Task processing method and device, computer equipment, medium and product

Abstract

The invention discloses a task processing method and device, computer equipment, a medium and a product. The method comprises the steps of determining predicted completion time of a target task through a first prediction model; if the predicted completion time of the target task is later than the configuration completion time of the target task, determining an abnormal task link and a normal task link from task links related to the target task; determining an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link and a resource adjustment strategy corresponding to the target task through a second prediction model; and according to the resource adjustment strategy, adjusting the allocation resource of the upstream task of the target task in the abnormal task link, the allocation resource of the upstream task of the target task in the normal task link and the allocation resource of the target task. Through the method, the resources of the target task can be dynamically adjusted, and the execution success rate of the task is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a task processing method, a task processing apparatus, a computer device, a computer-readable storage medium, and a computer program product. Background Art

[0002] In the big data scenario, the scheduling system is a relatively complex system, involving a wide range of content, targeting a variety of scenarios, and implementing solutions that vary greatly. A mature, easy-to-use, manage and maintain scheduling system needs to be connected to a large number of peripheral components, including not only various storage and computing frameworks, but also various services or transactions such as permission control, load flow control, management alarms, and quality analysis.

[0003] In the process of scheduling system executing scheduling tasks, how to quickly discover the execution status of scheduling tasks and reasonably allocate resources is a major research hotspot. At present, the management of scheduling tasks and resource allocation are directly carried out through historical data, which cannot cope with emergencies and leads to low efficiency of task execution.

[0004] Therefore, how to improve the execution efficiency of scheduling tasks has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The embodiments of the present application provide a task processing method, apparatus, computer equipment, medium, and product, which can dynamically adjust the resources of the target task and improve the success rate of task execution by predicting the target task running data.

[0006] On the one hand, an embodiment of the present application discloses a task processing method, the method comprising:

[0007] Determine the predicted completion time of the target task by using the first prediction model;

[0008] If the predicted completion time of the target task is later than the configured completion time of the target task, determining an abnormal task link and a normal task link from task links related to the target task;

[0009] Determine, by a second prediction model, an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link, and a resource adjustment strategy corresponding to the target task;

[0010] The allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted according to the resource adjustment strategy.

[0011] The present application discloses a task processing device, which includes:

[0012] A determination unit, configured to determine a predicted completion time of a target task by using a first prediction model;

[0013] The determining unit is further configured to determine an abnormal task link and a normal task link from task links related to the target task if the predicted completion time of the target task is later than the configured completion time of the target task;

[0014] The determining unit is further configured to determine, by means of a second prediction model, an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link, and a resource adjustment strategy corresponding to the target task;

[0015] A processing unit is used to adjust the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task according to the resource adjustment strategy.

[0016] On the one hand, an embodiment of the present application discloses a computer device, which includes a processor suitable for implementing one or more computer programs; and a computer-readable storage medium, wherein the computer-readable storage medium stores one or more computer programs, and the one or more computer programs are suitable for being loaded by the processor and executed in accordance with the above-mentioned task processing method.

[0017] In one aspect, the present application discloses a computer-readable storage medium storing one or more computer programs, wherein the one or more computer programs are suitable for being loaded by a processor and executing the above-mentioned task processing method.

[0018] The present application discloses a computer program product, which includes a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device performs the above-mentioned task processing method.

[0019] In an embodiment of the present application, in the current operation cycle of the target task, the predicted completion time of the target task is determined by the first prediction model. The predicted completion time of the target task in the current operation cycle can be predicted in advance by the prediction model. If the predicted completion time of the target task is later than the configuration completion time of the target task, the abnormal task link and the normal task link are determined from the task link related to the target task. The upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task are determined by the second prediction model. When the predicted completion time of the target task is later than the configuration completion time, the task link that affects the target task can be determined, and the resources of each task are adjusted in time based on the prediction model. According to the resource adjustment strategy, the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted so that the predicted completion time of the target task is not later than the configuration completion time of the target task. In summary, the task processing method provided in the embodiment of the present application can, on the one hand, predict the operating data of the target task through a prediction model; on the other hand, when the predicted completion time of the target task is later than the configured completion time, the prediction model can be used to dynamically adjust the resources of the target task to improve the operation success rate of the target task. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in 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.

[0021] Figure 1 It is a schematic diagram of a network architecture of a task processing system disclosed in an embodiment of the present application;

[0022] Figure 2 It is a flowchart of a task processing method disclosed in an embodiment of the present application;

[0023] Figure 3 It is a schematic diagram of a framework for determining predicted operation data of a target task using a first prediction model disclosed in an embodiment of the present application;

[0024] Figure 4 It is a task link monitoring schematic diagram disclosed in an embodiment of the present application;

[0025] Figure 5 It is a schematic diagram of a framework for determining a resource adjustment strategy using a second prediction model disclosed in an embodiment of the present application;

[0026] Figure 6 It is a schematic diagram of task link resource reallocation disclosed in an embodiment of the present application;

[0027] Figure 7 It is a schematic diagram of a task system disclosed in an embodiment of the present application in which the remaining resource amount does not satisfy the resource adjustment amount corresponding to the resource adjustment policy;

[0028] Figure 8 This is a schematic diagram of the correspondence between tasks, business domains, and management personnel disclosed in an embodiment of the present application;

[0029] Fig. 9 It is a flowchart of a training method of a first prediction model disclosed in an embodiment of the present application;

[0030] Fig.10 It is a flowchart of a training method for a second prediction model disclosed in an embodiment of the present application;

[0031] Fig.11 It is a flowchart of another task processing method disclosed in an embodiment of the present application;

[0032] Fig.12 is a structural schematic diagram of a task processing device disclosed in an embodiment of the present application;

[0033] Fig.13 It is a structural schematic diagram of a computer device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only 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 this application.

[0035] Before describing the embodiments of the present application, the following terms related to the task processing method are explained:

[0036] Scheduling system: In the embodiment of the present application, it refers to the scheduled task scheduling system of big data. Whether the current task is scheduled to run is determined by multiple conditions such as the expected scheduling running time of the task, the running status of the upstream dependent tasks, and scheduling resources.

[0037] Task dependency: If the data required to run the current task B is generated by another task A, then there is an upstream and downstream dependency relationship between the two tasks, which is recorded as task B depends on task A, or task A is the upstream task of task B. Task dependency includes strong dependency and weak dependency. Strong dependency: When the upstream task A runs successfully, it is a necessary condition for the downstream task B to be scheduled and run. Weak dependency: The successful running of the upstream task A is not a necessary condition for the scheduling and running of the downstream task B.

[0038] Scheduling task monitoring system: monitors various indicators of the scheduling system, including system operation indicators, task operation indicators, etc., and sends alarms to relevant recipients for abnormal or risky events.

[0039] Task link: Starting from a certain task, connect it with the upstream tasks according to the dependency relationship until the top node. The path between the top node and the current task node is the task link.

[0040] AI (Artificial Intelligence): Artificial Intelligence. In the embodiments of the present application, the first prediction model and the second prediction model involved are both AI models. The AI ​​model is one of the core technologies in the field of artificial intelligence. It solves complex problems by simulating human intelligent behavior and thinking processes. The AI ​​model can process and analyze large amounts of data, provide intelligent solutions, and is widely used in various fields, such as natural language processing, computer vision, machine learning, etc.

[0041] In the big data scenario, the scheduling system is a relatively complex system, involving a wide range of content, targeting a variety of scenarios, and implementing solutions that vary greatly. A mature, easy-to-use, manage and maintain scheduling system needs to be connected to a large number of peripheral components, including not only various storage and computing frameworks, but also various services or transactions such as permission control, load flow control, management alarms, and quality analysis.

[0042] In the process of scheduling tasks being executed by the scheduling system, how to quickly discover the execution status of the scheduling tasks and reasonably allocate resources is a major research hotspot. At present, the management of scheduling tasks and resource allocation are directly carried out through historical data. We generally cannot know in advance how long a task actually takes to run and how much resources it consumes. Only after the task is completed can we get the running time and resource consumption data from the running indicators. In this way, the resources of the scheduling tasks cannot be managed in real time, and emergencies cannot be dealt with, resulting in low efficiency of task execution and the risk of failure to execute successfully.

[0043] Based on this, an embodiment of the present application provides a task processing method, which includes: ① determining the predicted completion time of the target task through a first prediction model. ② If the predicted completion time of the target task is later than the configured completion time of the target task, then determining the abnormal task link and the normal task link from the task link related to the target task. ③ Determining the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through the second prediction model. When the predicted completion time of the target task is later than the configured completion time, the task link that affects the target task can be determined, and the resources of each task can be adjusted in time based on the prediction model. ④ According to the resource adjustment strategy, the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted, so that the predicted completion time of the target task is not later than the configured completion time of the target task. This method can predict the running data of the target task, find problems in time, and dynamically adjust the resources of the target task through the prediction model when the predicted completion time of the target task is later than the configured completion time, so as to improve the running success rate of the target task.

[0044] See also Figure 1 , is a network architecture diagram of a task processing system disclosed in an embodiment of the present application, and the task processing system may include a first terminal device 101, a second terminal device 102, and a task processing service 103. The first terminal device 101 and the task processing service 103 are connected via a network, and the second terminal device 102 and the task processing service 103 are connected via a network, for example, via a local area network, a wide area network, a mobile Internet, etc.

[0045] Among them, the first terminal device 101 is the terminal device corresponding to the task initiator, and the task initiator can be a research and development personnel, who is mainly responsible for initiating a task monitoring request to the task processing service 103. The second terminal device 102 is the terminal device corresponding to the manager, and the manager's main responsibility is to manage the tasks in the task system. In an embodiment of the present application, when the task system issues an alarm message for the task, the manager receives the alarm message and performs corresponding processing based on the alarm message. The first terminal device 101 and the second terminal device 102 in the embodiment of the present application include but are not limited to: smart phones, tablet computers, smart wearable devices, smart voice interaction devices, smart home appliances, personal computers, vehicle-mounted terminals, smart cameras, virtual reality devices (such as AR (Augmented Reality, augmented reality) devices) and other devices, and this application does not limit this.

[0046] The task processing server 103 refers to the task processing system, which is mainly responsible for receiving the monitoring request sent by the first terminal device 101 and monitoring the corresponding tasks. In the embodiment of the present application, the task processing server mainly includes predicting the running data of the target task and adjusting the resources. The task processing server 103 in the embodiment of the present application can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. The embodiment of the present application does not limit this.

[0047] In the embodiment of the present application, in combination with the task processing system, the task processing method provided by the embodiment of the present application can generally include the following steps: in the current operation cycle of the target task, the task processing server 103 determines the predicted completion time of the target task through the first prediction model. The predicted completion time of the target task in the current operation cycle can be predicted in advance through the first prediction model. If the predicted completion time of the target task is later than the configuration completion time of the target task, the abnormal task link and the normal task link are determined from the task link related to the target task, and the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task are determined through the second prediction model. When the predicted completion time of the target task is later than the configuration completion time, the task link that affects the target task can be determined, and the resources of each task are adjusted in time based on the second prediction model. According to the resource adjustment strategy, the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted so that the predicted completion time of the target task is not later than the configuration completion time of the target task. In summary, the task processing method provided in the embodiment of the present application can, on the one hand, predict the operating data of the target task through a prediction model; on the other hand, when the predicted completion time of the target task is later than the configured completion time, the prediction model can be used to dynamically adjust the resources of the target task, thereby improving the operation success rate of the target task.

[0048] In actual business scenarios, combined with the task processing system, the process of the task processing method includes: the first terminal device 101 sends a task monitoring request to the task processing server 103, the task processing server 103 monitors the target task in the task monitoring request, and uses the first prediction model to predict the operation data of the target task. When the predicted completion time of the target task is later than the configured completion time, the second prediction model is used to adjust the resources of the target task. If the remaining resources of the task system do not meet the resource adjustment strategy, the target task is determined to be abnormal, and the alarm information of the target task is generated, and the alarm information is sent to the second terminal device 102, so that the management personnel corresponding to the second terminal device 102 can adjust the configuration completion time of the target task based on the alarm information, or adjust the remaining resources of the task system, and the task system continues to monitor the target task until the target task is successfully executed.

[0049] It should be noted that the collection and processing of relevant data in this application should be strictly in accordance with the requirements of relevant laws and regulations when applied in examples, and the informed consent or separate consent of the personal information subject should be obtained, and subsequent data use and processing should be carried out within the scope of authorization of laws and regulations and the personal information subject.

[0050] The task processing method proposed in the embodiment of the present application is introduced in detail below.

[0051] See also Figure 2 , is a flowchart of a task processing method disclosed in an embodiment of the present application. The task processing method can be executed by a computer device, and the computer device can be Figure 1 The task processing server 103 shown, correspondingly, the task processing method may specifically include but is not limited to the following steps:

[0052] S201: Determine a predicted completion time of a target task using a first prediction model.

[0053] Among them, the target task is any task to be monitored in the task system, and the target task can be set by the R&D personnel. During the operation cycle, the predicted completion time of the target task is determined by the first prediction model, wherein the operation cycle can be days, weeks or hours, and the operation cycle is also set by the R&D personnel. For different tasks, the operation cycle may be different. The task system specifically refers to the scheduling task monitoring system, which is responsible for monitoring various indicators of each scheduling task in the scheduling system, including system operation indicators, task operation indicators, etc., and sending alarms to relevant recipients for abnormal or risky events.

[0054] In one possible implementation, after the target task is determined, before the current operation cycle of the target task ends, the task is cyclically monitored to monitor whether the target task can be executed and completed within the configured completion time. Specifically, a predictive analysis is performed through a model, that is, the predicted completion time of the target task is determined by calling a first prediction model. In addition to the predicted completion time, the first prediction model can also simultaneously predict the predicted end time, predicted running time, and predicted allocated resource amount of the target task. Among them, the predicted completion time of the target task, the predicted end time of the target task, the predicted running time, and the predicted allocated resource amount can be collectively referred to as the predicted operation data of the target task. After determining the predicted operation data, possible problems in the task system can be discovered in a timely manner based on the predicted operation data, such as discovering that the predicted completion time of the target task is later than the configured completion time, and solving the problem in a timely manner to ensure the smooth completion of the target task.

[0055] The specific process of determining the predicted completion time of the target task by using the first prediction model may include:

[0056] 1. Obtain the actual operation data of the target task in multiple operation cycles before the current operation cycle. The actual operation data includes one or more of the start time, end time, running time and allocated resource amount of the target task in these multiple operation cycles, that is, the historical operation data in multiple operation cycles. For example, the current operation cycle is April 1, 2024. The multiple operation cycles before the current operation cycle refer to the multiple days before April 1, 2024. This number of days can be customized according to the R&D personnel. It can be a week before April 1, 2024 (multiple operation cycles refer to 7 operation cycles), or it can be a month before April 1, 2024 (multiple operation cycles refer to 30 operation cycles). There are as many start times, end times, running times and allocated resource amounts as there are operation cycles.

[0057] 2. Obtain the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system. Among them, the task execution status includes not started, executing, and completed; the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status. If the task execution status of the upstream task is completed, the corresponding running data includes the actual start time, actual end time and actual running time of the upstream task; if the task execution status of the upstream task is executing, the corresponding running data includes the actual start time, predicted end time and predicted running time of the upstream task; if the task execution status of the upstream task is not started, the corresponding running data includes the predicted start time, predicted end time and predicted running time of the upstream task. The expected running time of the target task is the time set by the R&D personnel, which can be one hour, or 40 minutes, etc. The remaining resources of the task system are to first obtain the resource saturation of the task system, and then analyze the resource saturation to determine the remaining resources of the task system. In the task system, the upstream task of the target task refers to a task whose execution time is before the execution time of the target task and is associated with the target task. The completion of the target task depends on the upstream task.

[0058] 3. Input the actual operation data of the target task, the task status of the target task, the expected operation time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system into the first prediction model for prediction processing to obtain the predicted operation data of the target task. Before inputting the actual operation data into the first prediction model, the actual operation data needs to be preprocessed. The preprocessing mainly involves detecting abnormal data on the actual operation data and deleting the abnormal data.

[0059] See also Figure 3 , is a framework diagram of a method of determining predicted running data of a target task using a first prediction model disclosed in an embodiment of the present application, wherein it can be seen that the first prediction model includes two inputs, namely, the actual running data of the target task and the current data of the target task, the actual running data of the target task includes one or more of the start time of the target task, the end time of the target task, the running time of the target task, and the allocated resources of the target task in multiple running cycles before the current running cycle, the current data of the target task includes the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream tasks of the target task, and the remaining resources of the task system; the output is the predicted running data of the target task, mainly including the predicted start time of the target task, the predicted end time of the target task, the predicted running time of the target task, and the predicted allocated resources of the target task.

[0060] For any task in the task system, the predicted operation data of these tasks can be determined through the first prediction model. During the operation cycle, these predicted operation data are analyzed, problems are discovered in time, and the problems are handled to ensure the successful execution of the target tasks.

[0061] S202: If the predicted completion time of the target task is later than the configured completion time of the target task, an abnormal task link and a normal task link are determined from task links related to the target task.

[0062] In a possible implementation, after predicting the predicted completion time of the target task, the predicted completion time of the target task is compared with the configured completion time of the target task. If the predicted completion time of the target task is not later than the configured completion time of the target task, the target task is continuously monitored, that is, the predicted operation data of the target task is determined cyclically. Once a problem is found, it is promptly processed to ensure that the target task can be successfully executed. Among them, the configured completion time of the target task is defined by the developer, stored in a local or external database, and can be directly obtained during the operation cycle of the target task.

[0063] In another possible implementation, if the predicted completion time of the target task is later than the configured completion time of the target task, it indicates that the target task cannot be successfully executed. Therefore, the present application can dynamically adjust the resources in the task system in this case to ensure that the target task can be successfully executed. First, before adjusting the resources, it is necessary to determine the abnormal task links and normal task links from the task links related to the target task included in the task system, and then reallocate the resources of the task system based on the abnormal task links and the normal task links to ensure the successful execution of the target task.

[0064] Among them, the specific process of determining abnormal task links and normal task links from the task links related to the target task included in the task system is as follows: obtain the predicted completion time of the first upstream task of the target task in each task link related to the target task included in the task system, the first upstream task being the upstream task adjacent to the target task; for each task link, determine the time interval between the predicted completion time of the first upstream task and the predicted start time of the target task; determine the task link in each task link whose time interval is less than the time interval threshold as an abnormal task link associated with the target task; because the predicted end time of the first upstream task is too close to the predicted start time of the target task, the target task can only be executed after the first upstream task is completed, and therefore, the predicted start time of the target task is later, so that the predicted completion time of the target task is also postponed, exceeding the configured completion time, thereby affecting the successful execution of the target task. The task links in each task link whose time interval is greater than or equal to the time interval threshold are determined as normal task links associated with the target task. The predicted end time of the first upstream task in the normal task link is far away from the predicted start time of the target task. After the completion of this first upstream task, there is a long time interval that can meet the timely start of the target task, and therefore will not affect the successful execution of the target task.

[0065] See also Figure 4 , is a task link monitoring schematic diagram disclosed in an embodiment of the present application, from Figure 4 It can be seen that the target task is task M, and the task links associated with the target task are task link 1 and task link 2, where task link 1 includes task A, task B, and task C (task A, task B, and task C are all upstream tasks of task M), and task link 2 includes task H, task I, and task J (task H, task I, and task J are also upstream tasks of task M). Figure 4It can be seen from the time displayed in that the predicted end time of task M (03:30) is later than the latest completion time allowed by the business (3:00), so it is necessary to determine the abnormal task link and normal task link of task M. According to the displayed time, the time interval between the predicted end time of task C and the predicted start time of task M is 1 hour, and the time interval between the predicted end time of task J and the predicted start time of task M is 10 minutes. Assuming that the time interval threshold is 30 minutes, it can be determined that the task link 2 corresponding to task J is an abnormal task link, and the task link 1 corresponding to task C is a normal task link. Furthermore, it can be analyzed that task J is affected by task I. If the running time of task I and task J can be shortened, then the corresponding start time of task M can be earlier, so as not to affect the execution of task M. Based on this idea, since the task execution status of both task C and task M is not started, the resources allocated to task I and task J can be adjusted through the prediction model to shorten the running time. The specific steps can be referred to step S203.

[0066] S203: Determine the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through the second prediction model.

[0067] In one possible implementation, after determining the normal task link and the abnormal task link of the target task, the prediction model can be used to infer how many allocated resources are needed to add to the tasks on the abnormal task link to shorten the running time; it can also be inferred how many allocated resources are needed to reduce the tasks on the normal task link, so that the task system can reserve more resources for tasks that need additional resources, so that the target task can be executed and completed within the configured completion time. Among them, for tasks whose task execution status is completed, that is, tasks that have been completed in the current operating cycle, there is no need to add or delete resources (no need to adjust resources). In an embodiment of the present application, resources mainly include kernel resources and memory resources, and the amount of resources refers to one or more of the number of kernels and memory capacity.

[0068] In an embodiment of the present application, determining the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task by the second prediction model may specifically include the following steps:

[0069] 1. Obtain the impact scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task.

[0070] Get the first distance between each upstream task and the target task in the abnormal task link, such as Figure 4In task link 2, the first distance between task J and target task M is 1, the first distance between task I and target task M is 2, and the first distance between task H and target task M is 3. Then, according to the first weight coefficient and the first distance, the influence score of each upstream task in the abnormal task link on the target task is determined. The first weight coefficient is inversely proportional to the first distance, that is, the smaller the value of the first distance, the larger the first weight coefficient.

[0071] Get the second distance between each upstream task and the target task in the normal task chain, such as Figure 4 In task link 1, the second distance between task C and target task M is 1, the second distance between task B and target task M is 2, and the second distance between task A and target task M is 3. Then, according to the second weight coefficient and the second distance, the influence score of each upstream task on the target task in the normal task link is determined. The second weight coefficient is inversely proportional to the second distance, that is, the smaller the value of the second distance, the larger the second weight coefficient, wherein the second weight coefficient is greater than the first weight coefficient.

[0072] Optionally, weights may be jointly assigned based on the task execution status and distance of the upstream tasks. For example, the task execution status of the upstream tasks is first determined, and the tasks whose task execution status is not completed are determined. Then, the first distance between each upstream task whose task execution status is not completed and the target task in the abnormal task link associated with the target task is obtained, and the impact score of each upstream task in the abnormal task link on the target task is determined based on the first weight coefficient and the first distance. The second distance between each upstream task whose task execution status is not completed and the target task in the normal task link associated with the target task is obtained, and the impact score of each upstream task in the abnormal task link on the target task is determined based on the second weight coefficient and the second distance.

[0073] 2. Obtain the task execution status of the target task, the expected running time of the target task, the feature data of the upstream tasks of the target task, and the remaining resources of the task system. These data are consistent with the input of the first prediction model and will not be repeated here.

[0074] 3. Obtain the predicted operation data of the target task, including the predicted start time of the target task, the predicted end time of the target task, the predicted operation time of the target task, and the predicted allocated resources of the target task. These predicted operation data are the output of the first prediction model.

[0075] 4. Input the impact score of each upstream task on the target task, the configuration completion time of the target task, the predicted operation data of the target task, the task execution status of the target task, the expected operation time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system into the second prediction model for prediction processing, and obtain the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task. The resource adjustment strategy is the indication information of resource increase or decrease. In addition to obtaining the resource adjustment strategy of each task, the predicted operation data of each task after resource adjustment can also be determined through the second prediction model.

[0076] See also Figure 5 , is a framework diagram of determining a resource adjustment strategy using a second prediction model disclosed in an embodiment of the present application, wherein it can be seen that the second prediction model includes three inputs, namely, the predicted operation data of the target task, the current data of the target task, and the influence relationship between the upstream task and the target task. The predicted operation data of the target task includes the predicted start time of the target task, the predicted end time of the target task, the predicted operation duration of the target task, and the predicted allocated resource amount of the target task. The current data of the target task includes the task execution status of the target task, the expected operation duration of the target task, the characteristic data of the upstream task of the target task, and the remaining resource amount of the task system. The influence relationship between the upstream task and the target task includes the influence score of each upstream task on the target task and the configuration completion time of the target task; the output is the resource adjustment strategy and the predicted operation data of each task, mainly including the resource adjustment strategy of each task, the predicted start time of each task after the resource adjustment, the predicted end time of each task after the resource adjustment, and the predicted operation duration of each task after the resource adjustment.

[0077] S204: adjusting the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task according to the resource adjustment strategy.

[0078] Since the configuration completion time of the target task is added to the second prediction model, the resource adjustment strategy recommended by the second prediction model is more optimal, which can ensure that the adjustment can be completed within the configuration completion time. When the second prediction model is predicting, it is also necessary to consider whether the remaining resources of the task system meet the resource adjustment amount corresponding to the resource adjustment strategy. The input of the second prediction model includes the remaining resources of the task system. Therefore, after determining the resource allocation strategy, the second prediction model can simultaneously determine whether the remaining resources of the task system meet the resource adjustment amount corresponding to the resource adjustment strategy.

[0079] In one possible implementation, if the remaining resources of the task system meet the resource adjustment amount corresponding to the resource adjustment policy, the steps of adjusting the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are executed according to the resource adjustment policy.

[0080] like Figure 6 FIG. 1 is a schematic diagram of task link resource reallocation disclosed in an embodiment of the present application. Figure 6 is Figure 4 In this case, the remaining resources of the task system meet the resource adjustment amount corresponding to the resource adjustment strategy, see Figure 6 , it can be seen that the resource adjustment strategy based on the second prediction model is: Task B reduces 50vCore and 100GiB, Task I adds 100vCore and 200GiB, and Task J adds 50vCore and 200GiB. It can be seen that after adding resources, the predicted running time of Task I is shortened from 1 hour to 50 minutes, and the predicted running time of Task J is shortened from 40 minutes to 30 minutes; after reducing resources, the predicted running time of Task B increases from 40 minutes to 50 minutes. Correspondingly, the predicted running data of Task C and Task M also change. The reduction of resources of Task B will not affect Task M. The increase of resources of Task J and Task I makes the predicted end time of Task M change from 3:30 to 2:50. In this way, the predicted completion time of the target task is not later than the configured completion time of the target task. Therefore, under such a resource adjustment strategy, the predicted end time of Task M can be made earlier than the configured completion time, so that Task M can be successfully executed. Among them, vCore is a virtual core, which represents a logical CPU. It can be used to select the physical characteristics of the hardware (such as the number of cores, memory and storage size); GiB and GB are units for data storage capacity. GB is a decimal unit, usually used to describe the data storage capacity of hard disks, memory, etc., while GiB is a binary unit, usually used to accurately describe the storage capacity in computers. The conversion relationship between the two is 1GB = 0.9313GiB, 1GiB = 1.074GB.

[0081] In a possible implementation, if the remaining amount of resources of the task system does not meet the resource adjustment amount corresponding to the resource adjustment strategy, an alarm message of the target task is generated, and the alarm message of the target task is sent to the management user corresponding to the target task, so that the management user corresponding to the target task adjusts the configuration completion time of the target task, or increases the remaining amount of resources of the task system. Adjusting the configuration completion time of the target task generally means delaying the time so that the target task can be completed; increasing the remaining amount of resources of the task system generally means expanding the capacity so that the amount of resources required by the resource adjustment strategy is met, so that the resources can be reasonably allocated, in order to make the predicted completion time of the target task earlier than the configuration completion time. In addition, in this case, the abnormal task can also be determined from the upstream tasks of the target task in the abnormal task link according to the resource adjustment strategy, and the alarm message of the abnormal task is generated. The alarm message of the abnormal task is sent to the management user corresponding to the abnormal task, so that the management user corresponding to the abnormal task increases the remaining amount of resources of the task system. Different tasks may be in different business domains, and the corresponding management users are different for tasks in different business domains. If the target task and the abnormal task are in different, then the management user corresponding to the target task and the management user corresponding to the abnormal task are different. The abnormal task here refers to the task that requires additional resources in the resource adjustment strategy, but no resources can be added. Whether sending an alarm message to the manager corresponding to the target task or to the manager corresponding to the abnormal task, the purpose is to allow the manager to intervene manually to ensure the successful execution of the target task.

[0082] See also Figure 7 , is a schematic diagram of a task system disclosed in an embodiment of the present application in which the remaining resource amount does not satisfy the resource adjustment amount corresponding to the resource adjustment strategy, from Figure 7 It can be seen that the resource adjustment strategy based on the second prediction model is: Task B reduces 50vCore and 100GiB, Task I adds 100vCore and 200GiB, Task J adds 50vCore and 200GiB but no resources can be added. In this case, the predicted completion time of the target task M (03:20) is still later than the configured completion time of the target task M (3:00). Therefore, it is necessary to send an alarm message to the manager corresponding to the target task M (that is, the manager corresponding to the business domain D), and inform the manager which upstream tasks affect Task M and cannot be completed. Among them, since Task J needs to add resources, but no resources can be added, it is also a task that directly affects Task M. Therefore, Task J is an abnormal task. At this time, it is also necessary to send an alarm message to the manager corresponding to Task J (that is, the manager corresponding to the business domain C), and inform the manager which downstream tasks are affected by Task J.

[0083] In one possible implementation, since the tasks of each business domain are developed and maintained by the R&D personnel of the corresponding business domain, these developers are not familiar with the tasks of other business domains and do not have relevant permissions. Therefore, when tasks on the abnormal task link need to send alarms, they need to be grouped according to the business domain. In the implementation of this application, the target tasks and the corresponding upstream tasks can be grouped according to the business domain. After grouping, different business domains correspond to different managers. When problems occur in the tasks, alarm information can be sent to the corresponding managers. Please refer to Figure 7 , we can see that tasks H and I belong to business domain A, tasks A and B belong to business domain B, tasks C and J belong to business domain C, and task M belongs to business domain D. The binding relationship between tasks, business domains, and managers can be shown as follows: Figure 8 As shown, it is a schematic diagram of the corresponding relationship between tasks, business domains and managers disclosed in the embodiment of the present application. Task H and task I belong to business domain A, which is managed by manager A; task A and task B belong to business domain B, which is managed by manager B; task C and task J belong to business domain C, which is managed by manager C; task M belongs to business domain D, which is managed by manager D. According to this binding relationship, once any problem occurs in any task, it can be quickly sent to the corresponding manager.

[0084] In the embodiments of the present application, the monitoring of the target task is mainly explained. Specifically, the actual operation data and the current operation data of the target task are predicted and processed by the first prediction model to obtain the current predicted operation data of the target task, and when the predicted completion time of the target task is later than the configured completion time, the operation resources of the abnormal task link are automatically optimized in real time by the second prediction model. For links with surplus resources, some operation resources are automatically reduced. On the one hand, cluster resources can be maximized, and on the other hand, human workload can be greatly reduced. Human intervention is only required when resources are insufficient and the system cannot be automatically optimized, thereby improving the efficiency of resource adjustment and the success rate of target task execution.

[0085] The above embodiment describes how to use the first prediction model to predict the operation data of the target task, and how to use the second prediction model to adjust the resources, but does not describe the training process of the first prediction model and the second prediction model. Based on this, the training process of the first prediction model and the second prediction model will be described below:

[0086] See also Fig. 9 , is a flowchart of a training method for a first prediction model disclosed in an embodiment of the present application. The training method can be executed by a computer device, and the computer device can be Figure 1Corresponding to the task processing server 103 shown, the training method of the first prediction model may specifically include but is not limited to the following steps:

[0087] S901: Obtain sample running data of the sample task in multiple running cycles before the first running cycle, the running time of the sample task in the first running cycle, the task execution status, the actual running data, the characteristic data of the upstream task and the remaining resources of the task system.

[0088] In a possible implementation, the sample task is any completed task obtained from the database. During the model training process, the number of sample tasks is not limited. The first operation cycle is also an arbitrary cycle for the sample task. Since the execution log information of the sample task is recorded in the database, when training is performed, the running time, task execution status, actual operation data, characteristic data of the upstream task, and the remaining resources of the task system are obtained at any time within the first operation cycle. Among them, the characteristic data of the upstream task includes the task execution status of the upstream task and the operation data corresponding to the task execution status.

[0089] It should be noted that in order to achieve better training results for the model, after obtaining the above data, these data can be preprocessed in order to identify and delete abnormal data to avoid negative impacts of these abnormal data on model training.

[0090] S902: Input the sample operation data, task execution status, actual operation data, feature data of the upstream task, and the remaining resources of the task system into the first initial model for prediction processing to obtain the predicted operation data of the sample task.

[0091] The first initial model is a supervised learning model, such as a naive Bayes classifier, a logistic regression, a support vector machine, a decision tree, a random forest, etc.

[0092] S903: Adjust model parameters of the first initial model according to the predicted operation data of the sample task and the actual operation data of the sample task to obtain a first prediction model.

[0093] In one possible implementation, after obtaining the predicted running data of the sample task, the loss value between the predicted running data of the sample task and the actual running data of the sample task is determined. If the loss value does not reach the condition for stopping model training, the parameters of the first initial model are adjusted, and then when the loss value reaches the condition for stopping model training, the training is stopped to obtain the first prediction model.

[0094] In order to reflect the accuracy of the first prediction model, after the first prediction model is obtained, a large amount of sample data may be used to conduct experiments to ensure the accuracy of the first prediction model.

[0095] See also Fig.10 , is a flow chart of a training method for a second prediction model disclosed in an embodiment of the present application. The training method can be executed by a computer device, and the computer device can be Figure 1 Corresponding to the task processing server 103 shown, the training method of the second prediction model may specifically include but is not limited to the following steps:

[0096] S1001: Obtain the first running data of the sample task, running time, characteristic data of the upstream task, task execution status, configuration completion time, remaining resources of the task system, the impact score of the upstream task in the task link related to the sample task in the task system on the sample task, and the actual allocated resources of the upstream task and each task in the sample task.

[0097] In one possible implementation, the characteristic data of the upstream task includes the task execution status of the upstream task and the operating data corresponding to the task execution status. Similar to the training process of the first prediction model, the sample task is any completed task obtained from the database. During the model training process, the number of sample tasks is not limited. The first operating data of the sample task is the predicted operating data of the sample task determined by calling the first prediction model. Specifically, the first prediction model is called to predict the sample operating data within multiple operating cycles of the sample task, the running time of the sample task, the task execution status, the configuration completion time, and the remaining resources of the task system to obtain the predicted operating data of the sample task. The running time of the sample task, the characteristic data of the upstream task, the task execution status, the configuration completion time, and the remaining resources of the task system all refer to the data at the current moment.

[0098] S1002: The first running data of the sample task, the running time, the characteristic data of the upstream task, the task execution status, the configuration completion time, the remaining resources of the task system, and the influence score of the upstream tasks in the task link related to the sample task in the task system on the sample task are input into the second initial model prediction processing to obtain the upstream tasks in the task link associated with the sample task in the task system and the predicted allocated resources for the sample task.

[0099] The second initial model also belongs to a supervised learning model, for example, it can be a naive Bayes classifier, logistic regression, support vector machine, decision tree, random forest, etc.

[0100] S1003: Adjust model parameters of the second initial model according to the actual allocated resource amounts of each task in the upstream task and the sample task, the upstream tasks in the task link associated with the sample task in the task system, and the predicted allocated resource amounts of the sample task to obtain a second prediction model.

[0101] In one possible implementation, after obtaining the predicted allocated resource amounts of the upstream tasks and the sample tasks in the task link associated with the sample task in the task system, determine the loss value between the predicted allocated resource amounts of the upstream tasks and the sample tasks in the task link associated with the sample task in the task system, and the actual allocated resource amounts of each task in the upstream tasks and the sample tasks. If the loss value does not reach the condition for stopping model training, adjust the parameters of the second initial model, and then stop training when the loss value reaches the condition for stopping model training to obtain the second prediction model.

[0102] In order to reflect the accuracy of the second prediction model, after the second prediction model is obtained, a large amount of sample data may be used to conduct experiments to ensure the accuracy of the second prediction model.

[0103] Based on the trained first prediction model and the second prediction model, the task processing method of the embodiment of the application can be summarized as follows: Fig.11 See the flowchart shown in Fig.11 , is a flowchart of another task processing method disclosed in an embodiment of the present application, which is simplified into the following steps:

[0104] S1101: Calling a first prediction model to predict the running data of a target task, where the predicted running data includes a predicted completion time.

[0105] In this process, in addition to predicting the end time of the target task, the start time, running time and allocated resource amount of the target task can also be predicted. Therefore, in an optional embodiment, in addition to monitoring the target task according to the configured completion time, it can also be detected according to other indicators, for example, it can be directly monitored according to the resource amount.

[0106] S1102: Determine whether the predicted completion time of the target task exceeds the configured completion time.

[0107] If yes, execute step S1103, if no, jump back to step S1101.

[0108] S1103: Calling the second prediction model to adjust resources for the target task and the upstream tasks of the target task, and obtaining resource adjustment strategies for each task.

[0109] S1104: Determine whether the remaining resources of the task system meet the resource requirements of the resource adjustment strategy of each task.

[0110] If so, execute step S1105; if not, execute step S1106.

[0111] S1105: Perform resource adjustment according to the resource adjustment policy. Meanwhile, continue to execute step S1101.

[0112] S1106: Generate warning information.

[0113] In this step, if any task has an exception, an alarm message for the corresponding task is generated.

[0114] S1107: Send an alarm message to the administrator of the business domain to which the task belongs.

[0115] The alarm information can be sent to the management personnel in a preset form, including SMS, picture, and voice broadcast, etc. Furthermore, different sending forms can be adopted according to the abnormality degree of the task. If the abnormality degree indicates that the task is in an emergency state, the alarm can be repeatedly issued. If the abnormality degree indicates that the task is in a non-emergency state, the alarm information can be issued once.

[0116] S1108: Determine whether the target task has been completed.

[0117] If completed, the process ends; if not completed, execute step S1101.

[0118] Based on the description of the above embodiments, the implementation of this application can achieve the following technical effects:

[0119] 1. Predictable running data of target tasks. We generally cannot know in advance how long a task will take to run and how many resources it will consume. Only after the task is finished can we get the running time and resource consumption data from the running data. Through deep learning algorithms and AI's predictive capabilities, we can know the start and end time of a task and the amount of resources required in advance more accurately. Therefore, when there may be potential problems in a task that cannot be completed in time, an early warning can be issued to notify relevant personnel to intervene manually to ensure the normal completion of the task's running goals.

[0120] 2. Automatically optimize abnormal task links. Based on the predicted task operation data, the operation resources of abnormal task links are automatically optimized in real time. For links with surplus resources, some operation resources are automatically reduced. On the one hand, cluster resources can be maximized, and on the other hand, human workload can be greatly reduced. Human intervention is only required when the system cannot be automatically optimized.

[0121] 3. Group alarms according to the business domain to which the tasks belong. The tasks of each business domain are developed and maintained by the R&D personnel of the corresponding business domain. These developers are not familiar with the tasks of other business domains and do not have relevant permissions. Therefore, when tasks on the abnormal task link need to send alarms, they need to be grouped according to the business domain. After grouping, they are sent to the on-duty personnel of the corresponding business domain to improve the efficiency of problem handling.

[0122] Based on the above method embodiment, the present application embodiment also provides a structural diagram of a task processing device. Fig.12 , which is a structural diagram of a task processing device disclosed in an embodiment of the present application.

[0123] Fig.12 The task processing device 1200 shown can run the following units:

[0124] A determination unit 1201 is used to determine a predicted completion time of a target task by using a first prediction model;

[0125] A determining unit 1201 is configured to determine an abnormal task link and a normal task link from task links related to the target task if the predicted completion time of the target task is later than the configured completion time of the target task;

[0126] A determining unit 1201 is used to determine the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through a second prediction model;

[0127] The processing unit 1202 is used to adjust the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task according to the resource adjustment strategy.

[0128] In a possible implementation, when the determining unit 1201 determines the abnormal task link and the normal task link from the task links related to the target task, it is specifically configured to perform the following steps:

[0129] Obtaining a predicted completion time of a first upstream task of the target task in each task link related to the target task, where the first upstream task is an upstream task adjacent to the target task;

[0130] The abnormal task link and the normal task link are determined from task links related to the target task according to the predicted completion time of the first upstream task and the predicted start time of the target task.

[0131] In a possible implementation, when the determination unit 1201 determines the abnormal task link and the normal task link from the task links related to the target task according to the predicted completion time of the first upstream task and the predicted start time of the target task, it is specifically configured to perform the following steps:

[0132] Determine the time interval between the predicted completion time of the first upstream task and the predicted start time of the target task in each task link;

[0133] Determine the task link whose time interval is less than the time interval threshold as an abnormal task link associated with the target task;

[0134] The task link whose time interval is greater than or equal to the time interval threshold is determined as a normal task link associated with the target task.

[0135] In a possible implementation, when the determining unit 1201 determines the predicted completion time of the target task by using the first prediction model, it is specifically configured to perform the following steps:

[0136] Acquire actual operation data of the target task in multiple operation cycles before the current operation cycle, wherein the actual operation data includes one or more of a start time, an end time, a running time, and an allocated resource amount;

[0137] Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status;

[0138] The actual running data of the target task, the task status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task and the remaining resources of the task system are input into the first prediction model for prediction processing to obtain the predicted completion time of the target task.

[0139] In a possible implementation, when the determination unit 1201 determines the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through the second prediction model, it is specifically used to perform the following steps:

[0140] Obtaining the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task;

[0141] Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status;

[0142] The impact scores of each upstream task on the target task, the configuration completion time of the target task, the predicted running data of the target task, the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream tasks of the target task and the remaining resources of the task system are input into the second prediction model for prediction processing to obtain the upstream tasks of the target task in the abnormal task link, the upstream tasks of the target task in the normal task link and the resource adjustment strategy corresponding to the target task.

[0143] In a possible implementation, when the acquisition unit 1203 acquires the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task, it is specifically configured to perform the following steps:

[0144] Acquire a first distance between each upstream task of the target task in the abnormal task link and the target task;

[0145] Acquire a second distance between each upstream task of the target task in the normal task link and the target task;

[0146] Determining, according to the first weight coefficient and the first distance, an influence score of each upstream task in the abnormal task link on the target task;

[0147] The influence score of each upstream task in the normal task link on the target task is determined according to the second weight coefficient and the second distance; wherein the first weight coefficient is greater than the second weight coefficient.

[0148] It can be understood that the functions of each functional unit of the task processing device provided in the embodiment of the present application can be specifically implemented according to the method in the above method embodiment, and its specific implementation process can refer to the relevant description in the above method embodiment, which will not be repeated here.

[0149] In a feasible embodiment, the task processing device provided in the embodiment of the present application can be implemented in software. The task processing device can be stored in a memory. It can be software in the form of programs and plug-ins, and includes a series of units, including a determination unit and a processing unit; wherein the determination unit and the processing unit are used to implement the task processing method provided in the embodiment of the present application.

[0150] In other feasible embodiments, the task processing device provided in the embodiment of the present application can also be implemented in a combination of software and hardware. As an example, the task processing device provided in the embodiment of the present application can be a processor in the form of a hardware decoding processor, which is programmed to implement the task processing method provided in the embodiment of the present application. For example, the processor in the form of a hardware decoding processor can adopt one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs) or other electronic components.

[0151] In the embodiment of the present application, in combination with the determination unit and the processing unit, the method includes: the determination unit determines the predicted completion time of the target task through the first prediction model. The predicted completion time of the target task in the current operation cycle can be predicted in advance through the prediction model. If the predicted completion time of the target task is later than the configuration completion time of the target task, the determination unit determines the abnormal task link and the normal task link from the task link related to the target task. The determination unit determines the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through the second prediction model. When the predicted completion time of the target task is later than the configuration completion time, the task link that affects the target task can be determined, and the resources of each task are adjusted in time based on the prediction model. The processing unit adjusts the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task according to the resource adjustment strategy, so that the predicted completion time of the target task is not later than the configuration completion time of the target task. In summary, the task processing method provided in the embodiment of the present application can, on the one hand, predict the operating data of the target task through a prediction model; on the other hand, when the predicted completion time of the target task is later than the configured completion time, the prediction model can be used to dynamically adjust the resources of the target task to improve the operation success rate of the target task.

[0152] Based on the description of the above method embodiment and the device embodiment, the present application embodiment further provides a computer device; the computer device may be the task processing server 103 mentioned above. Fig.13The computer device may include at least a processor 1301, an input interface 1302, an output interface 1303 and a computer-readable storage medium 1304; wherein the processor 1301, the input interface 1302, the output interface 1303 and the computer-readable storage medium 1304 may be connected via a bus or other connection methods.

[0153] The computer-readable storage medium 1304 may be stored in the memory of the computer device, the computer-readable storage medium 1304 is used to store computer programs, and the processor 1301 is used to execute the computer programs stored in the computer-readable storage medium 1304. The processor 1301 (or CPU (Central Processing Unit)) is the computing core and control core of the computer device, which is suitable for implementing one or more computer programs, and is specifically suitable for loading and executing one or more computer programs to implement corresponding method flows or corresponding functions.

[0154] The embodiment of the present application also provides a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space that stores the operating system of the computer device. In addition, one or more computer programs (including program codes) suitable for being loaded and executed by the processor are also stored in the storage space. It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (Non-Volatile Memory), such as at least one disk storage; optionally, it can also be at least one computer-readable storage medium located away from the aforementioned processor.

[0155] In the embodiment of the present application, the processor 1301 may load and execute one or more computer programs stored in the computer-readable storage medium 1304 to implement the corresponding steps of the method shown in the above embodiment. In a specific implementation, the computer program in the computer-readable storage medium 1304 is loaded by the processor 1301 and executes the following steps:

[0156] Determine the predicted completion time of the target task by using the first prediction model;

[0157] If the predicted completion time of the target task is later than the configured completion time of the target task, determining an abnormal task link and a normal task link from task links related to the target task;

[0158] Determine, by a second prediction model, an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link, and a resource adjustment strategy corresponding to the target task;

[0159] The allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted according to the resource adjustment strategy.

[0160] In a possible implementation, when the computer program in the computer-readable storage medium 1304 is loaded and executed by the processor 1301 to determine abnormal task links and normal task links from the task links related to the target task, the computer program is specifically used to perform the following steps:

[0161] Obtaining a predicted completion time of a first upstream task of the target task in each task link related to the target task, where the first upstream task is an upstream task adjacent to the target task;

[0162] According to the predicted completion time of the first upstream task and the predicted start time of the target task, an abnormal task link and a normal task link are determined from task links related to the target task.

[0163] In a possible implementation, the computer program in the computer-readable storage medium 1304 is loaded and executed by the processor 1301 to determine abnormal task links and normal task links from task links related to the target task according to the predicted completion time of the first upstream task and the predicted start time of the target task, and is specifically used to perform the following steps:

[0164] Determine the time interval between the predicted completion time of the first upstream task and the predicted start time of the target task in each task link;

[0165] Determine the task link whose time interval is less than the time interval threshold as an abnormal task link associated with the target task;

[0166] The task link whose time interval is greater than or equal to the time interval threshold is determined as a normal task link associated with the target task.

[0167] In a possible implementation, when the computer program in the computer-readable storage medium 1304 is loaded and executed by the processor 1301 to determine the predicted completion time of the target task by using the first prediction model, the computer program is specifically configured to perform the following steps:

[0168] Acquire actual operation data of the target task in multiple operation cycles before the current operation cycle, wherein the actual operation data includes one or more of a start time, an end time, a running time, and an allocated resource amount;

[0169] Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status;

[0170] The actual running data of the target task, the task status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task and the remaining resources of the task system are input into the first prediction model for prediction processing to obtain the predicted completion time of the target task.

[0171] In a possible implementation, when the computer program in the computer-readable storage medium 1304 is loaded and executed by the processor 1301 to determine the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task through the second prediction model, it is specifically used to perform the following steps:

[0172] Obtaining the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task;

[0173] Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status;

[0174] The impact scores of each upstream task on the target task, the configuration completion time of the target task, the predicted running data of the target task, the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream tasks of the target task and the remaining resources of the task system are input into the second prediction model for prediction processing to obtain the upstream tasks of the target task in the abnormal task link, the upstream tasks of the target task in the normal task link and the resource adjustment strategy corresponding to the target task.

[0175] In a possible implementation, when the computer program in the computer-readable storage medium 1304 is loaded and executed by the processor 1301 to obtain the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task, it is specifically used to perform the following steps:

[0176] Acquire a first distance between each upstream task of the target task in the abnormal task link and the target task;

[0177] Acquire a second distance between each upstream task of the target task in the normal task link and the target task;

[0178] Determining, according to the first weight coefficient and the first distance, an influence score of each upstream task in the abnormal task link on the target task;

[0179] The influence score of each upstream task in the normal task link on the target task is determined according to the second weight coefficient and the second distance; wherein the first weight coefficient is greater than the second weight coefficient.

[0180] According to one aspect of the present application, a computer program product is provided, the computer program product comprising a computer program, the computer program being stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the task processing method provided in the above-mentioned various optional modes.

[0181] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0182] In the several embodiments provided in this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, and the division of the modules described above is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0183] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0184] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A task processing method, characterized in that: The method comprises: Determine the predicted completion time of the target task by using the first prediction model; If the predicted completion time of the target task is later than the configured completion time of the target task, determining an abnormal task link and a normal task link from task links related to the target task; Determine, by a second prediction model, an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link, and a resource adjustment strategy corresponding to the target task; The allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task are adjusted according to the resource adjustment strategy.

2. The method according to claim 1, characterized in that The determining of abnormal task links and normal task links from task links related to the target task includes: Obtaining a predicted completion time of a first upstream task of the target task in each task link related to the target task, where the first upstream task is an upstream task adjacent to the target task; According to the predicted completion time of the first upstream task and the predicted start time of the target task, an abnormal task link and a normal task link are determined from task links related to the target task.

3. The method according to claim 2, characterized in that The determining, according to the predicted completion time of the first upstream task and the predicted start time of the target task, abnormal task links and normal task links from task links related to the target task comprises: Determine the time interval between the predicted completion time of the first upstream task and the predicted start time of the target task in each task link; Determine the task link whose time interval is less than the time interval threshold as an abnormal task link associated with the target task; The task link whose time interval is greater than or equal to the time interval threshold is determined as a normal task link associated with the target task.

4. The method according to any one of claims 1 to 3, characterized in that Determining the predicted completion time of the target task by using the first prediction model includes: Acquire actual operation data of the target task in multiple operation cycles, wherein the actual operation data includes one or more of a start time, an end time, a running time, and an allocated resource amount; Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status; The actual running data of the target task, the task status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task and the remaining resources of the task system are input into the first prediction model for prediction processing to obtain the predicted completion time of the target task.

5. The method according to any one of claims 1 to 3, characterized in that: The determining, by using the second prediction model, the upstream task of the target task in the abnormal task link, the upstream task of the target task in the normal task link, and the resource adjustment strategy corresponding to the target task includes: Obtaining the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task; Acquire the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream task of the target task, and the remaining resources of the task system; wherein the characteristic data of the upstream task includes the task execution status of the upstream task and the running data corresponding to the task execution status; The impact scores of each upstream task on the target task, the configuration completion time of the target task, the predicted running data of the target task, the task execution status of the target task, the expected running time of the target task, the characteristic data of the upstream tasks of the target task and the remaining resources of the task system are input into the second prediction model for prediction processing to obtain the upstream tasks of the target task in the abnormal task link, the upstream tasks of the target task in the normal task link and the resource adjustment strategy corresponding to the target task.

6. The method according to claim 5, characterized in that The obtaining of the influence scores of each upstream task of the target task in the abnormal task link and the normal task link on the target task includes: Acquire a first distance between each upstream task of the target task in the abnormal task link and the target task; Acquire a second distance between each upstream task of the target task in the normal task link and the target task; Determining, according to the first weight coefficient and the first distance, an influence score of each upstream task in the abnormal task link on the target task; The influence score of each upstream task in the normal task link on the target task is determined according to the second weight coefficient and the second distance; wherein the first weight coefficient is greater than the second weight coefficient.

7. A task processing device, characterized in that: The device comprises: A determination unit, configured to determine a predicted completion time of a target task by using a first prediction model; The determining unit is further configured to determine an abnormal task link and a normal task link from task links related to the target task if the predicted completion time of the target task is later than the configured completion time of the target task; The determining unit is further configured to determine, by means of a second prediction model, an upstream task of the target task in the abnormal task link, an upstream task of the target task in the normal task link, and a resource adjustment strategy corresponding to the target task; A processing unit is used to adjust the allocated resources of the upstream task of the target task in the abnormal task link, the allocated resources of the upstream task of the target task in the normal task link, and the allocated resources of the target task according to the resource adjustment strategy.

8. A computer device, characterized in that: The computer device comprises: a processor adapted to implement one or more computer programs; and, A computer-readable storage medium storing one or more computer programs, wherein the one or more computer programs are suitable for being loaded by the processor and executing the task processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more computer programs, and the one or more computer programs are suitable for being loaded by a processor and executing the task processing method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The computer program product includes a computer program, which is stored in a computer-readable storage medium. The processor of the computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the task processing method according to any one of claims 1 to 6.

Images