High-performance scheduling method and system, program product and storage medium

Abstract

The invention discloses a high-performance scheduling method and system, a program product and a storage medium. In the method, after an order creation task request of a cross-border e-commerce system or an external system is received, tasks are executed in sequence, initial cache data are generated when the tasks are executed, and the state is monitored; when the execution state meets a preset error condition, determining an error type; when retry can be carried out and the cached data are healthy, retry is carried out according to strategy retry; if the cached data is not healthy, extracting a healthy cache for replacement; and if the retry fails, the order creation task is degraded to obtain and execute the degraded task, so that the order creation task has a perfect coping mechanism when facing various error conditions, the success rate and stability of cross-border e-commerce system order creation are improved, and the efficiency of order creation task execution is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a high-performance scheduling method, system, program product and storage medium. Background Art

[0002] In today's digital age, e-commerce platforms are widely popular and have become an important carrier of commercial activities. Many companies use e-commerce platforms to carry out sales, marketing and other businesses. It is of great significance for e-commerce platforms to obtain third-party data. It can enrich their own product information, optimize the recommendation system, improve user experience, and assist companies in market analysis, precision marketing, and supply chain management, etc., which helps companies stand out in fierce competition and achieve sustainable development.

[0003] At present, e-commerce platforms usually adopt a distributed architecture to handle a large number of order creation tasks. Tasks are assigned to multiple server nodes, and data validation rules are set at the same time to perform data checks at each stage of order creation, such as product information, user information, delivery address, etc. If an error occurs, the system will immediately trigger an alarm, mark the erroneous order and suspend processing. Then the technicians will make corrections based on the type of error, which may be to modify data errors or adjust system parameters to ensure smooth processing of subsequent orders.

[0004] However, during the peak period of cross-border e-commerce business, a large number of order creation task requests pour in. At this time, when the system faces a complex task environment and potential system resource fluctuations, the overall execution success rate of the order creation task will be reduced. However, when an error occurs in the order creation task, it is necessary to wait for technical personnel to correct it, which reduces the efficiency of the order creation task execution. Summary of the invention

[0005] The present application provides a high-performance scheduling method, system, program product and storage medium for improving the efficiency of order creation task execution.

[0006] In a first aspect, the present application provides a high-performance scheduling method for a cross-border e-commerce system, the method comprising: after receiving an order creation task request from a business module or an external system of the cross-border e-commerce system, executing the order creation tasks in the task request in sequence; after the current order creation task starts to execute, generating initial cache data of the order to be created, and monitoring the execution status of the current order creation task; when it is detected that the execution status meets a preset execution error condition, determining that the current order creation task has an execution error, locking the structure of the current cache data, and judging whether the error type of the current order creation task execution error belongs to a preset retryable error type based on the operation error keywords found from the operation log; the current cache data is the cache data obtained after reading and writing the initial cache data during task execution; the preset execution error condition at least includes that the weighted average of the CPU usage and the memory occupancy exceeds the preset average threshold; in the case of a preset retryable error type, checking whether the current cache data is in a healthy state; the healthy state is the difference between the current cache data and the initial cache data. The data structure and data format of are consistent; when the current cache data is not in the healthy state, extract the healthy cache data recorded in the system log of the current order creation task that was in the healthy state the most recently, and use the healthy cache data to replace the current cache data; when the current cache data is in the healthy state, determine the retry strategy according to the preset retry rule corresponding to the error type, and start retrying; the retry strategy includes the maximum number of retries and the retry interval time; during the retry process, monitor the execution status of the current order creation task, and when no execution error of the current order creation task is detected, continue to execute the current order creation task; when an execution error of the current order creation task is detected, retry again, and when the number of retries reaches the maximum number of retries and the execution error of the current order creation task is still detected, downgrade the current order creation task to obtain a downgraded task, and execute the downgraded task. During the execution of the downgraded task, when no execution error of the downgraded task is detected, the current order creation task is marked as a downgraded successful task; the downgrade process is to adjust the preset average value threshold.

[0007] In the above embodiment, after receiving the order creation task request, the tasks are executed in sequence and the status is monitored. When an error is encountered during the execution process, the cached data is locked, the error type and retryability are determined, and retries are performed according to the preset retry strategy when the cached data is healthy. If the retry fails, the preset average value threshold is adjusted for downgrade processing, thereby improving the system's ability to deal with errors and ensure business continuity and the efficiency of order creation task execution.

[0008] In combination with some embodiments of the first aspect, in some embodiments, after receiving an order creation task request from a business module or an external system of a cross-border e-commerce system, the order creation tasks in the task request are executed in sequence, specifically including: after receiving an order creation task request from a business module or an external system of a cross-border e-commerce system, each order creation task in the order creation task request is included in an order creation task queue; a preset reinforcement learning model is trained using an order creation task processing sequence training data group to obtain an order creation task processing sequence model; the order creation task processing sequence model is used to output order creation task processing sequence data based on a set of input core processor load ratio data, remaining memory ratio data, order item quantity data of the order creation task, and order creation task priority data; the order creation task processing sequence training data group includes multiple order creation task processing sequence training data, and each training data includes a a corresponding relationship between a core processor load ratio data, a remaining memory ratio data, an order commodity quantity data of an order creation task, an order creation task priority data and a corresponding marked completed order creation task processing sequence data; collect the real-time core processor load ratio data, the real-time remaining memory ratio data, the order commodity quantity data and the order creation task priority data of each order creation task in the order creation task queue, and retrieve the real-time core processor load ratio data, the real-time remaining memory ratio data, the order commodity quantity data and the order creation task priority data of each order creation task in the order creation task queue, and input them into the order creation task processing sequence model in sequence to obtain the processing sequence data of each order creation task in the order creation task queue; adjust the sequence of each order creation task in the order creation task queue according to the order creation task processing sequence data, and take out the order creation tasks from the order creation task queue in sequence and execute them.

[0009] In the above embodiment, by incorporating order tasks into a queue, training a reinforcement learning model and using it to determine the task processing order based on multiple types of data, collecting real-time data input into the model to obtain the queue task order and then adjusting and executing it, the cross-border e-commerce order task allocation is made more scientific and reasonable, and the order processing efficiency and the degree of optimization of system resource utilization are improved.

[0010] In combination with some embodiments of the first aspect, in some embodiments, after adjusting the order of each order creation task in the order creation task queue according to the order creation task processing sequence data, taking out the order creation tasks from the order creation task queue in sequence and executing them, it also includes: after each task is taken out from the order creation task queue, obtaining a real-time order creation task queue and collecting the core processor load ratio data and the remaining memory ratio data at that time; retrieving the core processor load ratio data at that time, the remaining memory ratio data at that time, the order commodity quantity data and the order creation task priority data of each order creation task in the real-time order creation task queue, and inputting them into the task processing sequence model in sequence to obtain the processing sequence data of each order creation task in the real-time order creation task queue, and adjusting the order of each task in the real-time task queue according to the processing sequence data of each order creation task in the real-time order creation task queue.

[0011] In the above embodiment, after taking out the task, data such as the core processor, memory and other task-related information are collected, the task processing sequence model is input, and the real-time task queue sequence is adjusted according to the result, so that the high-performance scheduling system can dynamically optimize the task processing sequence according to the real-time status, thereby improving the system's flexibility and adaptability to the execution of order tasks.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after receiving an order creation task request from a business module or an external system of a cross-border e-commerce system, and executing the order creation tasks in the task request in sequence, it also includes: using a predicted order peak training data group to train a preset neural network model to obtain a predicted order peak model; the predicted order peak model is used to output predicted order peak data based on a set of input network status data and data synchronization status data; the predicted order peak training data group includes multiple predicted order peak training data, and each training data includes a corresponding relationship between a network status data, a data synchronization status data and a corresponding marked completed predicted order peak data; real-time network status data and real-time data synchronization status data are obtained in real time, and input into the predicted order peak model to obtain predicted order peak data, and derive a predicted order peak time period; when the time difference between the real-time time and the predicted order peak time is less than a preset time difference range, the data collection frequency is adjusted by a multiple of the predicted peak duration ratio value; the predicted peak duration ratio value is the ratio of the duration of the predicted order peak time period to the preset peak duration.

[0013] In the above embodiment, the neural network model is trained using the training data of the predicted order peak period, and the network and data synchronization data are input into the model in real time to obtain the predicted order peak period. When approaching the peak period, the data collection frequency is adjusted proportionally, so that the cross-border e-commerce system can predict and prepare in advance, thereby improving the timeliness of data processing and system stability when dealing with order peaks.

[0014] In combination with some embodiments of the first aspect, in some embodiments, when it is detected that the execution status satisfies a preset execution error condition, it is determined that the current order creation task has an execution error, the structure of the current cached data is locked, and based on the execution error keywords found in the operation log, it is determined whether the error type of the current order creation task execution error belongs to a preset retryable error type, specifically including: when it is detected that the execution status satisfies the preset execution error condition, the current order creation task is marked as a suspected error state, and the structure of the current cached data is locked; the preset execution error condition at least includes that the weighted average of the CPU usage rate and the memory occupancy rate exceeds the preset average value threshold; the operation log of the current order creation task is retrieved, the operation log is parsed, and the error type is determined by matching the operation error keywords corresponding to the preset error type in the operation log, and it is determined whether the error type belongs to the preset retryable error type.

[0015] In the above embodiment, by marking suspected errors when the order task execution status is abnormal, locking the cached data, parsing the operation log and determining the error type and retryability based on keyword matching, the cross-border e-commerce system can accurately locate the problem when an error occurs in order creation, thereby ensuring the stability and reliability of the order creation process and improving the efficiency of processing order creation tasks.

[0016] In combination with some embodiments of the first aspect, in some embodiments, during the retry process, the execution status of the current order creation task is monitored, and when no execution error of the current order creation task is detected, the current order creation task continues to be executed, specifically including: during the retry process, after waiting for the retry interval time, re-executing the current order creation task, and detecting whether the execution status of the current order creation task during the retry process meets the preset execution error condition; if the execution status of the current order creation task meets the preset execution error condition, wait for the retry interval time to retry again; if the execution status of the current order creation task does not meet the preset execution error condition after the retry is completed, continue to execute the current order creation task.

[0017] In the above embodiment, by following specific retry interval rules during retries, detecting the execution status and determining subsequent operations accordingly, the stability and fault tolerance of the system in dealing with errors are improved, and the risk of order creation interruption due to errors is reduced.

[0018] In combination with some embodiments of the first aspect, in some embodiments, when an execution error of the current order creation task is detected, a retry is performed again. When the number of retries reaches the maximum number of retries and the execution error of the current order creation task is still detected, the current order creation task is downgraded to obtain a downgraded task, and the downgraded task is executed. During the execution of the downgraded task, when no execution error of the downgraded task is detected, the current order creation task is marked as a downgraded successful task, and it also includes: when it is detected that the execution status of the downgraded task meets the preset execution error condition, the order commodity quantity data of each order creation task in the order creation task queue at that time is retrieved; all order creation tasks in the order creation task queue at that time are sorted from low to high according to the order commodity quantity data of each order creation task in the order creation task queue at that time, and each order creation task in the order creation task queue at that time is processed in turn.

[0019] In the above embodiment, when an error occurs in the execution of a downgraded task, the order product quantity data is retrieved and the task queue is sorted according to the order product quantity data, and the tasks are processed in sequence. This allows the cross-border e-commerce system to flexibly adjust the task execution order according to the product quantity when an exception occurs in the downgraded task processing, thereby improving the system's ability to respond to complex situations and the overall order processing success rate.

[0020] In a second aspect, an embodiment of the present application provides a high-performance scheduling system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the high-performance scheduling system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions. When the above-mentioned computer program product runs on a high-performance scheduling system, the above-mentioned high-performance scheduling system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions. When the instructions are executed on a high-performance scheduling system, the high-performance scheduling system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0023] It is understandable that the high-performance scheduling system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the high-performance scheduling system provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This application receives an order creation task request and executes the task in sequence and monitors the status. When an error occurs during the execution process, it locks the cached data, determines the error type and retryability, and retries according to the preset retry strategy when the cached data is healthy. If the retry fails, the preset average value threshold is adjusted for downgrade processing, thereby improving the system's ability to deal with errors and ensure business continuity and the efficiency of order creation task execution.

[0025] 2. This application trains a neural network model using training data from predicted order peak periods, obtains network and data synchronization data in real time to input the model to obtain predicted order peak periods, and proportionally adjusts the data collection frequency when approaching the peak period, so that the cross-border e-commerce system can predict and prepare in advance, thereby improving the timeliness of data processing and system stability when responding to order peaks.

[0026] 3. This application retrieves the order item quantity data and sorts the task queue according to the order item quantity data when an error occurs in the execution of the downgraded task, and processes the tasks in sequence. This enables the cross-border e-commerce system to flexibly adjust the task execution order according to the item quantity when an error occurs in the execution of the downgraded task, thereby improving the system's ability to respond to complex situations and the overall order processing success rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a structural diagram of an applicable system architecture of the high-performance scheduling method in the embodiment of the present application; Figure 2 This is a flowchart of a high-performance scheduling method in an embodiment of the present application; Figure 3 is another flowchart of the high-performance scheduling method in an embodiment of the present application; Figure 4 It is an exemplary hardware structure diagram of a high-performance scheduling system in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to and includes any or all possible combinations of one or more listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0030] Figure 1 It is a structural diagram of a system architecture to which the high-performance scheduling method in the embodiment of the present application can be applied.

[0031] See also Figure 1 The system includes a server cluster 110, a network device 120, a cache device 130 and a monitoring device 140.

[0032] The server cluster 110 mainly includes a computing server 111 and a storage server 112, which are the core computing and storage units of the high-performance scheduling system, responsible for executing order creation tasks, storing related data, and running model training, etc., to ensure the normal operation of order processing and data management. Among them, the computing server 111 is used to execute the processing of order creation tasks; the storage server 112 is mainly responsible for storing various data in the system, including order creation task queue data, training data, operation logs, etc.

[0033] The network equipment 120 mainly includes a switch 121 and a router 122, which are used to connect the internal servers of the high-performance scheduling system and communicate with the external network to ensure that the order creation task request and data can be stably and efficiently transmitted between different devices and different networks. Among them, the switch 121 is used to connect the server cluster and other hardware devices to ensure the fast and stable transmission of data between different devices; the router 122 is responsible for connecting the cross-border e-commerce system with external networks, such as the Internet or the partner's network, and can realize data forwarding and routing selection between different networks, ensuring that the system can receive order creation task requests from external systems, and can also send data in the system to the outside.

[0034] The cache device 130 is used to quickly store and read cache data during the order creation process, reduce data access latency, and assist in recovering data when a task error occurs, thereby improving system performance and data processing consistency.

[0035] The monitoring device 140 mainly includes a performance monitoring device 141 and a log collection device 142, which are used to monitor the performance indicators of the server in real time, collect operation logs, and provide data support for the system. Among them, the performance monitoring device 141 is used to monitor the key performance indicators of the server in real time, collect performance data at a certain frequency, and send the data to the system's monitoring software for analysis; the log collection device 142 is responsible for collecting the system's operation logs, including the execution logs of the order creation task, the model training logs, etc.

[0036] The present application provides a high-performance scheduling method, which executes tasks in sequence after receiving an order creation task request, generates and monitors cache data and task status during execution, locks the cache structure when an error occurs in task execution, determines that the error type can be retried and the cache is healthy, and retry according to the strategy. If the retry fails, the process is downgraded, thereby improving the order creation success rate and system stability, and improving the efficiency of order creation task execution.

[0037] The following describes a high-performance scheduling method in one embodiment of the present application: like Figure 2 FIG. 1 is a flow chart of a high-performance scheduling method provided in an embodiment of the present application, which can be applied to Figure 1 The system architecture shown includes the following steps: S201, after receiving an order creation task request, executing the order creation tasks in the task request in sequence; When the cross-border e-commerce system receives an order creation task request from a business module or an external system, the high-performance scheduling system will process the order creation tasks one by one in sequence.

[0038] S202: after the current order creation task starts to be executed, generate initial cache data of the order to be created, and monitor the execution status of the current order creation task; When the order creation task begins to execute, the system will open up a specific area in the memory to generate the initial cache data of the order creation task, which is used to temporarily store order-related information, such as product details, user information, etc.; at the same time, through a special monitoring program or thread, it continuously obtains and analyzes various indicators during the task execution process, including CPU utilization, memory usage, etc., to determine whether the task execution status is normal, ensuring that potential problems can be discovered in time and corresponding measures can be taken.

[0039] S203: When it is detected that the execution status meets the preset execution error condition, the structure of the current cache data is locked, and it is determined whether the error type of the current order creation task execution error belongs to the preset retryable error type; It can be understood that the current cache data is the cache data obtained after reading and writing the initial cache data during task execution.

[0040] When the system detects that the execution status reaches a preset error condition, such as the weighted average of CPU usage and memory usage exceeding the preset threshold, the error handling mechanism is immediately activated. First, the current cache data structure is locked to prevent further data modification or misoperation; then, the system will deeply analyze the operation log, extract the error keywords, and match them with the preset keyword library of retryable error types to accurately determine whether the error is within the retry range.

[0041] In some embodiments of the present application, the preset execution error condition is that the weighted average of the CPU usage and the memory occupancy exceeds the preset average threshold; in other embodiments of the present application, the preset execution error condition may also be that the network delay data value exceeds the preset network delay data threshold, etc., which is not limited here.

[0042] S204: In the case of a preset retryable error type, check whether the current cached data is in a healthy state; If the current cache data is not in a healthy state, the following step S205 is executed; If the current cache data is in a healthy state, the following step S206 is executed; After determining that the error type belongs to the preset retryable error type, the system will check the health status of the current cache data through data structure verification and data format comparison. First, compare the structure definition of the current cache data with the initial cache data, such as whether the number and type of data fields are consistent; secondly, check the data format field by field, such as whether the date format is standardized, whether the numerical precision is correct, etc. If there is no difference between the two in structure and format, the current cache data is judged to be in a healthy state, otherwise it is unhealthy. If it is judged that the current cache data is not in a healthy state, the cache data that was most recently in a healthy state in the system log is extracted to replace the current cache data; if it is judged that the current cache data is in a healthy state, the subsequent retry steps are continued.

[0043] In some embodiments of the present application, the health status is that the data structure and data format of the current cache data are consistent with the initial cache data; in other embodiments of the present application, the health status may also be that the core order information is complete and conflict-free, etc., which is not limited here.

[0044] S205: when the current cache data is not in the healthy state, extract the healthy cache data in the healthy state recorded in the system log of the current order creation task the most recent time to replace the current cache data; When the current cache data is determined to be unhealthy, the system will retrieve the system log of the current order creation task. In the log, based on the timestamp and health status mark, the cache data information recorded when it was in the most recent healthy state is accurately located and extracted, and the current erroneous cache data is completely replaced with the extracted healthy cache data using the data recovery or replacement mechanism.

[0045] S206, when the current cache data is in a healthy state, determine a retry strategy according to a preset retry rule corresponding to the error type, and start retrying; After confirming that the current cache data is healthy, the system will match the preset retry rules according to the error type. First, find the retry settings corresponding to the error type in the rule library to obtain the maximum number of retries and the retry interval; second, start the retry mechanism, use the scheduled task or a dedicated retry thread, and re-execute the order creation task according to the set interval. During the retry, continue to monitor the task execution status and count the number of retries.

[0046] S207. During the retry process, monitor the execution status of the current order creation task to see if there is an execution error; If there is no execution error in the execution status of the current order creation task, then execute the above step S202; If the execution status of the current order creation task is an execution error, the following step S208 is executed; During the retry process, the execution status of the order creation task is tracked in real time through the monitoring module. The module continuously collects key indicator data during task operation, such as whether the weighted average of the CPU usage and memory usage exceeds the preset average threshold. Once it is found that the preset average threshold is exceeded, it can be determined that the current order creation task has been executed incorrectly again. If there is no execution error in the execution status of the current order creation task, execute the above step S202; continue to execute the current order creation task, and monitor the execution status of the current order creation task; if there is an execution error in the execution status of the current order creation task, execute the following step S208 to determine whether the current number of retries has reached the maximum number of retries.

[0047] S208. In case of an execution error, determine whether the current number of retries has reached the maximum number of retries; If the current number of retries does not reach the maximum number of retries, the following step S209 is executed; If the current number of retries reaches the maximum number of retries, the following step S210 is executed; When an error in the execution of an order creation task is detected, a special counting mechanism will be built into the system to record the current number of retries. This counting mechanism will increase the value accordingly with each retry operation. At the same time, the system pre-stores the parameter of the maximum number of retries set for this error type, and then compares the current number of retries with the maximum number of retries to determine whether the maximum number of retries has been reached. If the current number of retries does not reach the maximum number of retries, execute the following step S209, retry again and monitor the task execution status during the retry process; if the current number of retries reaches the maximum number of retries, execute the following step S210, downgrade the task and then retry.

[0048] S209, retry again and execute the above step S207; When it is determined that another retry is required, the system waits for a preset retry interval time, restarts the execution process of the order creation task, and executes the above step S207 to monitor the execution status of the current order creation task during the retry process to see if there is an execution error.

[0049] S210, downgrading the current order creation task to obtain a downgraded task, and executing the downgraded task. When no execution error is detected during the execution of the downgraded task, marking the current order creation task as a downgraded successful task.

[0050] When the number of retries reaches the upper limit and an error still occurs, the system starts downgrading. First, the task-related parameters are adjusted, such as the preset average value threshold, to form a downgraded task, and the task is executed again. During this period, the monitoring module continuously monitors and determines whether an error occurs during the execution through the new parameters. If no error occurs during the execution, the system marks the original order creation task as downgraded successfully, so that the task can be completed with limited functions and the business process is not interrupted.

[0051] In some embodiments of the present application, the downgrade processing is to increase the preset average value threshold; in other embodiments of the present application, the downgrade processing may also be to simplify the data processing logic in the order creation task, skip some non-core business verification steps, etc., which is not limited here.

[0052] In the above embodiment, after receiving an order creation task request from a cross-border e-commerce system or an external system, tasks are executed in sequence, initial cache data is generated and the status is monitored during task execution; when the execution status meets a preset error condition, the error type is determined; when retry is possible and the cache data is healthy, retries are performed according to the strategy; if the cache data is unhealthy, a healthy cache is extracted instead; if the retry fails, the order creation task is downgraded to obtain a downgraded task and executed, so that the order creation task has a complete response mechanism when facing various error situations, thereby improving the success rate and stability of order creation in the cross-border e-commerce system and improving the efficiency of order creation task execution.

[0053] In some embodiments, some special situations may occur during the execution of order creation tasks. For example, during certain shopping peak periods, the order volume will increase significantly, and the corresponding number of order creation tasks will increase significantly. The high-performance scheduling system can adjust the data collection frequency in advance by predicting the order peak period. Figure 3 FIG. 1 is another flow chart of a high-performance scheduling method provided in an embodiment of the present application, which can be used to Figure 1 The high-performance scheduling system architecture shown includes the following steps: S301, after receiving an order creation task request, adding each order creation task in the task request into an order creation task queue; When an order creation task request is received, the system will first start the task parsing module, which will extract and analyze the data in the request, split and identify each independent order creation task; then, the system will call the task queue management program, which will open up a specific storage area in the memory to store these split order creation tasks, and put them into the order creation task queue in a first-in-first-out order so that they can be processed and scheduled in sequence later.

[0054] S302, training a preset reinforcement learning model using an order creation task processing sequence training data group to obtain an order creation task processing sequence model; It can be understood that the order creation task processing sequence model is used to output order creation task processing sequence data based on a set of input core processor load ratio data, remaining memory ratio data, order item quantity data of the order creation task and order creation task priority data; the order creation task processing sequence training data group includes multiple order creation task processing sequence training data, each training data includes a core processor load ratio data, a remaining memory ratio data, an order item quantity data of the order creation task, an order creation task priority data and a corresponding marked completed order creation task processing sequence data.

[0055] First, we sorted out a training data set covering the relationship between core processor load, remaining memory, number of order items, task priority data and the corresponding order creation task processing sequence data, and used various data such as core processor load ratio as input features, and order creation task processing sequence data as labeled output. These data are input into the preset reinforcement learning model, and the model continuously adjusts parameters based on the relationship between features and outputs. After multiple rounds of iterative training, a model that can generate processing sequence data based on input data is obtained.

[0056] S303, collect real-time data and input the order creation task processing sequence model in sequence to obtain each order creation task processing sequence data in the order creation task queue; First, the system's built-in data acquisition module is used to collect data on the core processor load, remaining memory, the number of items in each order creation task, and task priority in real time. Then, these data are sorted and input into the model in the input format required by the order creation task processing sequence model. The model uses the trained algorithm logic to analyze and calculate the input data, and then outputs the processing sequence data corresponding to each order creation task.

[0057] S304, adjusting the order of each order creation task in the order creation task queue according to the order creation task processing sequence data, and taking out the order creation tasks in sequence for execution; After the system obtains the processing sequence data output by the order creation task processing sequence model, it will traverse the order creation task queue. It will match each order creation task in the queue with the sequence data and rearrange the order of order creation tasks according to the order indicated. Then, it will take out the order creation tasks from the head of the queue in the new order to ensure that the order creation tasks are executed in the arranged order.

[0058] S305, after taking out a task from the order creation task queue each time, obtain the real-time order creation task queue and collect the core processor load ratio data and the remaining memory ratio data at that time; At the moment when the order creation task is taken out from the order creation task queue, the data collection mechanism is triggered. Through the system's performance monitoring module, the load ratio data of the core processor at the current moment is obtained, such as calculating the proportion of core processor resources occupied by the running process; at the same time, the information in the memory management unit is read to determine the remaining memory ratio data at that time.

[0059] S306, retrieve the required data and input them into the task processing sequence model in sequence, obtain the processing sequence data of each order creation task in the real-time order creation task queue, and adjust the sequence of each task in the real-time task queue; First, retrieve the core processor load ratio data, remaining memory ratio data, and the order quantity data and task priority data of the order creation task itself; then, input these data in sequence according to the input requirements of the task processing sequence model. Based on the logic and algorithm obtained from its internal training, the model analyzes and processes the input data and outputs the processing sequence data of each order creation task; finally, based on these data, change the arrangement order of each order creation task in the real-time task queue.

[0060] S307, training a preset neural network model using the order peak period prediction training data group to obtain an order peak period prediction model; It can be understood that the order peak prediction model is used to output predicted order peak data based on a set of input network status data and data synchronization status data; the predicted order peak training data group includes multiple predicted order peak training data, and each training data includes a corresponding relationship between a network status data, a data synchronization status data and a corresponding marked completed predicted order peak data.

[0061] First, a large amount of historical data related to the order peak period, such as network status data and data synchronization data, is collected and organized into a training data set for predicting the order peak period. The data in this data set is used as input to the preset neural network model. The model continuously adjusts the connection weights and thresholds between neurons through learning and analyzing the data features. After multiple rounds of iterative training, the model can accurately predict the order peak period based on the input network status data and data synchronization data, thereby obtaining a model for predicting the order peak period.

[0062] S308, real-time network status data and real-time data synchronization data are acquired in real time, and input into the order peak prediction model to obtain the order peak prediction data and obtain the order peak prediction time period; With the help of network monitoring tools and data synchronization monitoring modules, the system continuously captures the current real-time network status data and real-time data synchronization status data, inputs these data into the trained order peak prediction model, obtains the predicted data of future order peak periods, and further determines the predicted order peak period time period.

[0063] S309, when the time difference between the real time and the predicted order peak period is less than the preset time difference range, the data collection frequency is adjusted by a multiple of the predicted peak duration ratio value; It can be understood that the predicted peak duration ratio value is the ratio of the predicted order peak period time period to the preset peak period time period.

[0064] The system first compares the real-time time with the predicted order peak time, calculates the difference between the two, and compares it with the preset time difference range. When the difference is less than the range, the predicted peak duration ratio is obtained. The multiple relationship is determined based on this ratio, and then the original data collection frequency is multiplied by the multiple by modifying the parameter settings of the data collection module, such as the trigger interval of the timer, so as to speed up the data collection speed when the order peak is approaching, and make better preparations for the peak.

[0065] S310, after the current order creation task starts to execute, generate initial cache data of the order to be created, and monitor the execution status of the current order creation task; S311. When it is detected that the execution status meets the preset execution error condition, the current order creation task is marked as a suspected error state, and the structure of the current cache data is locked; It is understandable that the preset execution error condition at least includes that the weighted average value of the CPU usage rate and the memory occupancy rate exceeds the preset average value threshold.

[0066] When the system monitors the execution status of the order creation task in real time, it will compare various operating indicators with the preset execution error conditions, such as detecting whether the CPU usage rate, memory usage, etc. exceed the established thresholds. Once these preset conditions are found to be met, the system will mark the current order creation task as a suspected error state and trigger the data protection mechanism to prevent the modification of the current cache data structure through specific code logic, thereby locking its structure.

[0067] S312, calling the running log of the current order creation task, determining the error type by searching the running error keyword corresponding to the preset error type in the running log for matching, and judging whether the error type belongs to the preset retryable error type; After marking the current order creation task as a suspected error state, the system will call the log retrieval module to obtain the operation log corresponding to the task, and search for operation error keywords associated with the preset error type in the log content, such as specific error codes, exception description statements, etc. The specific error type is determined based on the matched keywords, and then compared with the preset retryable error type list to determine whether the error can be retried.

[0068] S313: In the case of a preset retryable error type, check whether the current cached data is in a healthy state; If the current cache data is not in a healthy state, the following step S314 is executed; If the current cache data is in a healthy state, the following step S315 is executed; S314, when the current cache data is not in the healthy state, extract the healthy cache data in the healthy state recorded in the system log of the current order creation task the most recent time to replace the current cache data; S315, when the current cache data is in a healthy state, determine a retry strategy according to a preset retry rule corresponding to the error type, and start retrying; S316. During the retry process, monitor the execution status of the current order creation task to see if there is an execution error; If there is no execution error in the execution status of the current order creation task, then execute the above step S310; If the execution status of the current order creation task is an execution error, execute the following step S317; S317, whether the number of retries reaches the maximum number of retries; If the current number of retries does not reach the maximum number of retries, the following step S318 is executed; If the current number of retries reaches the maximum number of retries, the following step S319 is executed; S318, if the current number of retries does not reach the maximum number of retries, retry again and execute the above step S316; S319, downgrading the current order creation task to obtain a downgraded task, and executing the downgraded task. When no execution error is detected during the execution of the downgraded task, marking the current order creation task as a downgraded successful task; S320: when it is detected that the execution status of the downgraded task meets the preset execution error condition, the order commodity quantity data of each order creation task in the order creation task queue at that time is retrieved; When the system continuously monitors the execution status of the downgraded task, it will collect the CPU usage and memory usage in real time, and calculate the weighted average of the CPU usage and memory usage and compare them with the preset average threshold. Once the weighted average is found to exceed the preset average threshold, the downgraded task execution error occurs. At this time, the system will start the data retrieval mechanism, and accurately locate and extract the order item quantity data of each order creation task by accessing the database or cache area that stores information related to the order creation task queue.

[0069] S321. Sort all order creation tasks in the current order creation task queue from low to high according to the order commodity quantity data of each order creation task in the current order creation task queue, and process each order creation task in the current order creation task queue in sequence.

[0070] After obtaining the order item quantity data of each order creation task in the order creation task queue at that time, the system calls the sorting algorithm, compares all the tasks in the queue based on the data, and rearranges the order from low to high by quantity; then, through the task scheduling module, tasks are taken out one by one from the head of the queue according to the newly arranged order, and processing operations are carried out on each order creation task in turn.

[0071] Steps S310, S313-S319 and Figure 2 Steps S202 - S210 in the illustrated embodiment are similar, and the descriptions of steps S202 - S210 may be referred to, and will not be repeated here.

[0072] In the above embodiment, the high-performance scheduling method provided by the present application first puts the task into the queue after receiving the order creation task request of the cross-border e-commerce system, and uses the reinforcement learning model to determine the task processing order and execute it according to the core processor load, remaining memory, number of goods and task priority data, and generates initial cache data and monitors the status during execution; when an execution error occurs, the error type and retryability are determined by locking the cache data and parsing the operation log, and the cache data is retried according to the strategy when it is retriable and the cache data is healthy. If the cache data is unhealthy, the most recent healthy cache data is extracted to replace it; if the retry fails, the preset average value threshold is increased for downgrade processing, and the order peak period is predicted based on the network status and data synchronization status through the neural network model to adjust the data collection frequency, and the order creation task is sorted by the number of goods when the downgraded task fails again. The cross-border e-commerce system improves the intelligence and flexibility of task processing during the order creation process, enhances the ability to deal with errors and peak periods, effectively improves the overall performance, stability and order processing efficiency of the system, reduces the risk of business interruption and order loss caused by errors, and improves the efficiency of order creation task execution as a whole.

[0073] The following introduces an exemplary high-performance scheduling system 400 provided in an embodiment of the present application. Figure 4 4 is a schematic diagram of an exemplary hardware structure of a high-performance scheduling system 400 provided in an embodiment of the present application.

[0074] In some embodiments, the high-performance scheduling system 400 is a computer device or the high-performance scheduling system 400 includes a computer device. The computer device includes a processor, a memory and a network interface connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with other external terminals or servers through a network connection. In some embodiments, the network interface can be a wired network interface, and in some embodiments, the network interface can also be a wireless network interface. When the computer program is executed by the processor, the method in the embodiment of the present application is implemented.

[0075] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0076] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0077] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0078] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk), etc.

[0079] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A high-performance scheduling method for a cross-border e-commerce system, characterized in that: include: After receiving an order creation task request from a business module of a cross-border e-commerce system or an external system, the order creation tasks in the task request are executed in sequence; After the current order creation task starts to execute, initial cache data of the order to be created is generated, and the execution status of the current order creation task is monitored; When it is detected that the execution status meets the preset execution error condition, it is determined that the current order creation task has an execution error, the structure of the current cache data is locked, and based on the operation error keywords found in the operation log, it is determined whether the error type of the current order creation task execution error belongs to the preset retryable error type; The current cache data is the cache data obtained after reading and writing the initial cache data during task execution; The preset execution error condition at least includes that the weighted average of the CPU usage and the memory usage exceeds the preset average threshold; In the case of a preset retryable error type, checking whether the current cache data is in a healthy state; the healthy state is that the data structure and data format of the current cache data are consistent with those of the initial cache data; When the current cache data is not in the healthy state, extracting the most recent healthy cache data in the healthy state recorded in the system log of the current order creation task, and using the healthy cache data to replace the current cache data; When the current cache data is in the healthy state, a retry strategy is determined according to a preset retry rule corresponding to the error type, and a retry is started; the retry strategy includes a maximum number of retries and a retry interval; During the retry process, monitoring the execution status of the current order creation task, and when no execution error of the current order creation task is detected, continuing to execute the current order creation task; When an execution error of the current order creation task is detected, a retry is performed again. When the number of retries reaches the maximum number of retries and an execution error of the current order creation task is still detected, the current order creation task is downgraded to obtain a downgraded task, and the downgraded task is executed. During the execution of the downgraded task, when no execution error of the downgraded task is detected, the current order creation task is marked as a successfully downgraded task; the downgrade processing is to adjust the preset average value threshold.

2. The method according to claim 1, characterized in that After receiving an order creation task request from a business module of a cross-border e-commerce system or an external system, the order creation tasks in the task request are executed in sequence, specifically including: After receiving an order creation task request from a business module of a cross-border e-commerce system or an external system, each order creation task in the order creation task request is included in an order creation task queue; The preset reinforcement learning model is trained using the order creation task processing sequence training data group to obtain an order creation task processing sequence model; the order creation task processing sequence model is used to output order creation task processing sequence data based on a set of core processor load ratio data, remaining memory ratio data, order commodity quantity data of the order creation task and the order creation task input; the order creation task processing sequence training data group includes a plurality of order creation task processing sequence training data, each training data includes a core processor load ratio data, a remaining memory ratio data, an order commodity quantity data of the order creation task, an order creation task and a corresponding relationship between a corresponding marked completed order creation task processing sequence data; Collect real-time core processor load ratio data, real-time remaining memory ratio data, order commodity quantity data of each order creation task in the order creation task queue, and order creation tasks, retrieve the real-time core processor load ratio data, real-time remaining memory ratio data, order commodity quantity data of each order creation task in the order creation task queue, and order creation tasks, and input them into the order creation task processing sequence model in sequence to obtain the processing sequence data of each order creation task in the order creation task queue; The order of each order creation task in the order creation task queue is adjusted according to the order creation task processing sequence data, and the order creation tasks are taken out from the order creation task queue in sequence and executed.

3. The method according to claim 2, characterized in that After adjusting the order of each order creation task in the order creation task queue according to the order creation task processing sequence data, and sequentially taking out the order creation tasks from the order creation task queue and executing them, the method further includes: After taking out a task from the order creation task queue each time, a real-time order creation task queue is obtained and the core processor load ratio data and the remaining memory ratio data at that time are collected; Retrieve the core processor load ratio data at that time, the remaining memory ratio data at that time, the order commodity quantity data of each order creation task in the real-time order creation task queue, and the order creation task, and input them into the task processing sequence model in sequence to obtain the processing sequence data of each order creation task in the real-time order creation task queue, and adjust the order of each task in the real-time task queue according to the processing sequence data of each order creation task in the real-time order creation task queue.

4. The method according to claim 1, characterized in that: After receiving an order creation task request from a business module of a cross-border e-commerce system or an external system, and sequentially executing the order creation tasks in the task request, the method further includes: The preset neural network model is trained using the predicted order peak period training data group to obtain a predicted order peak period model; the predicted order peak period model is used to output predicted order peak period data based on a set of input network status data and data synchronization status data; the predicted order peak period training data group includes a plurality of predicted order peak period training data, each training data includes a corresponding relationship between a network status data, a data synchronization status data and a corresponding marked completed predicted order peak period data; Real-time network status data and real-time data synchronization status data are acquired in real time, and input into the predicted order peak period model to obtain predicted order peak period data and obtain predicted order peak period time period; When the time difference between the real time and the predicted order peak period is less than the preset time difference range, the data collection frequency is adjusted by the predicted peak duration ratio value as a multiple; the predicted peak duration ratio value is the ratio of the duration of the predicted order peak period time period to the preset peak duration.

5. The method according to claim 1, characterized in that When it is detected that the execution status meets the preset execution error condition, the current order creation task execution error is determined, the structure of the current cache data is locked, and based on the operation error keyword found in the operation log, it is determined whether the error type of the current order creation task execution error belongs to the preset retryable error type, specifically including: When it is detected that the execution status meets the preset execution error condition, the current order creation task is marked as a suspected error state, and the structure of the current cache data is locked; the preset execution error condition at least includes that the weighted average of the CPU usage rate and the memory occupancy exceeds the preset average value threshold; Retrieve the operation log of the current order creation task, parse the operation log, and determine the error type by searching for operation error keywords corresponding to preset error types in the operation log, and judge whether the error type belongs to a preset retryable error type.

6. The method according to claim 1, characterized in that During the retry process, monitoring the execution status of the current order creation task, and continuing to execute the current order creation task when no execution error of the current order creation task is detected, specifically includes: During the retry process, after waiting for the retry interval time, the current order creation task is re-executed, and the execution status of the current order creation task during the retry process is detected to determine whether it meets the preset execution error condition; When the execution status of the current order creation task meets the preset execution error condition, retry again after waiting for the retry interval time; If the execution status of the current order creation task does not meet the preset execution error condition after the retry is completed, continue to execute the current order creation task.

7. The method according to claim 1, characterized in that When an execution error of the current order creation task is detected, a retry is performed again. When the number of retries reaches the maximum number of retries and an execution error of the current order creation task is still detected, the current order creation task is downgraded to obtain a downgraded task, and the downgraded task is executed. During the execution of the downgraded task, when no execution error of the downgraded task is detected, the current order creation task is marked as a downgraded successful task, and the method further includes: When it is detected that the execution status of the downgraded task meets the preset execution error condition, the order commodity quantity data of each order creation task in the order creation task queue at that time is retrieved; All order creation tasks in the current order creation task queue are sorted from low to high according to the order commodity quantity data of each order creation task in the current order creation task queue, and each order creation task in the current order creation task queue is processed in sequence.

8. A high-performance scheduling system, characterized in that: The high-performance scheduling system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the high-performance scheduling system to execute the method described in any one of claims 1-7.

9. A computer program product comprising instructions, characterized in that When the computer program product is run on a high-performance scheduling system, the high-performance scheduling system is caused to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on a high-performance scheduling system, the high-performance scheduling system is caused to execute the method according to any one of claims 1 to 7.

Images