant personnel.

[0067] It is evident that while the relevant technologies can initially locate the errors reported by the test scripts and extract and output the error messages, they cannot further analyze the root cause of the errors. For a given error, it is impossible to accurately determine whether it is a hardware or software problem. It is also impossible to determine whether an abnormal test value for a component is caused by a problem with a connected component or an indirectly connected component. For some test items, the test scripts are provided by the client or a third party. Due to special circumstances, such as confidentiality requirements, the error parsing is unclear. Based solely on the error message, relevant personnel cannot locate the root cause of the error, requiring extensive and repetitive communication and verification, wasting a significant amount of time and energy, and posing a substantial challenge to production and maintenance work.

[0068] Therefore, this invention provides a multi-node server testing method and apparatus to solve some of the above-mentioned problems. Specifically, by constructing a test set according to the application scenario of the multi-node server to be tested, the correlation between the test set and the application scenario is improved, thereby ensuring that the actual application scenario can be effectively simulated during testing, improving the testing effect and reliability of the multi-node server; and by testing each node in the multi-node server through the test set and recording the operating data of each node, and in the case of abnormal nodes, node anomaly detection is performed based on the operating data of the abnormal nodes to obtain test results. The abnormal nodes can be automatically analyzed to determine the cause of the failure, assisting testers in discovering problems so that they can be resolved subsequently.

[0069] The following is combined Figures 1-5 This invention describes a multi-node server testing method and apparatus.

[0070] Figure 1 This is one of the flowcharts illustrating the multi-node server testing method provided by this invention. See [link / reference]. Figure 1 As shown, steps 101-103 are included, wherein:

[0071] Step 101: Construct a test suite based on the application scenario of the multi-node server to be tested.

[0072] First, it should be noted that the subject of this invention can be any electronic device used to test a multi-node server, such as a smartphone, smartwatch, desktop computer, laptop, or any other type.

[0073] Specifically, application scenarios refer to the types of services provided by multi-node servers, such as gaming, artificial intelligence (AI), and conferencing services. A test suite is used to evaluate the performance of a multi-node server and is a collection of test cases used to test the server. Each test case describes a test task for a specific function or node, reflecting the test plan, methods, techniques, and strategies.

[0074] Since different multi-node servers serve different content, i.e., different application scenarios, the test sets used when testing different multi-node servers should also be different. To improve testing effectiveness, a corresponding test set can be obtained based on the application scenario of the multi-node server.

[0075] In practical applications, first determine the multi-node server to be tested, and then obtain the application scenario of the multi-node server.

[0076] For example, the service content of a multi-node server can be parsed to obtain the application scenarios of the multi-node server.

[0077] For example, scenario analysis can be performed on at least one of the web pages, software, and applications associated with the multi-node server to obtain the application scenario of the multi-node server.

[0078] Furthermore, after obtaining the application scenario, a test set is built for the multi-node server based on the application scenario.

[0079] For example, a test library is pre-built, containing pre-built test sets for different application scenarios. After obtaining the application scenario of the multi-node server to be tested, the corresponding test set can be found in the test library according to the application scenario.

[0080] For example, in a multi-node server application scenario, test cases associated with that application scenario can be found, and then a test set can be formed.

[0081] Step 102: Test each node in the multi-node server based on the test set, and record the running data of each node.

[0082] Specifically, runtime data refers to the data generated by a node when executing the tasks indicated by the test set, such as start execution time, execution end time, execution duration, storage space occupied by the execution, and execution parameters.

[0083] In practical applications, after obtaining the test set, the multi-node server is tested: the test set is loaded, and each node in the multi-node server is tested using the test set. The running data of each node during the test is recorded.

[0084] Step 103: If there are abnormal nodes among the nodes, perform node anomaly detection based on the running data of the abnormal nodes to obtain test results.

[0085] Specifically, an abnormal node refers to a node that reports an error. Node anomaly detection refers to detecting the cause or fault of an abnormal node's error.

[0086] In practical applications, if an error occurs during the test, it indicates the presence of an abnormal node in the multi-node server. For each abnormal node, node anomaly detection is performed based on the corresponding operational data to determine the cause of the fault or anomaly. In this case, the test results can include the abnormal node and its corresponding fault or anomaly; alternatively, the test results can include the abnormal node and its corresponding fault or anomaly, as well as the normal nodes.

[0087] It should be noted that if no errors are reported during the test, it means that all nodes are normal, and the test result indicates that all nodes are normal.

[0088] The multi-node server testing method provided by this invention constructs a test set based on the application scenario of the multi-node server under test. This not only improves the correlation between the test set and the application scenario, thus ensuring that the test can effectively simulate the actual application scenario, improving the testing effect and reliability of the multi-node server; but also tests each node in the multi-node server through the test set, records the running data of each node, and performs node anomaly detection based on the running data of the abnormal nodes when abnormal nodes are present, obtaining test results. It can automatically analyze abnormal nodes, determine the cause of failure, and assist testers in discovering problems for subsequent problem-solving.

[0089] In one or more optional embodiments of the present invention, the step of performing node anomaly detection based on the operational data of the abnormal node to obtain test results can be implemented as follows:

[0090] The operational data of the abnormal node is input into a pre-trained fault cause analysis model to detect node anomalies and obtain the fault cause analysis results of the abnormal node.

[0091] Test results are generated based on the identifier of the abnormal node, its abnormal state, and the fault cause analysis results. The fault cause analysis results refer to the root causes and / or faults that caused the node to malfunction.

[0092] Specifically, the fault cause analysis model is used to analyze the causes of faults and / or the root causes of errors.

[0093] In practical applications, see Figure 2 , Figure 2 This is a flowchart illustrating the node anomaly detection process provided by the present invention: The runtime data of the abnormal node is input into a pre-trained fault cause analysis model for node anomaly detection, resulting in the fault cause analysis results of the abnormal node, including steps 201 and 202:

[0094] Step 201: Input the operational data of the abnormal node into the fault cause analysis model to detect node anomalies;

[0095] Step 202: Output the fault cause analysis results of the abnormal nodes from the fault cause analysis model.

[0096] Therefore, using pre-trained fault cause analysis results for node anomaly detection can not only improve the efficiency of node anomaly detection and save a lot of manpower and resources, but also avoid the problem of inaccurate results caused by subjective assumptions in manual node anomaly detection, thus improving the accuracy of node anomaly detection.

[0097] In one or more optional embodiments of the present invention, before inputting the operational data of the abnormal node into a pre-trained fault cause analysis model for node anomaly detection and obtaining the fault cause analysis result of the abnormal node, the method further includes:

[0098] Extract the target keywords from the runtime data of the abnormal node;

[0099] Match the target keyword with each preset keyword;

[0100] Accordingly, the step of inputting the operational data of the abnormal node into a pre-trained fault cause analysis model for node anomaly detection, and obtaining the fault cause analysis result of the abnormal node, includes:

[0101] In the event of a matching failure, the runtime data of the abnormal node is input into a pre-trained fault cause analysis model for node anomaly detection, thereby obtaining the fault cause analysis result of the abnormal node.

[0102] In one or more optional embodiments of the present invention, after matching the target keyword with each preset keyword, the method further includes:

[0103] If a match is successful, the fault cause analysis result corresponding to the target key is retrieved from the preset fault cause analysis result library and used as the fault cause analysis result of the abnormal node.

[0104] In this way, by matching the target keyword with each preset keyword, if a match is successful, the preset fault cause analysis result corresponding to the successfully matched preset keyword or target keyword can be retrieved from the preset fault cause analysis result database, where preset fault cause analysis results are stored in association with keywords. If the match fails, the fault cause analysis model is used to determine the fault cause analysis result. This can improve the efficiency of determining the fault cause analysis result.

[0105] In one or more optional embodiments of the present invention, the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component localization unit, a feature localization unit, and a fault cause matching unit; correspondingly, the step of inputting the operational data of the abnormal node into the pre-trained fault cause analysis model for node anomaly detection to obtain the fault cause analysis result of the abnormal node can be implemented as follows:

[0106] The operational data of the abnormal node is input to the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; the operational data of the abnormal node is input to the feature extraction unit for feature extraction to obtain initial abnormal features, which characterize the fault phenomenon occurring in the abnormal node.

[0107] The fault category is input into the component positioning unit for positioning processing to identify the abnormal component;

[0108] The at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal feature are input to the feature localization unit for feature localization to obtain the target abnormal feature, wherein the target abnormal feature is the preset abnormal feature that matches the initial abnormal feature;

[0109] The target anomaly features are input into the fault cause matching unit for fault cause matching to obtain the fault cause analysis results of the anomaly node.

[0110] Specifically, the category parsing unit refers to the unit in the fault cause analysis model that predicts the fault type. The feature extraction unit refers to the unit in the fault cause analysis model that identifies abnormal phenomena. The component location unit refers to the unit in the fault cause analysis model that identifies abnormal components. Preset abnormal features refer to the features corresponding to possible abnormal phenomena that are pre-set for different components. The feature location unit refers to the unit in the fault cause analysis model that identifies target abnormal features. The fault cause matching unit refers to the unit in the fault cause analysis model that identifies the target fault cause.

[0111] In practical applications, the operational data of abnormal nodes can be input into the category resolution unit. This unit performs category resolution on the operational data, i.e., fault type prediction. For example, it identifies whether the fault is software or hardware; within the software, the fault may lie in the firmware (FW) version, system logs, component logs, or other components; or within the hardware, the fault may lie in the central processing unit (CPU), memory, hard drive, or other components. This determines the fault category corresponding to the abnormal node. Fault categories include software fault categories and hardware fault categories. Software fault categories include FW version fault categories, system log categories, and component log fault categories, while hardware fault categories include CPU fault categories, memory fault categories, and hard drive fault categories, etc.

[0112] The operational data of abnormal nodes can be input into the feature extraction unit, which analyzes the operational data of the identified abnormal nodes to determine the data that guarantees the abnormal phenomenon, thereby obtaining the initial abnormal features.

[0113] The acquired fault category can be input into the component location unit, which will then locate the faulty component, i.e., the abnormal component, based on the fault category. For example, if the fault category is the hard disk fault category in the hardware fault category, then the hardware component corresponding to the hard disk will be identified as the abnormal component. Similarly, if the fault category is the hard disk fault category in the software fault category, then the software component (such as the hard disk program) corresponding to the hard disk will be identified as the abnormal component.

[0114] Furthermore, based on the identification of the abnormal component and initial abnormal features, the abnormal component and initial abnormal features are input to the feature localization unit. The feature localization unit acquires at least one preset abnormal feature corresponding to the abnormal component and performs feature matching (i.e., feature localization) between the initial abnormal feature and the at least one preset abnormal feature corresponding to the abnormal component. The successfully matched preset abnormal feature is then used as the target abnormal feature. Alternatively, based on the identification of the abnormal component and initial abnormal features, at least one preset abnormal feature corresponding to the abnormal component is acquired. The acquired at least one preset abnormal feature and the initial abnormal feature are input to the feature localization unit. The feature localization unit performs feature matching between the initial abnormal feature and the at least one preset abnormal feature corresponding to the abnormal component, and the successfully matched preset abnormal feature is used as the target abnormal feature.

[0115] It should be noted that obtaining at least one preset abnormal feature corresponding to the abnormal component can be achieved by: the fault cause analysis model obtaining it from the feature library associated with or built into the fault cause analysis model, such as matching the abnormal component with each preset abnormal feature in the feature library to determine at least one preset abnormal feature corresponding to the abnormal component; or by extracting preset abnormal features marked with an abnormal identifier from the feature library, where the abnormal identifier is the identifier of the abnormal component, such as name, ID, IP, etc.

[0116] In addition, successful feature matching can be achieved by: the feature similarity between the preset abnormal feature and the initial abnormal feature being greater than a set threshold, or by other feasible methods.

[0117] Furthermore, the target anomaly features are input into the fault cause matching unit, which then obtains the target fault cause matched by the target anomaly features from the fault causes associated with or inherent to the fault cause analysis model. All target fault causes constitute the fault cause analysis result of the anomaly node.

[0118] Thus, by inheriting the operational data of abnormal nodes through the category parsing unit, feature extraction unit, component location unit, feature location unit, and fault cause matching unit in the fault cause analysis results, the reliability of node anomaly detection and the accuracy of fault cause analysis results can be greatly improved.

[0119] It should be noted that the operational data of the abnormal node can be input into the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node, and then the operational data of the abnormal node can be input into the feature extraction unit for feature extraction to obtain the initial abnormal features; alternatively, the operational data of the abnormal node can be input into the feature extraction unit for feature extraction to obtain the initial abnormal features, and then the operational data of the abnormal node can be input into the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; alternatively, the operational data of the abnormal node can be input into the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node, and the operational data of the abnormal node can be input into the feature extraction unit for feature extraction to obtain the initial abnormal features. This invention does not impose any limitations on this, as long as the initial abnormal features are obtained before inputting at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal features into the feature localization unit for feature localization, and the fault category is obtained before inputting the fault category into the component localization unit for localization processing.

[0120] In one or more optional embodiments of the present invention, before inputting the running data of the abnormal node into the pre-trained fault cause analysis model for node anomaly detection, the method further includes: training the initial analysis model based on historical anomaly data and historical fault causes to obtain the fault cause analysis model.

[0121] Specifically, historical anomaly data refers to the historical operational data corresponding to historical anomaly nodes. Historical failure causes refer to the actual root causes of failures at historical anomaly nodes. The initial analysis model refers to a model with fault analysis capabilities that has not yet been trained.

[0122] In practical applications, historical anomaly data and historical failure causes collected from multi-node servers can be used as a sample set to train the initial analysis model, thereby improving the robustness, efficiency, and accuracy of the failure cause analysis model.

[0123] Optionally, the step of training the initial analysis model based on historical anomaly data and historical fault causes to obtain a fault cause analysis model can be implemented as follows:

[0124] Collect historical operational data and causes of historical failures from historical abnormal nodes;

[0125] The historical operational data is input into the initial analysis model to detect node anomalies and obtain the predicted causes of failures.

[0126] Calculate the loss value based on the predicted causes of failure and the historical causes of failure;

[0127] Adjust the model parameters of the initial analysis model based on the loss value, and continue to execute the steps of collecting historical operating data and historical fault causes of historical abnormal nodes until the preset training stop condition is reached to obtain the trained fault cause analysis model.

[0128] Specifically, historical abnormal nodes refer to nodes that have previously experienced anomalies or errors in multi-node services. Preset training stopping conditions can be: the loss value being less than a loss threshold, the number of iterations reaching a threshold, the rate of change of the loss value being less than a rate of change threshold, or a combination of these. This invention does not impose any limitations on these conditions.

[0129] In practical applications, multiple errors occurring on the multi-node server under test and / or other multi-node servers can be collected. The historical abnormal node corresponding to each error can be obtained, and then the historical operational data and historical failure causes of that historical abnormal node can be acquired. Each set of historical operational data and historical failure causes is treated as a sample, forming a sample set. This sample set can include historical operational data and historical failure causes corresponding to the same historical abnormal node for different errors, and can also include historical operational data and historical failure causes corresponding to different historical abnormal nodes.

[0130] Furthermore, any sample can be obtained from the sample set, and its historical operational data can be input into the untrained initial analysis model for node anomaly detection to obtain the predicted cause of failure. This predicted cause of failure is then compared with the historical cause of failure in the sample according to a preset loss calculation strategy to obtain the loss value. The parameters of the initial analysis model are then optimized based on the loss value, such as by adjusting the model parameters in reverse. This process continues, with any sample obtained from the sample set for parameter optimization of the initial analysis model, until a preset training stopping condition is met. The trained initial analysis model is then designated as the failure cause analysis model.

[0131] Therefore, training the initial analysis model with historical data (historical operation data and historical causes of failure) helps improve the learning efficiency of the failure cause analysis model in detecting abnormal nodes, thereby improving the robustness, analysis efficiency and accuracy of the failure cause analysis model.

[0132] When the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component localization unit, a feature localization unit, and a fault cause matching unit, the step of inputting the historical operating data into the initial analysis model for node anomaly detection to obtain the predicted fault cause includes: inputting the historical operating data into the category parsing unit for category parsing to obtain the predicted fault category corresponding to the abnormal node; inputting the historical operating data into the feature extraction unit for feature extraction to obtain initial predicted abnormal features; inputting the predicted fault category into the component localization unit for localization processing to determine the predicted abnormal component; inputting at least one preset abnormal feature corresponding to the predicted abnormal component and the initial predicted abnormal feature into the feature localization unit for feature localization to obtain the target predicted abnormal feature; and inputting the target predicted abnormal feature into the fault cause matching unit for fault cause matching to obtain the predicted fault cause. This improves the robustness of model training.

[0133] In one or more optional embodiments of the present invention, before constructing the test set according to the application scenario of the multi-node server to be tested, the method further includes obtaining the application scenario of the multi-node server to be tested. The specific implementation process can be as follows:

[0134] Obtain application data from the multi-node server to be tested;

[0135] The application scenario of the multi-node server is constructed based on the application data.

[0136] Specifically, application data refers to the configuration data for services provided by multi-node servers, such as music configuration data, game performance configuration data, shooting configuration data, and so on.

[0137] In practical applications, application data from multi-node servers can be acquired first, then analyzed to determine the various service functions of the multi-node servers. Based on these functions, the application scenarios of the multi-node servers can be determined. Thus, constructing application scenarios for multi-node servers using their application data improves the efficiency and accuracy of scenario building, thereby increasing the accuracy of testing.

[0138] In one or more optional embodiments of the present invention, the specific implementation process of obtaining the application data of the multi-node server to be tested can be as follows:

[0139] Obtain the model number of the multi-node server to be tested;

[0140] Retrieve application data corresponding to the model number.

[0141] Specifically, the model number refers to a series of indicators that reflect the nature, performance, quality, etc. of a multi-node server, and is generally composed of a set of letters and numbers numbered according to a certain pattern.

[0142] In practical applications, model numbers are typically labeled on the surface of objects or in prominent locations, making it easy to obtain the model number of the multi-node server under test. Then, searching by multi-node server model number allows for quick and accurate retrieval of application data. This method is simple and easy to implement, simplifies the data processing workflow, and ensures the comprehensiveness, accuracy, and speed of application data retrieval.

[0143] In one or more optional embodiments of the present invention, obtaining the application data of the multi-node server to be tested may be: parsing the multi-node server to obtain the application data of the multi-node server.

[0144] In one or more optional embodiments of the present invention, the specific implementation process of constructing the multi-node server application scenario based on the application data can be as follows:

[0145] Based on the application data, the multi-node server is analyzed to determine the working application scope corresponding to the multi-node server;

[0146] The application scope of the work is divided into scenarios to obtain the application scenarios of the multi-node server.

[0147] Specifically, the scope of application refers to the working environment of a multi-node server.

[0148] In practical applications, application data from multi-node servers can be analyzed to determine the corresponding working application scope. This working application scope can then be divided into scenarios to identify the application scenarios for the multi-node servers. For example, the working application scope can be matched with various application scenarios in a scenario library; the successfully matched scenarios are then identified as the application scenarios for the multi-node servers. In this way, determining the application scenarios for multi-node servers based on their working application scope improves the matching degree between application scenarios and the working application scope. This ensures that the testing environment for multi-node servers more closely resembles the real-world working environment of multi-node servers, thereby improving testing accuracy.

[0149] In one or more optional embodiments of the present invention, the specific implementation process of constructing a test set according to the application scenario of the multi-node server to be tested can be as follows:

[0150] The application scenario of the multi-node server to be tested is analyzed to determine at least one test case corresponding to the application scenario.

[0151] Construct a test suite based on each of the test cases described.

[0152] In practical applications, multi-node server application scenarios can be analyzed to determine at least one test case associated with that application scenario. The obtained test cases constitute a test set.

[0153] In this way, by constructing test sets using test cases corresponding to application scenarios, we can not only ensure the completeness of the test sets and their compatibility with multi-node servers, but also improve the correlation between the test sets and application scenarios. This will ensure that the actual application scenarios can be effectively simulated during testing, thereby improving the testing effect and reliability of multi-node servers.

[0154] Optionally, a test case library is pre-built, containing multiple test cases, each labeled with a corresponding application scenario. After obtaining the application scenario of the multi-node server to be tested, the corresponding test case can be found in the test case library based on that scenario.

[0155] Optionally, the application scenario of the multi-node server to be tested is analyzed to determine at least one test case corresponding to the application scenario. The specific implementation process can be as follows:

[0156] Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified;

[0157] Collect the test cases corresponding to each of the tasks to be tested.

[0158] In practical applications, scenario analysis can be performed on multi-node server application scenarios to determine the various tasks to be tested that need to be executed in that application scenario. Then, for each task to be tested, corresponding test cases can be obtained, thus forming a test set.

[0159] In this way, obtaining test cases through the tasks to be tested corresponding to the application scenario can not only improve the efficiency of obtaining test cases, but also improve the comprehensiveness of test cases. When using the test set composed of test cases for testing, multi-node servers can be tested in all aspects, thus improving the testing effect.

[0160] In one or more optional embodiments of the present invention, test cases can be loaded in a certain order. That is, each node in the multi-node server is tested based on the test set. The specific implementation process can be as follows:

[0161] Based on the application scenario, determine the execution order of each of the tasks to be tested;

[0162] According to the execution order, each test case in the test set is loaded to test each node in the multi-node server.

[0163] Specifically, the execution order refers to the order in which the tasks to be tested are executed.

[0164] In practical applications, after obtaining the test set, the application scenario is analyzed to determine the execution order of each task to be tested. The execution priority of each test case for each task is also determined. The execution order of the tasks and the execution priority of each test case are combined to determine the loading order of the test cases in the test set. Then, the test cases in the test set are loaded according to the loading order to test each node in the multi-node server.

[0165] In this way, by determining the execution order of the tasks to be tested and loading test cases based on the execution order to test each node in the multi-node server, the test is made more consistent with the working state of a multi-node server in a real environment, thereby improving the accuracy of the test and enhancing the efficiency, reliability and accuracy of the test.

[0166] In one or more optional embodiments of the present invention, a test script may be obtained during the testing process to determine whether there are abnormal nodes. That is, if abnormal nodes exist among the nodes, before obtaining the test results by performing node anomaly detection based on the running data of the abnormal nodes, the process further includes:

[0167] Obtain the test scripts generated during the testing process;

[0168] The return values ​​in the test script are parsed to determine whether there are any abnormal nodes among the nodes.

[0169] In practical applications, during the testing of nodes in a multi-node server based on a test set, test scripts are generated. The return values ​​from these scripts can be used to determine if any node is abnormal. For abnormal nodes, the test script will report an error through its return value. If the test script reports an error, it indicates the presence of an abnormal node; if it does not, it indicates the absence of an abnormal node. Thus, identifying abnormal nodes through test scripts and their return values ​​improves both efficiency and accuracy.

[0170] In one or more optional embodiments of the present invention, the status of each node in the multi-node server can also be displayed during the test, that is: during the test, the node status of each node is displayed in real time, and the node status includes normal status and abnormal status.

[0171] Specifically, an abnormal state refers to a node malfunctioning or reporting an error. A normal state refers to a node operating normally.

[0172] In practical applications, see Figure 3 , Figure 3 This is a flowchart illustrating the display of node status provided by the present invention: When testing each node of a multi-node server, the node status of each node is displayed in real time, including steps 301 and 302:

[0173] Step 301: Obtain the node status of each node on the multi-node server and display it in real time;

[0174] Step 302: If an anomaly occurs, display or highlight the abnormal node in the multi-node server.

[0175] This allows testers to easily view and monitor the testing process, which helps improve testing efficiency and user satisfaction.

[0176] In one or more optional embodiments of the present invention, after recording the running data of each node, the method further includes: generating a test log based on the running data of each node, so as to record the test process and facilitate testers to find the log.

[0177] The following is combined Figure 4 The multi-node server testing method provided by this invention will be further described below. See also... Figure 4 , Figure 4 This is a second flowchart of the multi-node server testing method provided by the present invention, including steps 401 to 407:

[0178] Step 401: Collect and obtain application data from the multi-node server.

[0179] Specifically, determine the model of the multi-node server to be tested, retrieve the application data for that model based on the model, and then test the multi-node server of that model.

[0180] Step 402: Construct the application scenario of the multi-node server based on the obtained application data of the multi-node server.

[0181] Specifically, based on the application data of the acquired multi-node server, an application scenario for the multi-node server is constructed, including: constructing at least one application scenario for the multi-node server based on the application data of the multi-node server, which can ensure that the testing environment of the multi-node server is more consistent with the working environment of the multi-node server in the real environment.

[0182] Step 403: Generate a test set based on the constructed application scenario.

[0183] Step 404: When testing a multi-node server, load the test set and test each node of the multi-node server using the test samples.

[0184] Step 405: During the test, obtain the running data of each node of the multi-node server.

[0185] Specifically, during the testing process, the operational data of each node on the multi-node server is acquired and recorded to generate test logs for testers to access. The system can also display the real-time status of each node on the multi-node server; if an anomaly occurs, the abnormal node on the multi-node server is displayed and / or highlighted.

[0186] Step 406: Perform node anomaly detection based on the obtained operating data of each node of the multi-node server.

[0187] Specifically, in the presence of abnormal nodes, the operational data of the abnormal nodes in the multi-node server are input into the fault cause analysis model. The fault cause analysis model is trained by using the collected historical abnormal data and historical fault causes of the multi-node server as a sample set, which enables the fault cause analysis model to automatically analyze the fault causes of abnormal nodes, and the accuracy of the fault cause analysis of abnormal nodes is good. Furthermore, the fault cause analysis model outputs the fault cause analysis results.

[0188] Step 407: Finally, output the test results, which include abnormal nodes and abnormal states.

[0189] Specifically, test results are generated and output based on node anomaly detection. The test results include abnormal nodes and abnormal states. At the same time as outputting the test results, if an abnormal node is found, the fault cause analysis results are output simultaneously to help testers find problems so that they can be resolved in the future.

[0190] The multi-node server testing method provided in this embodiment determines the model of the multi-node server, retrieves application data for that model, determines the working application scope of the server based on the obtained application data, constructs application scenarios for the multi-node server, generates a test set based on the constructed application scenarios, loads the test set when testing the multi-node server, and tests each node of the multi-node server using the test set. This effectively simulates real-world application scenarios and results in better testing performance.

[0191] In addition, node anomaly detection is performed based on the operational data of each node of the multi-node server. After a node is identified as an abnormal node, its operational data is input into the fault cause analysis model. The fault cause analysis model outputs the fault cause analysis results and finally outputs the test results. If an anomaly is detected, the fault cause analysis results are output synchronously. It can automatically analyze abnormal nodes and find the cause of the fault to assist testers in discovering problems so that they can be resolved in the future.

[0192] The multi-node server testing apparatus provided by the present invention is described below. The multi-node server testing apparatus described below can be referred to in correspondence with the multi-node server testing method described above.

[0193] Figure 5 This is a schematic diagram of the structure of the multi-node server testing device provided by the present invention, as shown below. Figure 5 As shown, the multi-node server testing device 500 includes: a construction module 501, a testing module 502, and a detection module 503, wherein:

[0194] Module 501 is configured to build a test suite based on the application scenario of the multi-node server to be tested;

[0195] The test module 502 is configured to test each node in the multi-node server based on the test set and record the running data of each node.

[0196] The detection module 503 is configured to perform node anomaly detection based on the running data of the abnormal node when there is an abnormal node among the nodes, and obtain test results.

[0197] The multi-node server testing method and apparatus provided by this invention constructs a test set based on the application scenario of the multi-node server to be tested. This not only improves the correlation between the test set and the application scenario, thus ensuring that the actual application scenario can be effectively simulated during testing, thereby improving the testing effect and reliability of the multi-node server; but also tests each node in the multi-node server through the test set, records the operating data of each node, and performs node anomaly detection based on the operating data of the abnormal nodes when abnormal nodes are present, obtaining test results. It can automatically analyze the abnormal nodes, determine the cause of the failure, and assist testers in discovering problems so that they can be resolved subsequently.

[0198] In one or more optional embodiments of the present invention, the detection module 503 is further configured to:

[0199] The operational data of the abnormal node is input into a pre-trained fault cause analysis model to detect node anomalies and obtain the fault cause analysis results of the abnormal node.

[0200] Test results are generated based on the identifier of the abnormal node, its abnormal state, and the analysis results of the cause of the failure.

[0201] In one or more optional embodiments of the present invention, the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component location unit, a feature location unit, and a fault cause matching unit;

[0202] Accordingly, the detection module 503 is further configured as follows:

[0203] The operational data of the abnormal node is input to the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; the operational data of the abnormal node is input to the feature extraction unit for feature extraction to obtain initial abnormal features, which characterize the fault phenomenon occurring at the abnormal node.

[0204] The fault category is input into the component positioning unit for positioning processing to identify the abnormal component;

[0205] The at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal feature are input to the feature localization unit for feature localization to obtain the target abnormal feature, wherein the target abnormal feature is the preset abnormal feature that matches the initial abnormal feature;

[0206] The target anomaly features are input into the fault cause matching unit for fault cause matching to obtain the fault cause analysis results of the anomaly node.

[0207] In one or more optional embodiments of the present invention, the multi-node server testing apparatus 500 further includes a training module configured to:

[0208] Collect historical operational data and causes of historical failures from historical abnormal nodes;

[0209] The historical operational data is input into the initial analysis model to detect node anomalies and obtain the predicted causes of failures.

[0210] Calculate the loss value based on the predicted causes of failure and the historical causes of failure;

[0211] Adjust the model parameters of the initial analysis model based on the loss value, and continue to execute the steps of collecting historical operating data and historical fault causes of historical abnormal nodes until the preset training stop condition is reached to obtain the trained fault cause analysis model.

[0212] In one or more optional embodiments of the present invention, the multi-node server testing device 500 further includes an acquisition module configured to:

[0213] Obtain the model number of the multi-node server to be tested;

[0214] Retrieve the application data corresponding to the model number;

[0215] Based on the application data, the multi-node server is analyzed to determine the working application scope corresponding to the multi-node server;

[0216] The application scope of the work is divided into scenarios to obtain the application scenarios of the multi-node server;

[0217] Accordingly, the construction module 501 is further configured as follows:

[0218] Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified;

[0219] Collect test cases corresponding to each of the tasks to be tested;

[0220] Construct a test suite based on each of the described test cases;

[0221] Accordingly, the test module 502 is further configured as follows:

[0222] Based on the application scenario, determine the execution order of each of the tasks to be tested;

[0223] According to the execution order, each test case in the test set is loaded to test each node in the multi-node server.

[0224] In one or more optional embodiments of the present invention, the multi-node server testing apparatus 500 further includes a determining module configured to:

[0225] Obtain the test scripts generated during the testing process;

[0226] The return values ​​in the test script are parsed to determine whether there are any abnormal nodes among the nodes.

[0227] In one or more optional embodiments of the present invention, the multi-node server testing device 500 further includes a display module configured to:

[0228] During the test, the node status of each node is displayed in real time, including normal status and abnormal status.

[0229] In one or more optional embodiments of the present invention, the multi-node server testing apparatus 500 further includes a generation module configured to:

[0230] Test logs are generated based on the operational data of each node.

[0231] In one or more optional embodiments of the present invention, the building module 501 is further configured to:

[0232] The application scenario of the multi-node server to be tested is analyzed to determine at least one test case corresponding to the application scenario.

[0233] Construct a test suite based on each of the test cases described.

[0234] In one or more optional embodiments of the present invention, the building module 501 is further configured to:

[0235] Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified;

[0236] Collect the test cases corresponding to each of the tasks to be tested.

[0237] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communications interface 620, a memory 630, and a communication bus 640, wherein the processor 610, communications interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a multi-node server testing method. This method includes: constructing a test set according to the application scenario of the multi-node server to be tested; testing each node in the multi-node server based on the test set and recording the operating data of each node; and, in the case of an abnormal node, performing node anomaly detection based on the operating data of the abnormal node to obtain test results.

[0238] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0239] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the multi-node server testing method provided by the above methods. The method includes: constructing a test set according to the application scenario of the multi-node server to be tested; testing each node in the multi-node server based on the test set and recording the running data of each node; and, in the case of abnormal nodes among the nodes, performing node anomaly detection based on the running data of the abnormal nodes to obtain test results.

[0240] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the multi-node server testing method provided by the above methods. The method includes: constructing a test set according to the application scenario of the multi-node server to be tested; testing each node in the multi-node server based on the test set and recording the operating data of each node; and, in the case of an abnormal node among the nodes, performing node anomaly detection based on the operating data of the abnormal node to obtain test results.

[0241] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0242] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0243] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-node server testing method, characterized in that, include: A test suite is constructed based on the application scenario of the multi-node server to be tested; the application scenario refers to the type of service provided by the multi-node server to be tested. The test set is used to test each node in the multi-node server, and the running data of each node is recorded. If there are abnormal nodes among the nodes, node anomaly detection is performed based on the operational data of the abnormal nodes to obtain test results; The steps for obtaining the test results include: Obtain a pre-trained fault cause analysis model; the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component localization unit, a feature localization unit, and a fault cause matching unit; The operational data of the abnormal node is input into the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; The running data of the abnormal node is input into the feature extraction unit for feature extraction to obtain initial abnormal features, which characterize the fault phenomenon that occurred in the abnormal node. The fault category is input into the component positioning unit for positioning processing to identify the abnormal component; The at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal feature are input to the feature localization unit for feature localization to obtain the target abnormal feature, wherein the target abnormal feature is the preset abnormal feature that matches the initial abnormal feature; The target anomaly features are input into the fault cause matching unit to perform fault cause matching, and the fault cause analysis results of the anomaly node are obtained. Test results are generated based on the identifier of the abnormal node, its abnormal state, and the analysis results of the cause of the failure. Before constructing the test set based on the application scenario of the multi-node server to be tested, the following steps are also included: Obtain the model number of the multi-node server to be tested; Retrieve the application data corresponding to the model number; Based on the application data, the multi-node server is analyzed to determine the working application scope corresponding to the multi-node server; The application scope of the work is divided into scenarios to obtain the application scenarios of the multi-node server; Accordingly, the step of constructing a test set based on the application scenario of the multi-node server to be tested includes: Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified; Collect test cases corresponding to each of the tasks to be tested; Construct a test suite based on each of the described test cases; Accordingly, the testing of each node in the multi-node server based on the test set includes: Based on the application scenario, determine the execution order of each of the tasks to be tested; According to the execution order, each test case in the test set is loaded to test each node in the multi-node server.

2. The multi-node server testing method according to claim 1, characterized in that, Before inputting the operational data of the abnormal node into the pre-trained fault cause analysis model for node anomaly detection, the method further includes: Collect historical operational data and causes of historical failures from historical abnormal nodes; The historical operational data is input into the initial analysis model to detect node anomalies and obtain the predicted causes of failures. Calculate the loss value based on the predicted causes of failure and the historical causes of failure; Adjust the model parameters of the initial analysis model based on the loss value, and continue to execute the steps of collecting historical operating data and historical fault causes of historical abnormal nodes until the preset training stop condition is reached to obtain the trained fault cause analysis model.

3. The multi-node server testing method according to any one of claims 1-2, characterized in that, Before obtaining the test result by performing node anomaly detection based on the operational data of the abnormal nodes when abnormal nodes exist among the nodes, the method further includes: Obtain the test scripts generated during the testing process; The return values ​​in the test script are parsed to determine whether there are any abnormal nodes among the nodes.

4. The multi-node server testing method according to any one of claims 1-2, characterized in that, Also includes: During the test, the node status of each node is displayed in real time, including normal status and abnormal status.

5. The multi-node server testing method according to any one of claims 1-2, characterized in that, After recording the operational data of each node, the method further includes: Test logs are generated based on the operational data of each node.

6. A multi-node server testing device, characterized in that, include: The build module is configured to build a test suite based on the application scenario of the multi-node server to be tested; the application scenario refers to the type of service provided by the multi-node server to be tested. The testing module is configured to test each node in the multi-node server based on the test set and record the running data of each node. The detection module is configured to perform node anomaly detection based on the running data of the abnormal nodes when there are abnormal nodes among the nodes, and obtain test results. The steps for obtaining the test results include: Obtain a pre-trained fault cause analysis model; the fault cause analysis model includes a category parsing unit, a feature extraction unit, a component localization unit, a feature localization unit, and a fault cause matching unit; The operational data of the abnormal node is input into the category parsing unit for category parsing to obtain the fault category corresponding to the abnormal node; The running data of the abnormal node is input into the feature extraction unit for feature extraction to obtain initial abnormal features, which characterize the fault phenomenon that occurred in the abnormal node. The fault category is input into the component positioning unit for positioning processing to identify the abnormal component; The at least one preset abnormal feature corresponding to the abnormal component and the initial abnormal feature are input to the feature localization unit for feature localization to obtain the target abnormal feature, wherein the target abnormal feature is the preset abnormal feature that matches the initial abnormal feature; The target anomaly features are input into the fault cause matching unit to perform fault cause matching, and the fault cause analysis results of the anomaly node are obtained. Test results are generated based on the identifier of the abnormal node, its abnormal state, and the analysis results of the cause of the failure. Before constructing the test set based on the application scenario of the multi-node server to be tested, the following steps are also included: Obtain the model number of the multi-node server to be tested; Retrieve the application data corresponding to the model number; Based on the application data, the multi-node server is analyzed to determine the working application scope corresponding to the multi-node server; The application scope of the work is divided into scenarios to obtain the application scenarios of the multi-node server; Accordingly, the step of constructing a test set based on the application scenario of the multi-node server to be tested includes: Based on the application scenario of the multi-node server to be tested, at least one task to be tested should be identified; Collect test cases corresponding to each of the tasks to be tested; Construct a test suite based on each of the described test cases; Accordingly, the testing of each node in the multi-node server based on the test set includes: Based on the application scenario, determine the execution order of each of the tasks to be tested; According to the execution order, each test case in the test set is loaded to test each node in the multi-node server.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the multi-node server testing method as described in any one of claims 1 to 5.

8. A non-transitory 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 multi-node server testing method as described in any one of claims 1 to 5.

Images