Method, device and equipment for monitoring and processing cluster and storage medium
Through timing decomposition and frequency domain attention mechanisms, the cluster's index data is screened and processed, and feature extraction and fusion is combined with multi-layer perception mechanisms, which solves the problem that the existing technology cannot effectively monitor and predict data processing cluster abnormalities, and achieves efficient and accurate anomaly prediction and monitoring.
Patent Information
- Application Number
- CN202411715677.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-27
- Publication Date
- 2025-03-18
AI Technical Summary
The prior art is unable to effectively monitor and predict potential anomalies in data processing clusters, resulting in low monitoring efficiency and poor accuracy.
By obtaining and recording the index data of the processing cluster, performing timing decomposition and frequency domain attention mechanism screening, using multi-layer perception mechanisms for feature extraction and feature fusion, predicting whether abnormalities occur in the processing cluster.
It realizes the advance prediction of abnormal situations in the handling cluster, improves monitoring efficiency and accuracy, and can promptly detect and repair potential problems and avoid safety accidents.
Smart Images
Figure CN119645761A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a method, apparatus, device and storage medium for monitoring a processing cluster. Background Art
[0002] With the development of technology, large-scale data processing clusters have been widely used in many fields. For example, in financial institutions such as banks, large-scale data processing clusters can be used to process user data and transaction data to better serve users. In addition, for security reasons, for branches and sub-branches of banks, centralized data processing clusters can be used to split and process data to ensure the security of the processed data.
[0003] Large-scale data processing clusters may have anomalies during operation. In order to ensure the normal processing and security of data, it is necessary to monitor the data processing cluster so as to promptly detect anomalies in the data processing cluster and repair them in time to avoid safety accidents. At present, most data monitoring methods rely on manual monitoring by technicians. Alternatively, data monitoring rules can be pre-set to analyze the data generated during the operation of the cluster through the preset rules.
[0004] However, the above two monitoring methods have the problems of low monitoring efficiency and poor accuracy. In particular, the existing monitoring methods cannot detect potential problems in the cluster and cannot predict possible abnormalities in the cluster. Summary of the invention
[0005] In view of this, the present application provides a method and apparatus for monitoring a processing cluster, aiming to predict possible anomalies in the processing cluster through a model.
[0006] In a first aspect, the present application provides a method for monitoring a processing cluster, the method comprising:
[0007] Obtain and record the indicator data of the processing cluster in chronological order;
[0008] Based on the time sequence, the indicator data is decomposed in time series to determine monitoring data parameter items of the indicator data;
[0009] Based on the frequency domain attention mechanism, the monitoring data parameter items are screened by importance to determine at least one target monitoring parameter item;
[0010] The at least one target monitoring parameter item is subjected to feature extraction and feature fusion through a multi-layer perception mechanism, and a predicted monitoring result is determined based on a result obtained by the feature fusion, wherein the predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
[0011] In some possible implementations, the extracting features of the at least one target monitoring parameter item by using a multi-layer perception mechanism includes:
[0012] Segmenting the target monitoring parameter items multiple times in chronological order;
[0013] The segmented target monitoring parameter items are subjected to feature extraction through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter items in a time series.
[0014] In some possible implementations, the step of performing time series decomposition on the indicator data based on the time series further includes:
[0015] Decomposing the indicator data in time series according to the time series to determine the trend item of the indicator data;
[0016] The feature fusion includes:
[0017] The trend item and the multi-scale feature are fused in time scale.
[0018] In some possible implementations, determining the predicted monitoring result based on the result obtained by the feature fusion includes:
[0019] Through the fully connected layer, the result obtained by fusion of the features is mapped to the time series in the time direction to obtain the predicted monitoring result.
[0020] In a second aspect, the present application provides a device for monitoring a processing cluster, the device comprising:
[0021] An acquisition unit, used to acquire and record the indicator data of the processing cluster in chronological order;
[0022] A decomposition unit, used to perform time series decomposition on the indicator data based on a time sequence, and determine monitoring data parameter items of the indicator data;
[0023] A screening unit, configured to screen the monitoring data parameter items by importance based on a frequency domain attention mechanism, and determine at least one target monitoring parameter item;
[0024] The prediction unit is used to extract and fuse features of the at least one target monitoring parameter item through a multi-layer perception mechanism, and determine a predicted monitoring result based on a result obtained by the feature fusion, wherein the predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
[0025] In some possible implementations, the prediction unit is specifically used to segment the target monitoring parameter items multiple times in chronological order; and perform feature extraction on the segmented target monitoring parameter items through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter items in the time series.
[0026] In some possible implementations, the decomposition unit is further used to perform time series decomposition on the indicator data based on a time series to determine a trend item of the indicator data;
[0027] The prediction unit is specifically used to perform feature fusion on the trend item and the multi-scale feature in a time scale.
[0028] In some possible implementations, the prediction module is specifically used to map the result obtained by the feature fusion to a time series in the time direction through a fully connected layer to obtain a predicted monitoring result.
[0029] In a third aspect, the present application provides a device, comprising a memory and a processor, wherein the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the method for monitoring a processing cluster as described in any one of the first aspects.
[0030] In a fourth aspect, the present application provides a computer storage medium having codes stored therein. When the codes are executed, a device executing the codes implements the method for monitoring a processing cluster as described in any one of the first aspects.
[0031] In a fifth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute the method for monitoring a processing cluster as described in any one of the first aspects.
[0032] The present application provides a method, apparatus, device and storage medium for monitoring a processing cluster. The method for monitoring a processing cluster can predict abnormal conditions of a processing cluster in advance. Specifically, during the operation of the processing cluster, the index data of the processing cluster can be obtained and recorded in chronological order, thereby converting the data generated by the processing cluster into a time series. Then, the index data can be decomposed into a time series based on the time sequence to determine the monitoring data parameter items of the index data. After determining the monitoring data parameter items, the monitoring data parameter items can be screened for importance based on the frequency domain attention mechanism, and at least one target monitoring parameter item with a high monitoring value can be determined. Finally, feature extraction and feature fusion can be performed on at least one extracted target monitoring parameter item through a multi-layer perception mechanism, and a predicted monitoring result can be determined based on the result obtained by the feature fusion. The predicted monitoring result indicates whether the processing cluster is abnormal. In this way, based on the frequency domain attention mechanism, data with monitoring value can be screened out from a large amount of data generated by the processing cluster, so that by analyzing the data with monitoring value, it is predicted whether the processing cluster will be abnormal. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the technical solutions in this embodiment or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0034] Figure 1 A schematic diagram of a flow chart of a method for monitoring and processing a cluster provided in an embodiment of the present application;
[0035] Figure 2 A schematic diagram of the structure of a processing cluster monitoring device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0036] For large institutions such as banks, large-scale batch data processing is an important part of daily operations. In order to ensure that data is processed normally and safely, it is necessary to monitor the cluster that processes data, that is, the data processing cluster, so as to detect and eliminate abnormal situations in time.
[0037] At present, the processing cluster can be monitored by designing data collection rules and data monitoring rules. Specifically, the data generated by the processing cluster during the data processing process can be collected according to the pre-designed data collection rules. In addition, the collected data can be analyzed and judged according to the pre-set data monitoring rules to determine whether the processing cluster is abnormal.
[0038] However, the above monitoring methods are insufficient in functionality and cannot meet the specific monitoring needs of users. They are also less flexible and difficult to adapt to the dynamic changes of the processing cluster. For example, as the business carried by the cluster develops, the data to be monitored and the abnormal standards may change, and the traditional method of setting rules cannot meet this demand.
[0039] In order to solve the above problems, an embodiment of the present application provides a method for monitoring a processing cluster. The method combines time domain analysis and frequency domain analysis to predict whether an abnormality will occur in the processing cluster, so as to promptly remind the technician to perform troubleshooting. The method for monitoring a processing cluster can be applied to a processing cluster monitoring device. Optionally, the processing cluster monitoring device can be run on a server, and can be specifically applied to a monitoring server of a processing cluster.
[0040] The following describes the method for monitoring a processing cluster provided by an embodiment of the present application from the perspective of a processing cluster monitoring device. Obviously, the described embodiment is only a part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.
[0041] See also Figure 1 , Figure 1 A method flow chart of a method for monitoring a processing cluster provided in an embodiment of the present application includes:
[0042] S101: Obtain and record indicator data of a processing cluster in chronological order.
[0043] In order to predict and monitor abnormalities of the processing cluster, it is first necessary to obtain the working status of the processing cluster. In an embodiment of the present application, the working status of the processing cluster can be characterized by the indicator data of the processing cluster. Therefore, when monitoring the processing cluster, it is first necessary to obtain the indicator data of the processing cluster. Optionally, the indicator data can be collected in real time. Accordingly, the indicator data can be recorded in the order in which the indicator data is collected.
[0044] In order to better process the indicator data of the processing cluster, in the embodiment of the present application, time domain analysis and frequency domain analysis can be introduced. To this end, when acquiring the indicator data, the indicator data of the processing cluster can be recorded in chronological order. Optionally, the indicator data of the processing cluster recorded in chronological order can be referred to as an indicator time series sequence or a time series sequence.
[0045] Optionally, the indicator data may include various data generated during the operation of the processing cluster, for example, the working status and operation status of each node in the processing cluster. The working status of the node may include data such as the CPU occupancy, memory occupancy, and storage space occupancy of the node. The operation status may include, for example, the average operation time, the completion rate of the operation, the delay of the operation, the log error information, etc. Through the working status and operation status of the node, the working status of the node can be analyzed to determine whether the cluster and each node in the cluster will be abnormal.
[0046] After obtaining the indicator data, the indicator data can be preprocessed to improve the efficiency of monitoring and the accuracy of monitoring results. Specifically, normalization, feature selection and data cleaning can be performed on indicator data of different scales to filter out data with lower value from a large amount of indicator data and retain data with higher value to improve the efficiency of subsequent data processing.
[0047] Optionally, early warning indicators can be pre-set for some indicator data. For example, an upper threshold can be set for indicator data such as the CPU usage of a node, the memory usage of a node, and the storage space usage of a node. When the indicator data reaches the upper threshold, an abnormal alarm message can be directly generated. For another example, corresponding thresholds can also be set for data such as the average duration of a job, the completion rate of a job, and the delay time of a job, so that the above indicator data can be controlled within a controllable range through the threshold.
[0048] After obtaining the indicator data, the indicator data can be analyzed by the model. Specifically, the indicator data can be input into the model, and the model implements the following steps S102 to S104 to obtain the prediction monitoring results. The model can include modules such as time series decomposition module, frequency domain attention mechanism module, multi-scale feature extraction module, feature fusion module and full connection layer.
[0049] S102: Based on the time sequence, the indicator data is decomposed in time series to determine monitoring data parameter items of the indicator data.
[0050] Specifically, after the indicator data is input into the model, the model can first perform time domain analysis on the indicator data. The time domain analysis module of the model can perform time series analysis on the input indicator data according to the time sequence of the data to determine the monitoring data parameter items of the indicator data.
[0051] Optionally, the changing trend of the indicator data can also be determined through time domain analysis. That is, the analysis module can also perform time series analysis on the input indicator data according to the time sequence of the data to determine the trend item of the indicator data. The trend item of the indicator data can be used to indicate the change of the indicator data. In the subsequent analysis process, the trend item of the indicator data can be combined to better perform feature extraction and feature fusion.
[0052] S103: Based on the frequency domain attention mechanism, importance screening is performed on monitoring data parameter items to determine at least one target monitoring parameter item.
[0053] After the monitoring data parameter items are determined, importance screening can be performed based on the frequency domain attention mechanism to determine at least one target monitoring parameter item with higher importance. Through the frequency domain attention mechanism module in the model, the importance of each monitoring data item can be analyzed based on the characteristics of the monitoring data item in the frequency domain, and then the monitoring data parameter items that meet the importance requirements can be determined as target monitoring parameter items.
[0054] In actual application scenarios, the processing cluster generates a large amount of data during the working process, of which a small part of the data can reflect the abnormal situation of the processing cluster or reflect that the processing cluster is about to have an abnormal situation. If all the data are processed, not only will the processing efficiency be reduced, but it may also be impossible to find truly valuable data. In an embodiment of the present application, by using the frequency domain attention mechanism for importance screening, valuable and worthy data can be found from a large amount of monitoring data parameter items, so as to monitor and predict the abnormal situation of the processing cluster more efficiently and accurately.
[0055] S104: Perform feature extraction and feature fusion on at least one target monitoring parameter item through a multi-layer perception mechanism, and determine a predicted monitoring result based on a result obtained by the feature fusion.
[0056] After determining at least one target monitoring parameter item, a feature extraction can be performed on the at least one target monitoring parameter item through a multi-layer perception mechanism, and the extracted features can be fused, and finally a predicted monitoring result can be determined based on the result of the feature fusion. The predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
[0057] Specifically, when performing feature extraction, the target monitoring parameter items can be segmented first. For example, the multi-scale feature extraction module in the model can segment the target monitoring parameter items multiple times in chronological order. Then, the segmented target monitoring parameter items can be feature extracted through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter items in the time series. These multi-scale features can reflect the information implicit in the target monitoring parameter items that characterizes the abnormality or impending abnormality of the processing cluster.
[0058] After feature extraction, multi-scale features can be fused. Specifically, the feature fusion module in the model can fuse multi-scale features of different target monitoring parameter items. Through feature fusion, the information of different target monitoring parameter items can be integrated to comprehensively judge whether the processing cluster is abnormal or whether there are signs of abnormality.
[0059] Optionally, feature fusion can also be performed in combination with the trend of changes in monitoring data. Specifically, if the trend item of the indicator data is determined in the aforementioned step S102. Then, during feature fusion, the trend item and the multi-scale feature can be fused on the time scale to obtain the features of the multivariate time series. In this way, not only the characteristics of the target monitoring parameter item itself are taken into account, but also the changing trend of the indicator data is taken into account, so that the working condition of the processing cluster can be evaluated more accurately.
[0060] After the feature fusion module fuses the features, the prediction monitoring result can be determined based on the result obtained by the feature fusion. Specifically, the fused features can be input into the fully connected layer of the model. The fully connected layer can map the fused features to the time series in the practical direction, thereby obtaining the prediction output result.
[0061] After obtaining the prediction output result, the prediction output result can be displayed through a visualization tool. Optionally, if the prediction output result is obtained based on one or several indicator data, the above indicator data can also be displayed through a visualization tool. Or, optionally, the target monitoring parameter item can also be displayed through a visualization tool.
[0062] The present application provides a method for monitoring a processing cluster, which can predict the abnormal situation of the processing cluster in advance. Specifically, in the process of processing the cluster, the index data of the processing cluster can be obtained and recorded in chronological order, so as to convert the data generated by the processing cluster into a time series. Then, the index data can be decomposed into a time series based on the time sequence to determine the monitoring data parameter items of the index data. After determining the monitoring data parameter items, the monitoring data parameter items can be screened for importance based on the frequency domain attention mechanism, and at least one target monitoring parameter item with high monitoring value can be determined. Finally, feature extraction and feature fusion can be performed on at least one extracted target monitoring parameter item through a multi-layer perception mechanism, and the predicted monitoring result can be determined based on the result obtained by the feature fusion. The predicted monitoring result indicates whether the processing cluster is abnormal. In this way, based on the frequency domain attention mechanism, data with monitoring value can be screened out from a large amount of data generated by the processing cluster, so as to predict whether the processing cluster will be abnormal by analyzing the data with monitoring value.
[0063] In the above implementation, the model indicator data can be analyzed to determine whether the processing cluster has or is about to have an abnormality. Adaptively, the embodiment of the present application also provides a corresponding model training method.
[0064] When training a model, you can first collect training data and preprocess the data. The training data can be data before the cluster anomaly occurs. Then, you can use supervised learning to train the model by calculating loss values and back propagation. Specifically, during the training process, you can follow the gradient back propagation process to obtain the direction of the gradient in the parameters in the network, and use stochastic gradient descent (SGD) to optimize the update parameters in the model.
[0065] The above are some specific implementations of the method for monitoring a processing cluster provided in the embodiment of the present application. Based on this, the present application also provides a corresponding processing cluster monitoring device. The processing cluster monitoring device provided in the embodiment of the present application will be introduced below from the perspective of functional modularization.
[0066] See also Figure 2 , Figure 2 A schematic diagram of a processing cluster monitoring device provided in an embodiment of the present application. Specifically, Figure 2 The processing cluster monitoring device 200 shown includes:
[0067] An acquisition unit 210 is used to acquire and record the indicator data of the processing cluster in chronological order;
[0068] A decomposition unit 220, configured to perform time series decomposition on the indicator data based on a time sequence to determine monitoring data parameter items of the indicator data;
[0069] A screening unit 230 is used to screen the monitoring data parameter items by importance based on a frequency domain attention mechanism to determine at least one target monitoring parameter item;
[0070] The prediction unit 240 is used to extract and fuse features of the at least one target monitoring parameter item through a multi-layer perception mechanism, and determine a predicted monitoring result based on the result obtained by the feature fusion, wherein the predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
[0071] In some possible implementations, the prediction unit 240 is specifically used to segment the target monitoring parameter item multiple times in chronological order; and perform feature extraction on the segmented target monitoring parameter item through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter item in the time series.
[0072] In some possible implementations, the decomposition unit 220 is further configured to perform time series decomposition on the indicator data based on a time series to determine a trend item of the indicator data;
[0073] The prediction unit 240 is specifically configured to perform feature fusion on the trend item and the multi-scale feature in a time scale.
[0074] In some possible implementations, the prediction module 240 is specifically configured to map the result obtained by the feature fusion to a time series in the time direction through a fully connected layer to obtain a prediction monitoring result.
[0075] The present application provides a processing cluster monitoring device, which can predict the abnormal situation of the processing cluster in advance. Specifically, in the process of processing cluster work, the index data of the processing cluster can be obtained and recorded in chronological order, so as to convert the data generated by the processing cluster into a time series. Then, the index data can be decomposed into a time series based on the time sequence to determine the monitoring data parameter items of the index data. After determining the monitoring data parameter items, the monitoring data parameter items can be screened for importance based on the frequency domain attention mechanism, and at least one target monitoring parameter item with high monitoring value can be determined. Finally, feature extraction and feature fusion can be performed on at least one extracted target monitoring parameter item through a multi-layer perception mechanism, and the predicted monitoring result can be determined based on the result obtained by feature fusion. The predicted monitoring result indicates whether the processing cluster is abnormal. In this way, based on the frequency domain attention mechanism, data with monitoring value can be screened out from a large amount of data generated by the processing cluster, so as to predict whether the processing cluster will be abnormal by analyzing the data with monitoring value.
[0076] The embodiments of the present application also provide corresponding devices, computer storage media and computer program products for implementing the technical solutions provided by the embodiments of the present application.
[0077] The device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the method for monitoring a processing cluster as described in any embodiment of the present application.
[0078] The computer storage medium stores codes, and when the codes are executed, the device executing the codes implements the method for monitoring a processing cluster described in any embodiment of the present application.
[0079] The computer program product includes instructions. When the instructions are executed on a computer, the computer is enabled to execute the method for monitoring a processing cluster described in any embodiment of the present application.
[0080] The "first" and "second" in the names such as "first" and "second" (if any) mentioned in the embodiments of the present application are only used as name identifiers and do not represent the first or second in order.
[0081] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as a read-only memory (ROM) / ROM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the methods described in each embodiment of the present application or some parts of the embodiments.
[0082] Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.
[0083] The above description is merely an exemplary embodiment of the present application and is not intended to limit the protection scope of the present application.
Claims
1. A method for monitoring a processing cluster, characterized in that: The method comprises: Obtain and record the metrics data of the processing cluster in chronological order; Based on the time sequence, the indicator data is decomposed in time series to determine monitoring data parameter items of the indicator data; Based on the frequency domain attention mechanism, the monitoring data parameter items are screened by importance to determine at least one target monitoring parameter item; The at least one target monitoring parameter item is subjected to feature extraction and feature fusion through a multi-layer perception mechanism, and a predicted monitoring result is determined based on a result obtained by the feature fusion, wherein the predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
2. The method according to claim 1, characterized in that The extracting features of the at least one target monitoring parameter item by using a multi-layer perception mechanism includes: Segmenting the target monitoring parameter items multiple times in chronological order; The segmented target monitoring parameter items are subjected to feature extraction through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter items in a time series.
3. The method according to claim 2, characterized in that The step of performing time series decomposition on the indicator data based on the time series further includes: Decomposing the indicator data in time series according to the time series to determine the trend item of the indicator data; The feature fusion includes: The trend item and the multi-scale feature are fused in time scale.
4. The method according to claim 3, characterized in that Determining the predicted monitoring result based on the result obtained by the feature fusion includes: Through the fully connected layer, the result obtained by fusion of the features is mapped to the time series in the time direction to obtain the predicted monitoring result.
5. A device for monitoring a processing cluster, characterized in that: The device comprises: An acquisition unit, used to acquire and record the indicator data of the processing cluster in chronological order; A decomposition unit, used to perform time series decomposition on the indicator data based on a time sequence, and determine monitoring data parameter items of the indicator data; A screening unit, configured to screen the monitoring data parameter items by importance based on a frequency domain attention mechanism, and determine at least one target monitoring parameter item; A prediction unit is used to extract and fuse features of the at least one target monitoring parameter item through a multi-layer perception mechanism, and determine a predicted monitoring result based on a result obtained by the feature fusion, wherein the predicted monitoring result indicates whether an abnormality occurs in the processing cluster.
6. The device according to claim 5, characterized in that The prediction unit is specifically used to segment the target monitoring parameter item multiple times in chronological order; and extract features of the segmented target monitoring parameter item through a multi-layer perception mechanism to obtain multi-scale features of the target monitoring parameter item in the time series.
7. The device according to claim 6, characterized in that The decomposition unit is further used to perform time series decomposition on the indicator data based on the time series to determine the trend item of the indicator data; The prediction unit is specifically used to perform feature fusion on the trend item and the multi-scale feature in a time scale.
8. The device according to claim 7, characterized in that The prediction module is specifically used to map the result obtained by the feature fusion to a time series in the time direction through a fully connected layer to obtain a prediction monitoring result.
9. A device, characterized in that: The device includes a memory and a controller, the memory is used to store instructions or codes, and the controller is used to execute the instructions or codes stored in the memory to implement the method for monitoring a processing cluster as described in any one of claims 1-4.
10. A computer storage medium, characterized in that: The computer storage medium stores codes, and when the codes are executed, a device executing the codes implements the method for monitoring a processing cluster according to any one of claims 1 to 4.