A conflict coordination method and apparatus, electronic device, and storage medium

Abstract

The application discloses a conflict coordination method and device, electronic equipment and storage medium, relates to the technical field of conflict coordination, and comprises the following steps: acquiring resource characteristics and logical characteristics of a plurality of running entities; generating resource competition characteristics of the plurality of running entities according to the resource characteristics, and generating a resource conflict detection result according to the resource competition characteristics and stored resource conflict characteristics; generating logical association characteristics of the plurality of running entities according to the logical characteristics, and generating a logical conflict detection result according to the logical association characteristics and stored logical conflict characteristics; in response to both indicating that there is a conflict, determining that the plurality of running entities are composite conflict entities, performing root cause determination processing on the composite conflict entities, determining a conflict handling action and executing. The application can comprehensively consider the coupling relationship between hardware resource conflicts and software logic conflicts among running entities, identify and handle composite conflicts, and solve the problem that in the related art, it is difficult to accurately distinguish the real cause of the conflict, resulting in poor conflict resolution effect.

Description

Technical Field

[0001] This application relates to the field of conflict coordination technology, and in particular to a conflict coordination method, apparatus, electronic device and storage medium. Background Technology

[0002] Cloud computing platforms typically handle a large number of concurrent services within the same resource pool, with multiple running entities operating simultaneously on the platform. Because these running entities share underlying computing, storage, and network resources, and engage in interactions such as calls, coordination, and dependency waiting during service execution, various conflicts can easily arise between them, leading to performance spikes, response timeouts, task blocking, and even service unavailability.

[0003] Existing conflict detection and governance solutions often focus on hardware resource conflicts, such as identifying and handling resource contention from the perspectives of resource utilization, quota limits, or preemption relationships. However, these solutions typically pay insufficient attention to software logic anomalies caused by call chains, dependencies, or timing anomalies between running entities, resulting in incomplete conflict detection and governance coverage. Furthermore, in real-world operating environments, hardware resource conflicts and software logic conflicts are often coupled and mutually reinforcing. That is, resource contention may induce logical anomalies such as timeouts, amplified retry, and cascading waits, while wait / dependency propagation at the logic level may cause prolonged resource occupation or exacerbate exclusive constraints, further amplifying resource-side conflicts. Since the external manifestations of the two types of conflicts can easily mask or amplify each other, single-level detection and handling often fail to accurately distinguish the true cause of the conflict, easily leading to root cause misjudgment. This results in mismatches between handling strategies and actual problems, poor handling effects, and even repeated or escalating conflicts. Summary of the Invention

[0004] This application provides a conflict coordination method that can comprehensively consider the coupling relationship between hardware resource conflicts and software logic conflicts between running entities in a cloud computing platform. This method aims to solve the problem in related technologies that it is difficult to accurately distinguish the true cause of the conflict when conflict detection and handling are performed from a single level, which can easily lead to misjudgment of the root cause and poor conflict resolution.

[0005] This application provides a conflict coordination method, including: Obtain entity operation data of multiple running entities in the cloud computing platform, and generate resource characteristics and logical characteristics of multiple running entities based on the entity operation data; Based on resource characteristics, resource contention characteristics of multiple running entities are generated. Based on the resource contention characteristics and the resource conflict characteristics stored in the pre-built conflict characteristic library, resource conflict detection is performed on multiple running entities to generate resource conflict detection results. Based on logical characteristics, generate logical association characteristics for multiple running entities, and based on logical association characteristics and logical conflict characteristics stored in the conflict characteristic library, perform logical conflict detection on multiple running entities and generate logical conflict detection results. If both the resource conflict detection result and the logical conflict detection result indicate the existence of a conflict, then multiple running entities are determined to be composite conflict entities, and composite conflict evidence of multiple composite conflict entities is generated based on the resource conflict detection result and the logical conflict detection result. Based on complex conflict evidence, root cause determination processing is performed on multiple complex conflict entities to generate root cause determination results. Based on the root cause determination results, conflict resolution actions for multiple complex conflict entities are determined and executed.

[0006] This application also provides a conflict coordination device, comprising: The feature generation module is used for entity operation data of multiple running entities in the cloud computing platform, and generates resource features and logical features of multiple running entities based on the entity operation data. The resource conflict detection module is used to generate resource competition characteristics of multiple running entities based on resource characteristics, and to perform resource conflict detection on multiple running entities based on the resource competition characteristics and the resource conflict characteristics stored in the pre-built conflict characteristic library, and generate resource conflict detection results. The logical conflict detection module is used to generate logical association features of multiple running entities based on logical characteristics, and to perform logical conflict detection on multiple running entities based on the logical association features and the logical conflict features stored in the conflict feature library, and generate logical conflict detection results. The composite determination module is used to determine multiple running entities as composite conflict entities if both the resource conflict detection result and the logical conflict detection result indicate that there is a conflict, and to generate composite conflict evidence for multiple composite conflict entities based on the resource conflict detection result and the logical conflict detection result. The conflict resolution module is used to perform root cause determination processing on multiple complex conflict entities based on complex conflict evidence, generate root cause determination results, and determine and execute conflict resolution actions for multiple complex conflict entities based on the root cause determination results.

[0007] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for implementing any of the above-described conflict coordination methods when executing the computer program.

[0008] This application also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of any of the above-described conflict coordination methods.

[0009] This application addresses the technical problem of incomplete conflict detection and governance caused by prioritizing hardware resource conflicts. By acquiring operational data from both the resource and logical dimensions and conducting conflict identification separately, conflict analysis no longer relies solely on a single resource indicator while ignoring logical factors such as call chains, dependency waiting, and timing anomalies. This solves the problem of existing technologies focusing only on hardware resource conflicts, resulting in incomplete conflict detection and governance. Furthermore, when both resource-side and logical-side detections indicate anomalies, related operational entities are grouped into composite conflict objects, and the two types of detection results are aggregated to form unified evidence. This allows scenarios where resource contention and logical anomalies are coupled and mutually reinforcing to be comprehensively characterized and incorporated into the same judgment criterion for root cause determination and conflict resolution. This solves the technical problem of existing technologies easily misjudging the root cause of conflicts when resource conflicts and software logic conflicts mutually mask or amplify each other, leading to poor resolution results. This reduces the risk of root cause misjudgment and improves the effectiveness and stability of conflict resolution. Attached Figure Description

[0010] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of a conflict coordination application environment provided in an embodiment of this application; Figure 2 A flowchart of a conflict coordination method provided in an embodiment of this application; Figure 3 A flowchart of conflict coordination steps is provided for an embodiment of this application; Figure 4 A structural block diagram of a conflict coordination device provided in an embodiment of this application; Figure 5 This is a schematic diagram of an electronic device structure provided in an embodiment of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0013] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0014] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] This application provides a fault early warning method that can be applied to, for example... Figure 1 The application environment shown includes a cloud computing platform, a control plane set up within the cloud computing platform, and multiple running entities (running entity 1, running entity 2... running entity n) carried by the cloud computing platform. The cloud computing platform can be a public cloud, private cloud, or hybrid cloud platform, or a data center cloud infrastructure. It can integrate or interface with virtualized / containerized operating environments and resource management capabilities, such as an infrastructure layer that provides computing, storage, and network resource provisioning and isolation, and a cloud management / container orchestration platform that provides orchestration, scheduling, scaling, and resource quota / limit management. The running entity is used to carry specific business or system components. The running entity can be any one or more of containers, virtual machines, processes or service instances. During the running process, the running entity generates entity running data. The entity running data includes at least resource running data that characterizes the resource usage status and logical running data that characterizes the interaction relationship between running entities, so as to support the generation of resource characteristics and logical characteristics of multiple running entities within a preset time window, and further form resource competition characteristics and logical association characteristics for conflict identification. The control plane can be the management control plane of a cloud computing platform or a set of control components that are logically equivalent to it. It is used to provide unified observation, policy decision-making, and orchestration execution capabilities across operating entities. The control plane and the operating entities form a two-way interaction: On the one hand, the control plane is used to aggregate or receive entity operation data (such as monitoring indicators, logs, call chain traces, and event data) from multiple operating entities, and based on this, generate features of the operation data, perform resource conflict detection and / or logical conflict detection based on the features, generate evidence of compound conflicts when compound conflicts exist, and determine the root cause; on the other hand, the control plane is also used to determine and issue conflict handling actions to the corresponding operating entities and / or resource objects after obtaining the root cause determination results (such as performing resource release / reclaim, constraint adjustment, dependency adjustment, restart / migration, etc. through orchestration and scheduling interfaces), and can continue to obtain updated operation data after handling to determine whether the conflict has been resolved and whether further action compensation is needed, thereby realizing the conflict coordination of "observation-detection-determination-handling-verification" within the cloud computing platform.

[0016] like Figure 2 As shown, an embodiment of this application provides a conflict coordination method, including: Step 201: Obtain entity operation data of multiple running entities in the cloud computing platform, and generate resource characteristics and logical characteristics of multiple running entities based on the entity operation data; Step 202: Generate resource contention characteristics for multiple running entities based on resource characteristics, and perform resource conflict detection on multiple running entities based on resource contention characteristics and resource conflict characteristics stored in a pre-built conflict characteristic library, and generate resource conflict detection results. Step 203: Generate logical association features of multiple running entities based on logical features, and perform logical conflict detection on multiple running entities based on logical association features and logical conflict features stored in the conflict feature library, and generate logical conflict detection results. Step 204: In response to the fact that both the resource conflict detection result and the logical conflict detection result indicate the existence of a conflict, the multiple running entities are determined to be composite conflict entities, and composite conflict evidence of the multiple composite conflict entities is generated based on the resource conflict detection result and the logical conflict detection result. Step 205: Based on the complex conflict evidence, perform root cause determination processing on multiple complex conflict entities, generate root cause determination results, and determine and execute conflict resolution actions for multiple complex conflict entities according to the root cause determination results.

[0017] Specifically, this application provides a conflict coordination method that, after acquiring the runtime data of the running entities, simultaneously forms corresponding representations from both the resource and logical dimensions and conducts conflict identification separately. This ensures that conflict analysis no longer relies solely on a single resource indicator while ignoring logical factors such as call chains, dependency waiting, and timing anomalies. This addresses the technical problem of existing technologies that only focus on hardware resource conflicts, leading to incomplete conflict detection and governance. It achieves the technical effect of improving conflict identification coverage and reducing missed logical conflicts. Furthermore, when both resource-side and logical-side detections indicate anomalies, by merging related running entities into composite conflict objects and aggregating the two types of detection results to form unified evidence, the scenario where resource contention and logical anomalies are coupled and mutually causal can be comprehensively characterized and incorporated into the same judgment criterion. This allows for root cause determination and conflict handling, thus solving the technical problem in existing technologies where resource conflicts and software logical conflicts mutually mask or amplify each other, easily leading to misjudgment of the root cause and poor handling results. This achieves the technical effect of reducing the risk of root cause misjudgment and improving the effectiveness and stability of conflict resolution.

[0018] In one embodiment, entity operation data of multiple running entities in a cloud computing platform are obtained, and resource characteristics and logical characteristics of the multiple running entities are generated based on the entity operation data, including: The system acquires entity operation data for multiple running entities within a preset time window. Entity operation data includes at least resource operation data characterizing resource usage status and logical operation data characterizing the interaction relationships between running entities. A running entity refers to any one or more of containers, virtual machines, processes, or service instances. The preset time window is preferably a sliding window or a scrolling window, with a configurable window length. Resource operation data preferably includes one or more of the following metrics: CPU utilization / quota usage, memory usage / page faults, storage I / O throughput or IOPS and queue length, network throughput / packet loss / connection count, and port occupancy status. Logical operation data preferably includes data characterizing interaction relationships, such as call tracing data, RPC / HTTP request / response logs, message queue production and consumption records, connection establishment / release events, thread pool / connection pool queuing events, and lock wait events. The resource operation data and logic operation data are respectively processed by time alignment, denoising, and normalization to obtain standard resource operation data and standard logic operation data. Time alignment refers to mapping data with different sampling frequencies or different sources to the same time base according to the timestamp, which is preferably achieved by resampling to a unified sampling interval. Denoising preferably includes one or more of the following: abnormal peak removal, missing value imputation, moving average / median filtering. Normalization preferably includes Min-Max normalization, Z-score normalization, or proportional processing according to the resource limit (quota / limit). Based on standard resource operation data of multiple operating entities, resource characteristics of these entities are extracted to characterize at least their resource utilization levels, resource allocation constraints, and resource change trends. Specifically, resource utilization levels preferably refer to the mean, peak value, or quantile (P95 / P99) within a time window; resource allocation constraints refer to constraints such as quotas / limits, CPU affinity, I / O rate limiting, and connection limit, which are preferably obtained by combining platform-side configuration parameters with monitoring indicators; and resource change trends refer to the direction and magnitude of indicator changes over time, which are preferably obtained through linear fitting slope, adjacent difference statistics, or abrupt change point counting. Based on the standard logical operation data of multiple running entities, interaction relationship representations of multiple running entities are generated respectively, and the logical characteristics of the running entities are determined according to the interaction relationship representations, so as to at least characterize the associated entity object information of the running entity. The associated entity object information includes at least: a set of associated entity object identifiers that interact with the corresponding running entity, the interaction direction and / or interaction type with each associated entity object, interaction timing summary information, and waiting / dependency characteristics.

[0019] Specifically, the interaction relationship representation is preferably a graph structure representation or an edge set representation, used to describe the "upstream → downstream" dependency edges between entities; the interaction direction refers to requester → requested party or producer → consumer; the interaction type includes at least one or more of RPC / HTTP calls, database access, message subscription, lock contention, and shared queue queuing; the interaction timing summary information preferably includes the number of calls, average / quantile latency, timeout ratio, retry ratio, and queuing waiting time statistics; the waiting / dependency characteristics preferably refer to the blocking waiting ratio, waiting chain length, whether a circular waiting is formed, sensitivity to downstream failures, etc., which can be obtained through the parent-child relationship of the call chain Span, queuing events, and lock waiting events statistics.

[0020] Specifically, in this embodiment, by simultaneously acquiring the resource operation data and logical operation data of the running entity within a preset time window, and performing time alignment, denoising, and normalization processing respectively, resource features that can characterize resource occupancy levels, resource allocation constraints, and resource change trends are extracted, as well as logical features that generate interaction relationship representations and determine the information of associated entity objects accordingly. This enables the resource-side and logical-side states of different running entities to be expressed in a standardized manner under the same time reference, and key behavioral features to be extracted in a structured way, improving feature comparability and noise resistance, and providing a more stable input for subsequent conflict identification.

[0021] In one embodiment, the method further includes, prior to performing resource conflict detection on multiple running entities: The virtualization environment parameters are set based on the cloud computing platform's runtime environment. A virtualization test sandbox is then constructed based on these parameters. The virtualization environment parameters include at least the following: computing resource specifications (CPU core count / quota, memory limit), storage and network configuration (disk type, I / O rate limiting, bandwidth limit, latency policy), isolation policies (namespaces, cgroups, virtual networks), and optional sampling period / log level / fault injection switch. The virtualization test sandbox preferably refers to an environment that is isomorphic or nearly isomorphic to the production environment in terms of critical runtime components and resource constraints, is isolated from production, and supports snapshot rollback. Deploy at least two test running entities in a virtualized test sandbox and apply preset conflict triggering conditions to at least two test running entities to generate at least one conflict scenario. The preset conflict triggering conditions refer to the configuration and load combination used to trigger resource contention or logical anomalies, preferably including one or more of the following: differentiated configuration of resource quotas, hotspot binding of shared resources, latency / packet loss injection, concurrency gate and queue length limit, and timeout retry parameter adjustment. Obtain the test entity runtime data corresponding to the conflict scenario, and annotate the test entity runtime data to obtain conflict annotation data. The test entity runtime data preferably comes from monitoring metrics, call chain tracing, event logs and injection records. The annotation preferably includes at least the conflict start and end time, the set of entities involved, the resource objects involved, the conflict type label and the severity level. The annotation method can be manual annotation, rule-based automatic annotation or semi-automatic annotation. Conflict features are generated based on conflict annotation data, and these features are classified and stored according to resource conflict features and / or logical conflict features to construct a conflict feature library. The conflict features are preferably vectorized representations or sets of structured fields. The classified storage is preferably organized according to "conflict type - feature version - environmental baseline parameters" to support subsequent matching and backtracking.

[0022] Specifically, in this embodiment, a virtualized test sandbox is constructed by setting virtualization environment parameters based on the platform's operating environment. Test entities are deployed in the sandbox and preset conflict induction conditions are applied to generate conflict scenarios. Then, test entity operation data is acquired and labeled to form conflict labeling data. Conflict features are generated from the conflict labeling data and classified and stored according to resource conflict features and / or logical conflict features to construct a conflict feature library. This allows conflict patterns to be systematically accumulated in a "labelable, precipitable, and reusable" form and used for subsequent comparisons, thereby improving the representativeness and reusability of the conflict feature library and enhancing the consistency of detection references.

[0023] In a preferred embodiment, at least two test execution entities are deployed in a virtualized test sandbox, and preset conflict induction conditions are applied to the at least two test execution entities to generate at least one conflict scenario, including: A set of typical runtime entity templates for generating conflict scenarios is pre-built. The set of typical runtime entity templates includes at least: computational entity templates, memory entity templates, storage I / O entity templates, network entity templates, lock / queue entity templates, and dependency call entity templates. Each entity template is used to represent different load profiles and supports adjusting load intensity and resource ratio through parameter configuration. Specifically, the load profile refers to the resource consumption pattern and interaction behavior pattern of an entity in the dimensions of CPU / memory / I / O / network / lock and queue. The preferred parameter configurations include concurrency, request rate, data block size, lock granularity, cache capacity, and number of retries. Based on the typical runtime entity template set, a set of test runtime entities consisting of preferably 2 to 4 test runtime entities is selected and deployed in the virtualization test sandbox. The set of test runtime entities includes at least: a resource pressure entity for applying resource pressure and a dependency interaction entity for building interaction dependencies, so as to have the basic conditions to reproduce resource contention and interaction waiting with as few entities as possible. Configure switchable interaction relationships for at least two test run entities to cover a variety of typical interaction topologies with a small number of test run entities. The interaction relationships include at least: serial dependency topology, fan-out dependency topology, fan-in dependency topology, shared downstream dependency topology, and ring dependency topology. Preferably, the interaction relationship is described by "configurable dependency edges", and adjustable parameters are configured for each dependency edge. The adjustable parameters include at least: concurrency, queue length, timeout, and number of retries, so that different types of conflict scenarios can be generated by switching parameters without increasing the size of the test run entities. Here, configurable dependency edges refer to the edges that abstract the interaction between entities as upstream → downstream and have configurable edge attributes; the interaction sequence is preferably obtained by event points such as request initiation / response and queue start / end. To generate resource conflict scenarios, resource conflict induction conditions are applied to at least two test entities. These conditions include: setting differentiated limits on CPU quotas / core counts, memory limits, storage I / O bandwidth, and / or network bandwidth; controlling the resource-pressured entities to gradually increase their corresponding resource usage to form resource competition; optionally, configuring at least two test entities to access the same shared resource object to form hotspot competition, wherein the shared resource object includes at least one or more of shared storage volumes, shared thread pools, shared connection pools, shared locks, or shared queues; optionally, periodic fluctuations or sudden rate limits are applied to resource supply capacity to make the resource competition process include trend changes and sudden spikes, thereby forming more representative resource competition behavior. The gradual increase in resource usage is preferably executed according to a preset load curve (boost-hold-blowdown); the periodic fluctuations or sudden rate limits are preferably implemented through dynamic adjustment of quota / rate limit strategies. To generate logical conflict scenarios, logical conflict induction conditions are applied to at least two test entities. These conditions include: configuring timeout and retry policy parameters on the interaction dependency chain to control anomaly propagation or convergence behavior; setting constraints on thread pool / connection pool / message consumption concurrency and queue length to induce queuing and cascading backpressure; optionally, setting shared locks or order constraints for multiple resource requests to induce circular waiting or lock contention amplification; and optionally, injecting delays, random jitter, packet loss, or temporary unreachability into the dependency edge to create interaction timing anomalies or observable changes in the interaction topology. Through these logical conflict induction conditions, typical logical conflict phenomena such as dependency wait propagation, cascading blocking, and call chain timeout propagation can be reproduced with a small number of test entities. Anomaly propagation or convergence behavior is preferably reflected through timeout thresholds, retry counts, and backoff interval configurations; observable changes in the interaction topology are preferably reflected through retry amplification, circuit breaker triggering, or call path switching. To cover the most comprehensive simulated conflict scenarios with the fewest possible entities, a scenario matrix is ​​used to organize the preset conflict induction conditions. The scenario matrix includes at least: resource dimension combinations of CPU / memory / storage I / O / network / shared lock or shared queue types, logic dimension combinations of serial / fan-out / fan-in / shared downstream / circular dependency topologies, and behavior dimension combinations of concurrency, queue length, timeout and retry parameters. While keeping the size of the test run entity set unchanged, at least one and multiple conflict scenarios are generated by sampling or traversing the parameter combinations in the scenario matrix. These scenarios include at least: single resource contention conflict scenarios, single logic exception conflict scenarios, and resource and logic coupling conflict scenarios, thereby achieving the simulation purpose of covering multiple types of conflict mechanisms with a smaller scale.

[0024] Furthermore, to ensure that conflict scenarios can be repeatedly generated, a scenario identifier and scenario parameter set are configured for each conflict scenario, and a fixed random seed, load curve, duration, and sampling period are set. This allows the same scenario to be repeatedly triggered in the same sandbox environment and maintain a consistent induction process, so as to obtain test entity running data and perform consistent annotation. The random seed is used to control the consistency of random processes such as random delay / random packet loss / random request distribution; the sampling period is used to specify the collection frequency and ensure the comparability of data from different scenarios.

[0025] Specifically, in this embodiment, multiple typical runtime entity templates are pre-built, and a minimal deployment set is formed in the virtualized test sandbox with a small number of test runtime entities. Switchable typical interaction topologies and adjustable parameters of dependency edges are configured. Resource conflict induction conditions and logical conflict induction conditions are applied respectively, and resource dimension, logical dimension, and behavioral dimension parameters are combined and generated in the form of a scenario matrix without increasing the entity volume. This allows for the coverage of multiple conflict mechanisms such as single resource competition, single logical anomaly, and resource-logic coupling with a small test volume. At the same time, by setting scenario identifiers and parameter sets, fixed random seeds, and load curves for each scenario, the conflict scenario triggering process is made consistent and reproducible, achieving the technical effects of reducing scenario construction costs, improving the breadth of conflict coverage, and improving the stability of scenario reuse.

[0026] In one embodiment, resource contention characteristics of multiple running entities are generated based on resource characteristics, and resource conflict detection is performed on the multiple running entities based on the resource contention characteristics and resource conflict characteristics stored in a pre-built conflict characteristic library, generating resource conflict detection results, including: Obtain the runtime environment resource configuration parameters of the cloud computing platform, and obtain the benchmark test environment resource configuration parameters from the conflict feature library. The runtime environment resource configuration parameters preferably include node specifications, entity quotas / limits, I / O and network rate limits, and available margins. The benchmark test environment resource configuration parameters refer to the corresponding parameters of the test sandbox when building the conflict feature library, and are stored in association with resource conflict features. Based on the resource configuration parameters of the operating environment and the benchmark test environment, resource scale difference parameters are determined to at least characterize the differences between the current operating environment and the test environment in terms of resource supply capacity and / or resource allocation constraints. Specifically, the resource scale difference parameters can be obtained in the following ways: for capacity-type resources, the scaling ratio is obtained by "current available upper limit / benchmark available upper limit", and for bandwidth-type resources, the scaling ratio is obtained by "current rate limit or rated bandwidth / benchmark rate limit or rated bandwidth", thereby forming a difference parameter vector or a comprehensive value. Based on the resource scale difference parameter, at least one resource competition value among the resource characteristics of multiple operating entities is corrected to generate corrected resource characteristics. The resource competition value preferably includes occupancy rate, request / supply ratio, queue length, waiting time, etc. The correction process is preferably converted into relative tension according to the scaling ratio or proportionally calibrated to the threshold. Based on the modified resource characteristics of multiple operating entities, resource competition characteristics are generated to represent at least the resource competition relationship between multiple operating entities. The resource competition relationship preferably refers to the joint competition for the same resource object or the same resource pool. Specifically, the resource competition characteristics are preferably obtained by identifying shared resource objects → statistically analyzing the modified occupancy / request of each entity for that object → calculating the competition intensity (such as the degree to which the occupancy exceeds the supply margin or the queuing growth rate). Multiple resource conflict features stored in the conflict feature library are obtained, and feature matching processing is performed on the resource competition features and multiple resource conflict features respectively to obtain the resource conflict feature matching score between the resource competition features and multiple resource conflict features. The feature matching processing preferably includes one or more of similarity calculation (cosine similarity or weighted distance) and rule constraint verification (same resource type / same constraint form), and the matching score is preferably a normalized score. If the maximum value in the resource conflict feature matching score is greater than or equal to a preset first matching threshold, it is determined that multiple running entities have resource conflicts. The resource conflict type of the multiple running entities is determined according to the resource conflict feature corresponding to the maximum value, so as to generate a resource conflict detection result based on the resource conflict type. The first matching threshold is configurable and can be adaptively adjusted according to the noise level. The resource conflict type includes at least port conflict, CPU quota squeezing, memory contention, I / O hotspot or network bandwidth contention. If the matching scores of multiple resource conflict features are all less than the first matching threshold, it is determined that there is no resource conflict among the multiple running entities and the corresponding resource conflict detection results are generated.

[0027] Specifically, in this embodiment, resource scale difference parameters are obtained by acquiring the resource configuration parameters of the operating environment and comparing them with the resource configuration parameters of the benchmark test environment. Then, at least one resource competition value in the resource features is corrected based on the difference parameters. Subsequently, resource competition features representing the resource competition relationship between operating entities are generated from the corrected resource features. These features are then matched, scored, and thresholded with resource conflict features in the conflict feature library. The resource conflict type can be determined based on the feature corresponding to the maximum matching score. This enables the resource-side discrimination to adapt to different resource supply capacity / allocation constraints and output typified results, improving the accuracy and environmental adaptability of resource conflict matching and enhancing the interpretability of the detection results.

[0028] In one embodiment, logical association features of multiple running entities are generated based on logical characteristics, and logical conflict detection is performed on the multiple running entities based on the logical association features and logical conflict features stored in the conflict feature library, generating logical conflict detection results, including: The system obtains the number of currently running entities and their dependency complexity in the current operating environment of the cloud computing platform. Based on these factors, it generates a correlation metric for the currently running entities. The dependency complexity refers to the complexity of the interaction topology and preferably includes one or more of the following indicators: average out-degree / in-degree, average length of dependency chains, presence of cycles, and proportion of shared downstream entities. The correlation metric is used to comprehensively characterize the entity size and dependency complexity and can be obtained by normalizing the number of entities and weighting and summing the complexity indicators. The number of benchmark test run entities and the dependency complexity of benchmark test run entities are obtained from the conflict feature library. Based on the number of benchmark test run entities and the dependency complexity of benchmark test run entities, test run entity association metrics are generated. The benchmark values ​​are associated with and stored with the corresponding logical conflict features. Based on the current running entity association metric and the test running entity association metric, a logical scale difference parameter is determined to at least characterize the difference between the current running environment and the test environment in terms of the degree of association of running entities. The logical scale difference parameter is preferably used to characterize the topology difference and timing jitter difference caused by scale changes. The logic detection correction parameters are determined based on the logic scale difference parameters. The logic detection correction parameters include at least: structural tolerance parameters and temporal tolerance window parameters. The structural tolerance parameters refer to the allowable deviation range of the topology at the node / edge level (such as allowing a small number of newly added or missing edges); the temporal tolerance window parameters refer to the allowable time jitter range of event alignment, and preferably increase with the increase of logic scale difference. Based on the logical characteristics of multiple running entities, an interaction topology and interaction timing information between multiple running entities are constructed, and logical association features are generated according to the interaction topology and interaction timing information. The interaction topology is preferably a graph composed of entity nodes and dependent edges; the interaction timing information preferably includes edge-level call delay distribution, timeout / retry statistics, queuing statistics, and event sequence relationships. The interaction topology is modified based on the structural tolerance parameter to obtain the modified interaction topology, and the interaction timing information is modified based on the timing tolerance window parameter to obtain the modified interaction timing information. The topology modification preferably includes filtering low-confidence edges, merging equivalent dependency edges, or reducing the weight of occasional edges; the timing modification preferably includes merging events within the alignment window, smoothing short jitters, and truncating or reducing the weight of abnormal spikes. Based on the modified interaction topology and modified interaction temporal information, modified logical association features are generated. Based on the modified logical association features, feature matching processing is performed with multiple logical conflict features to obtain the logical conflict feature matching score between the modified logical association features and multiple logical conflict features. The logical conflict feature matching score is preferably determined by both similarity calculation and structural / temporal constraints. If the maximum value in the logical conflict feature matching score is greater than or equal to the preset second matching threshold, it is determined that multiple running entities have logical conflicts. The logical conflict type of the multiple running entities is determined according to the logical conflict feature corresponding to the maximum value, so as to generate a logical conflict detection result based on the logical conflict type. The logical conflict type includes at least one or more of the following: cascading timeout, retry amplification, ring wait / potential deadlock, and shared downstream back pressure diffusion. If the matching scores of multiple logical conflict features are all less than the second matching threshold, it is determined that there is no logical conflict among the multiple running entities and the corresponding logical conflict detection results are generated.

[0029] Specifically, in this embodiment, a correlation metric for the current running entity is generated based on the number of current running entities and dependency complexity. This metric is then compared with the correlation metric for the benchmark running entity to obtain a logical scale difference parameter. This parameter determines logical detection correction parameters, including structural tolerance parameters and timing tolerance window parameters. Logical features are then analyzed to construct interaction topology and interaction timing information, generating logical correlation features. The interaction topology and interaction timing are corrected according to structural tolerance and timing tolerance windows, respectively. Based on the corrected logical correlation features and logical conflict features, a matching score is obtained and a threshold is determined. The type of logical conflict can be identified, enabling the logical detection side to tolerate and calibrate scale changes and normal timing jitter. Simultaneously, a typified discrimination result is output, improving the stability and accuracy of logical conflict matching and enhancing usability across scale scenarios.

[0030] In one embodiment, based on complex conflict evidence, root cause determination processing is performed on multiple complex conflict entities to generate a root cause determination result, including: The complex conflict evidence is analyzed to obtain orchestration event data and logical observation data corresponding to multiple complex conflict entities. Complex conflict evidence refers to a set of evidence that can simultaneously reflect resource-side anomalies and logical-side anomalies, including at least one or more of the following: conflict detection results, key indicator time series, call chain tracing summary, alarms / logs, and orchestration change records. Orchestration event data refers to records of orchestration actions such as creation, scaling up / down, migration, restart, configuration changes, and quota adjustments. Logical observation data refers to runtime observation events such as call chains, queuing / lock waiting, and timeouts / retries. The orchestration event data is parsed to obtain the orchestration events, and the orchestration events are associated with the running entity identifiers of multiple composite conflicting entities to generate an orchestration event sequence. The running entity identifiers preferably include instance ID, Pod UID, process ID, or service instance identifier for event attribution. Logical observation data is parsed to obtain logical observation events, and logical observation events are associated with the running entity identifiers of multiple composite conflict entities to generate a logical observation sequence. The logical observation events include at least one or more of the following: call start / end, exception return, timeout, retry, queue start / end, and lock wait start / end. Time alignment processing is performed on the orchestrated event sequence and the logical observation sequence to generate a unified time axis. A causal timeline is then constructed based on the unified time axis to at least characterize the temporal causal relationship between the related actions, state changes and observed anomalies of the complex conflict entity. The causal timeline is preferably obtained by using orchestrated events as candidate causes and anomaly observations as candidate effects on the unified time axis, and statistically analyzing the "anomaly increment after the action and the diffusion of the impact range" within a preset causal window to establish causal associations and their strength. Obtain the logical feature vectors corresponding to multiple composite conflicting entities and perform relation extraction processing to generate a waiting relation structure, so as to at least represent the waiting / dependency propagation relationship between multiple composite conflicting entities. The waiting relation structure refers to the relation structure with "running entity A waiting for running entity B" or "running entity A depends on running entity B" as the edge. The edge attributes include at least the waiting time, blocking ratio or retry amplification coefficient. The relation extraction is preferably obtained through the call chain Span relation, queuing events and lock waiting events statistics. Obtain resource feature vectors corresponding to multiple composite conflict entities and perform relation extraction processing to generate a resource ownership structure, so as to at least characterize the resource occupancy relationship and / or resource exclusive constraint relationship between multiple composite conflict entities. The resource ownership structure refers to the occupancy / exclusivity constraint relationship structure between entities and resource objects. The resource objects include at least one or more of the following: port, volume, thread pool, connection pool, lock or queue. A causal evidence set is generated based on the causal timeline, the waiting relationship structure, and the resource ownership structure. Root cause determination is performed on multiple complex conflict entities based on a causal evidence set to obtain root cause determination results that indicate at least the root cause of the complex conflict. The root cause determination results include at least: triggering factors that characterize the initial source of the complex conflict, mechanism factors that characterize the formation mechanism of the complex conflict, and control points that characterize the priority handling position for the complex conflict.

[0031] Specifically, triggering factors refer to the initial source event or object of the compound conflict, mechanism factors refer to the dominant mechanism of conflict formation and spread, and control points refer to the dependent edge / occupation constraint edge / causal node that takes priority in intervention.

[0032] Specifically, in this embodiment, data indexes are determined based on composite conflict evidence, and orchestrated event data and logical observation data are obtained. The orchestrated events and logical observation events are parsed into event sequences associated with the running entity identifiers, and then time-aligned to generate a unified timeline. Based on this, a causal timeline representing the temporal causal relationship between actions, state changes, and observed anomalies is constructed. At the same time, waiting / dependency propagation relationships are extracted from logical features to form a waiting relationship structure, and resource occupation / exclusivity constraints are extracted from resource features to form a resource ownership structure. Together with the causal timeline, a causal evidence set is formed. Based on the causal evidence set, a root cause determination result containing triggering factors, mechanism factors, and control points is output. This makes the temporal link and relationship propagation path of composite conflicts present as evidence and provides actionable key intervention locations, improving the interpretability and granularity of the root cause determination result and enhancing the directionality of subsequent handling decisions.

[0033] In a preferred embodiment, root cause determination processing is performed on multiple complex conflict entities based on a causal evidence set to obtain root cause determination results that at least indicate the root cause of the complex conflict, including: The trigger candidate set is determined based on the causal timeline, and the candidate that is located at the beginning of a unified time axis and has the greatest contribution to the subsequent propagation of anomalies is determined as the trigger factor. The contribution is preferably obtained by weighting the increment of anomaly indicators (timeout rate / retry rate / queue time, etc.) within a preset observation window after the candidate occurs, the increase in the number of affected entities, and the increase in the propagation path length. The conflict propagation path is determined based on the waiting relationship structure and the resource ownership structure, and the conflict formation mechanism corresponding to the conflict propagation path is determined as the mechanism factor. The preferred conflict propagation path is a chain structure in which the starting entity spreads downstream through the dependent edge / waiting edge or affects multiple entities through the resource occupation constraint. Based on the conflict propagation path, a set of intervention candidates is determined, and the candidate that can block the conflict propagation path and contributes the most to the mitigation of complex conflicts is determined as the control point. The mitigation contribution is preferably obtained by comparing the decrease of abnormal indicators and the convergence speed of the influence range after the action of different candidate control points through simulation evaluation or historical mapping.

[0034] Specifically, in this embodiment, a set of triggering candidates is first determined based on a causal timeline. Then, the candidate located at the beginning of a unified timeline and contributing the most to subsequent abnormal propagation is identified as the triggering factor. This allows the initial source of the complex conflict to be highlighted and prioritized within a unified temporal context. Simultaneously, the conflict propagation path is determined based on the waiting relationship structure and resource ownership structure, and the conflict formation mechanism corresponding to this propagation path is identified as the mechanism factor. This allows the complex conflict to be structurally expressed from the key link of "starting point—propagation—evolution." Furthermore, an intervention candidate set is determined based on the conflict propagation path, and the candidate that can block the conflict propagation path and contributes the most to mitigating the complex conflict is identified as the control point. This ensures that the selection of intervention locations has a clear propagation blocking direction and effect orientation, thereby improving the certainty of triggering source identification, enhancing the interpretability of conflict mechanism characterization, and improving the targeting and operability of control point selection.

[0035] In one embodiment, determining and executing conflict resolution actions for complex conflict entities based on root cause determination results includes: Analyze the root cause determination results to identify triggering factors, mechanistic factors, and control points; Based on control points, the object and location of the action are determined in the causal evidence set. The object of the action includes at least running entities and / or resource objects, and the location of the action indicates at least the dependency edge, occupancy constraint edge and / or causal node that should be intervened first. When the object of the action is a resource object, the resource object includes at least a port, volume, thread pool, connection pool, lock or queue. The location of the action corresponds one-to-one with the control point and is used to limit the boundary of the action. Based on the mechanism factor, a handling strategy is determined to indicate at least the intervention method for the conflict formation mechanism, and a corresponding set of conflict handling actions is generated based on the handling strategy. The handling strategy is preferably a set of intervention methods that match the mechanism, such as adjusting retry / backoff / circuit break for retry amplification, adjusting concurrency and queue or expanding downstream for back pressure diffusion, and releasing / recovering resources or adjusting quota limits for resource squeezing. The timing and / or scope of the response are determined based on the triggering factors. The set of conflict response actions is then parameterized and ordered based on the timing and / or scope to generate a response execution sequence. The timing of the response is preferably immediate or during a business downturn. The scope of the response is preferably limited to a set of key entities or key resource objects along the propagation path. The execution order preferably follows the principle of prioritizing low-risk and reversible actions followed by high-risk actions. Based on the execution sequence, corresponding conflict resolution actions are performed on complex conflict entities.

[0036] Optionally, conflict resolution actions include at least one or any combination of the following: releasing or reclaiming resources, adjusting resource allocation constraints, adjusting waiting / dependencies, adjusting scheduling priorities, terminating / restarting running entities, and rolling back or compensating orchestration actions.

[0037] Specifically, in this embodiment, triggering factors, mechanism factors, and control points are obtained by analyzing the root cause determination results. Based on the control points, the treatment object and treatment location are determined in the causal evidence set. Based on the mechanism factors, a treatment strategy oriented towards the formation mechanism is determined to generate a set of treatment actions. Then, based on the triggering factors, the treatment timing and / or treatment scope are determined. The set of treatment actions is parameterized and ordered to generate a treatment execution sequence and executed accordingly. This enables the treatment to be carried out in an orderly manner around key dependency edges, occupation constraint edges, or causal nodes, with controllable timing and scope, improving the targeting and consistency of the treatment actions, and reducing ineffective intervention and collateral impacts in the treatment process.

[0038] like Figure 3 As shown, in a preferred embodiment, after determining and executing the conflict resolution action for the complex conflict entity based on the root cause determination result, the method further includes: Step 301: In response to the completion of the conflict resolution action, the updated operation data of multiple composite conflict entities are obtained to obtain the updated resource characteristics and update logic characteristics, and the updated resource competition characteristics and update logic association characteristics are generated accordingly. The updated operational data is preferably collected within the first observation window after the processing is completed, and the length of the observation window is configurable. The generation methods of updated resource features and updated logical features can reuse the aforementioned process of "time alignment, noise reduction, normalization and feature extraction". Step 302: Based on the updated resource competition characteristics and updated logical association characteristics, determine whether one or more of the resource conflict and logical conflict still exist among the multiple composite conflict entities; Whether a conflict still exists is preferably determined by re-executing resource conflict detection and logical conflict detection, or by comparing whether the key matching scores and abnormal indicators before and after the handling have continuously fallen below the threshold. Step 303, the response is then based on the updated resource competition characteristics and updated logical association characteristics, the action compensation parameters for multiple composite conflict entities are adjusted to compensate for conflict handling actions and execute them. Among them, the action compensation parameters refer to the adjustable parameters of the intensity, range, duration and rhythm of the action, including at least the release / recovery amount, limit adjustment range, concurrency threshold, queue length, timeout and retry parameters, degradation ratio or migration batch size; the compensation is preferably adjusted gradually according to the preset step size, and the effect is evaluated in the next observation window after each adjustment, so as to reduce the risk of overcompensation. Step 304: If the response is no, it is determined that the conflict between multiple complex conflicting entities has been resolved.

[0039] Specifically, in this embodiment, the updated resource and logical characteristics are analyzed after the conflict resolution action is executed, thereby dynamically adjusting the compensation strategy in the conflict resolution process to ensure that the conflict is effectively resolved or further optimized, improve the accuracy and sustainability of conflict resolution, and enhance the system's adaptability in complex conflict scenarios.

[0040] In an optional embodiment, prior to conflict resolution, the method further includes a step of providing early warning of one or more of compound conflicts, logical conflicts, and resource conflicts through a configured conflict prediction engine, specifically including: The conflict prediction engine builds a prediction model based on historical conflict samples and online running data. The prediction model can optionally use a recurrent neural network model with attention mechanism (such as an improved LSTM) to learn the temporal pattern of conflict occurrence and output the conflict risk. Furthermore, the input features of the prediction model include at least: timestamp information, resource type identifiers (including one or more of port, memory, CPU, and storage), call stack fingerprints, system load metrics (including one or more of CPU utilization, memory usage, and network bandwidth), and business traffic features (including one or more of request volume and concurrency). The prediction model outputs the probability distribution and risk level of conflict, which can be used to provide early warning of potential conflicts within a preset time window (e.g., 1 to 5 minutes in advance).

[0041] Optionally, the conflict prediction engine also has an online feedback self-learning mechanism, which feeds back feedback information such as the conflict handling results and whether the conflict has been resolved to the prediction model for incremental updates to the model parameters, thereby improving the accuracy of subsequent predictions and the adaptability to different scenarios.

[0042] In an optional embodiment, when at least two types of conflicting entities are detected to be competing for the same resource or the same disposal opportunity, a dynamic arbitrator determines the disposal priority of the complex conflicting entities, specifically including: The dynamic arbitrator acquires system status parameters in real time and calculates system load factors. Combined with the business impact assessment results of each conflicting entity, it dynamically adjusts the weight factors of different conflicting entities. The weighting factors include at least one or more of the following: business level weight, system health weight, and dependency chain length weight. Resource scarcity weight and historical conflict record weight can be further introduced to achieve adaptive modification of the arbitration strategy. Furthermore, the dynamic arbitrator calculates the urgency score of each conflicting party based on the dynamically adjusted weighting factors, and introduces a conflict impact scope correction factor to correct the urgency score in order to obtain the final arbitration score. The arbitration result is determined based on the final arbitration score, and the arbitration result is used to select the action to deal with the winning party and the compensation or restriction action to the losing party. This enables the arbitration strategy to adapt in real time when the system load and business impact change, and reduces the probability of misjudgment caused by fixed weights.

[0043] In an optional embodiment, a compensation execution unit is provided to perform conflict resolution actions and compensation actions after the arbitration result is determined, specifically including: The compensation execution unit supports at least one or more of the following basic recovery modes: instant rollback mode, progressive migration mode, and topology reconstruction mode; Among them, the instant rollback mode is suitable for short transactions or stateless service scenarios to restore consistency within a short time; the progressive migration mode is suitable for long transaction scenarios, maintaining service continuity during the migration process through breakpoint resume or segmented migration mechanisms; the topology reconstruction mode can adjust the call topology and resource allocation based on service dependencies in conjunction with service mesh or service governance framework to reduce the risk of hotspots and cascading blocking.

[0044] Optionally, the compensation execution unit also includes an intelligent degradation mode: when resources are determined to be in an extremely strained state, the priority or service quality level of non-core functions is automatically reduced to ensure the operation of core business links; Furthermore, the compensation execution unit also includes a compensation effect evaluation module, which is used to verify key indicators (such as conflict resource availability, response latency, error rate, retry rate, etc.) after the compensation action is executed, thereby confirming the effectiveness of the compensation measures and providing a basis for adjusting the compensation parameters for subsequent actions.

[0045] In one optional embodiment, a resource fingerprint database is set up in the cloud computing platform to store and maintain resource fingerprint information related to conflict identification, root cause determination and arbitration. Specifically, resource fingerprint information includes at least one or more of the following: port number characteristics, memory hash characteristics, process call tree characteristics, CPU affinity identifiers, and storage I / O signatures, to form a multi-dimensional resource profile. The resource fingerprint database may employ a distributed storage architecture to support millisecond-level queries and high-frequency updates, thereby meeting the needs of high sampling frequency and concurrent access by multiple entities. Furthermore, the resource fingerprint database has data lifecycle management capabilities, automatically cleaning up or downsampling and archiving expired fingerprint data to reduce storage consumption and maintain the timeliness and availability of the fingerprint database.

[0046] In one optional embodiment, an event report is automatically generated after the arbitration and compensation execution are completed for audit trail and post-mortem optimization. Specifically, the incident report should include at least one or more of the following: conflict type and severity, list of affected services, impact assessment information, arbitration result and score, type of compensation action and execution time, compensation status, and suggestions for follow-up actions. The suggestions for follow-up actions may include monitoring duration and set of monitoring indicators, as well as optimization suggestions for configuration parameters. This will structure the closed-loop information of "detection-arbitration-handling-verification-optimization" for easy model self-learning and strategy iteration.

[0047] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0048] like Figure 4 As shown, embodiments of this application also provide a conflict coordination device, the device comprising: The feature generation module is used to obtain the entity operation data of multiple running entities in the cloud computing platform, and generate resource features and logical features of multiple running entities based on the entity operation data. The resource conflict detection module is used to generate resource competition characteristics of multiple running entities based on resource characteristics, and to perform resource conflict detection on multiple running entities based on the resource competition characteristics and the resource conflict characteristics stored in the pre-built conflict characteristic library, and generate resource conflict detection results. The logical conflict detection module is used to generate logical association features of multiple running entities based on logical characteristics, and to perform logical conflict detection on multiple running entities based on the logical association features and the logical conflict features stored in the conflict feature library, and generate logical conflict detection results. The composite determination module is used to determine multiple running entities as composite conflict entities if both the resource conflict detection result and the logical conflict detection result indicate that there is a conflict, and to generate composite conflict evidence for multiple composite conflict entities based on the resource conflict detection result and the logical conflict detection result. The conflict resolution module is used to perform root cause determination processing on multiple conflicting entities based on composite conflict evidence, generate root cause determination results, and determine and execute conflict resolution actions for multiple composite conflicting entities based on the root cause determination results.

[0049] For a description of the features in the embodiment corresponding to the conflict coordination device, please refer to the relevant description of the embodiment corresponding to the conflict coordination method, which will not be repeated here.

[0050] like Figure 5As shown, embodiments of this application also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the conflict coordination method embodiments described above.

[0051] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the conflict coordination method embodiments described above when it is run.

[0052] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0053] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the conflict coordination method embodiments described above.

[0054] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the conflict coordination method embodiments described above.

[0055] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0056] The foregoing has provided a detailed description of a conflict coordination method, apparatus, electronic device, and storage medium provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A conflict coordination method, characterized in that, include: Obtain entity operation data of multiple running entities in a cloud computing platform, and generate resource characteristics and logical characteristics of multiple running entities based on the entity operation data; Based on the resource characteristics, resource contention characteristics of multiple running entities are generated, and resource conflict detection is performed on multiple running entities based on the resource contention characteristics and resource conflict characteristics stored in the pre-built conflict characteristic library, generating resource conflict detection results; Based on the logical features, generate logical association features for multiple running entities, and based on the logical association features and the logical conflict features stored in the conflict feature library, perform logical conflict detection on multiple running entities to generate logical conflict detection results. In response to the fact that both the resource conflict detection result and the logical conflict detection result indicate the existence of a conflict, the multiple running entities are determined to be composite conflict entities, and composite conflict evidence of the multiple composite conflict entities is generated based on the resource conflict detection result and the logical conflict detection result. Based on the composite conflict evidence, root cause determination processing is performed on multiple composite conflict entities to generate root cause determination results, and conflict resolution actions for multiple composite conflict entities are determined and executed based on the root cause determination results.

2. The conflict coordination method according to claim 1, characterized in that, The cloud computing platform contains entity operation data for multiple running entities, and generates resource characteristics and logical characteristics for multiple running entities based on the entity operation data, including: Obtain entity operation data of multiple running entities within a preset time window, wherein the entity operation data includes at least resource operation data representing resource usage status and logical operation data representing the interaction relationship between running entities; The resource operation data and the logic operation data are respectively processed by time alignment, noise reduction and normalization to obtain standard resource operation data and standard logic operation data; Based on the standard resource operation data of multiple operating entities, resource characteristics of multiple operating entities are extracted respectively to at least characterize the resource occupancy level, resource allocation constraints and resource change trends of multiple operating entities; Based on the standard logical operation data of multiple running entities, interaction relationship representations of multiple running entities are generated respectively, and logical characteristics of the running entities are determined according to the interaction relationship representations, so as to at least characterize the associated entity object information of the running entities.

3. The conflict coordination method according to claim 1, characterized in that, Before performing resource conflict detection on the plurality of said running entities, the method further includes: Based on the operating environment of the cloud computing platform, virtualization environment parameters are set, and a virtualization test sandbox is constructed according to the virtualization environment parameters. At least two test running entities are deployed in the virtualized test sandbox, and preset conflict induction conditions are applied to the at least two test running entities to generate at least one conflict scenario; Obtain the test entity running data corresponding to the conflict scenario, and annotate the test entity running data to obtain conflict annotation data; Conflict features are generated based on the conflict annotation data, and the conflict features are classified and stored according to resource conflict features and / or logical conflict features to construct the conflict feature library.

4. The conflict coordination method according to claim 1, characterized in that, The process of generating resource contention characteristics for multiple running entities based on the resource characteristics, and performing resource conflict detection on the multiple running entities based on the resource contention characteristics and resource conflict characteristics stored in a pre-built conflict characteristic library, to generate resource conflict detection results, includes: Obtain the runtime environment resource configuration parameters of the cloud computing platform, and obtain the benchmark test environment resource configuration parameters from the conflict feature library; Based on the resource configuration parameters of the operating environment and the resource configuration parameters of the benchmark test environment, resource scale difference parameters are determined to at least characterize the differences between the current operating environment and the test environment in terms of resource supply capacity and / or resource allocation constraints. Based on the resource scale difference parameter, at least one resource contention value among the resource characteristics of the multiple operating entities is corrected to generate corrected resource characteristics. Based on the modified resource characteristics of the multiple running entities, at least resource competition characteristics representing the resource competition relationship between the multiple running entities are generated; Multiple resource conflict features stored in the conflict feature library are obtained, and feature matching processing is performed between the resource competition feature and the multiple resource conflict features respectively to obtain the resource conflict feature matching score between the resource competition feature and the multiple resource conflict features. If the maximum value in the resource conflict feature matching score is greater than or equal to a preset first matching threshold, it is determined that multiple running entities have resource conflicts, and the resource conflict type of multiple running entities is determined according to the resource conflict feature corresponding to the maximum value, so as to generate a resource conflict detection result based on the resource conflict type; If the matching scores of multiple resource conflict features are all less than the first matching threshold, it is determined that there is no resource conflict among the multiple running entities and a corresponding resource conflict detection result is generated.

5. The conflict coordination method according to claim 1, characterized in that, The step of generating logical association features for multiple running entities based on the logical features, and performing logical conflict detection on the multiple running entities based on the logical association features and logical conflict features stored in the conflict feature library, and generating logical conflict detection results, includes: Obtain the number of currently running entities and the dependency complexity of the currently running entities in the current operating environment of the cloud computing platform, and generate the association metric value of the currently running entities based on the number of currently running entities and the dependency complexity of the currently running entities; The number of benchmark test running entities and the dependency complexity of benchmark test running entities are obtained from the conflict feature library, and test running entity association metrics are generated based on the number of benchmark test running entities and the dependency complexity of benchmark test running entities. Based on the current running entity association metric and the test running entity association metric, a logical scale difference parameter is determined to at least characterize the difference between the current running environment and the test environment in terms of running entity association. Based on the logic scale difference parameters, logic detection correction parameters are determined, and the logic detection correction parameters include at least: structural tolerance parameters and timing tolerance window parameters. The logical characteristics of multiple running entities are parsed to construct the interaction topology and interaction timing information between the multiple running entities, and the logical association features are generated based on the interaction topology and the interaction timing information. The interaction topology is modified based on the structural tolerance parameter to obtain the modified interaction topology, and the interaction timing information is modified based on the timing tolerance window parameter to obtain the modified interaction timing information. Based on the modified interaction topology and the modified interaction timing information, a modified logical association feature is generated, and based on the modified logical association feature, feature matching processing is performed with multiple logical conflict features to obtain the logical conflict feature matching score between the modified logical association feature and the multiple logical conflict features. If the maximum value in the logical conflict feature matching score is greater than or equal to a preset second matching threshold, it is determined that multiple running entities have logical conflicts, and the logical conflict type of multiple running entities is determined according to the logical conflict feature corresponding to the maximum value, so as to generate a logical conflict detection result based on the logical conflict type. If the matching scores of multiple logical conflict features are all less than the second matching threshold, it is determined that there is no logical conflict among the multiple running entities and a corresponding logical conflict detection result is generated.

6. The conflict coordination method according to claim 1, characterized in that, The step of performing root cause determination processing on multiple conflicting entities based on the composite conflict evidence, and generating root cause determination results, includes: Based on the composite conflict evidence, a data index corresponding to the multiple composite conflict entities is determined, and orchestration event data and logical observation data are obtained according to the data index; The orchestration event data is parsed to obtain orchestration events, and the orchestration events are associated with the running entity identifiers of multiple composite conflict entities to generate an orchestration event sequence; The logical observation data is parsed to obtain logical observation events, and the logical observation events are associated with the running entity identifiers of multiple composite conflict entities to generate a logical observation sequence; The orchestrated event sequence and the logical observation sequence are time-aligned to generate a unified timeline, and a causal timeline is constructed based on the unified timeline to at least characterize the temporal causal relationship between the related actions, state changes and observed anomalies of the composite conflict entity. Obtain the logical features corresponding to multiple composite conflict entities and perform relation extraction processing to generate a waiting relation structure, so as to at least characterize the waiting / dependency propagation relationship between multiple composite conflict entities; The resource characteristics corresponding to the multiple composite conflict entities are obtained and the relationship extraction process is performed to generate a resource ownership structure, so as to at least characterize the resource occupation relationship and / or resource exclusive constraint relationship between the multiple composite conflict entities; Based on the causal timeline, the waiting relationship structure, and the resource ownership structure, a causal evidence set is generated. Based on the causal evidence set, root cause determination processing is performed on multiple composite conflict entities to obtain root cause determination results that indicate at least the root cause of the composite conflict. The root cause determination results include at least: triggering factors characterizing the initial source of the composite conflict, mechanism factors characterizing the formation mechanism of the composite conflict, and control points characterizing the priority handling position for the composite conflict.

7. The conflict coordination method according to claim 6, characterized in that, The step of determining and executing conflict resolution actions for multiple composite conflict entities based on the root cause determination result includes: Analyze the root cause determination results to identify the triggering factor, the mechanism factor, and the control point; Based on the control points, the disposal objects and disposal locations are determined in the causal evidence set. The disposal objects include at least running entities and / or resource objects, and the disposal locations indicate at least the dependency edges, occupancy constraint edges, and / or causal nodes that require priority intervention. Based on the aforementioned mechanism factors, a response strategy is determined to at least indicate an intervention method targeting the conflict formation mechanism, and a corresponding set of conflict response actions is generated based on the response strategy. Based on the triggering factor, the timing and / or scope of the action are determined, and the set of conflict handling actions is parameterized and ordered for execution based on the timing and / or scope of the action to generate a handling execution sequence; Based on the aforementioned execution sequence, corresponding conflict resolution actions are performed on the composite conflict entity.

8. A conflict coordination device, characterized in that, include: The feature generation module is used to generate resource features and logical features of multiple running entities in a cloud computing platform based on the entity running data. The resource conflict detection module is used to generate resource competition characteristics of multiple running entities based on the resource characteristics, and to perform resource conflict detection on multiple running entities based on the resource competition characteristics and the resource conflict characteristics stored in the pre-built conflict characteristic library, and generate resource conflict detection results. The logical conflict detection module is used to generate logical association features of multiple running entities based on the logical features, and to perform logical conflict detection on multiple running entities based on the logical association features and the logical conflict features stored in the conflict feature library, and generate logical conflict detection results. The composite determination module is used to determine that multiple running entities are composite conflict entities in response to both the resource conflict detection result and the logical conflict detection result indicating the existence of a conflict, and to generate composite conflict evidence of multiple composite conflict entities based on the resource conflict detection result and the logical conflict detection result. The conflict resolution module is used to perform root cause determination processing on multiple composite conflict entities based on the composite conflict evidence, generate root cause determination results, and determine and execute conflict resolution actions on multiple composite conflict entities according to the root cause determination results.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the conflict coordination method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the conflict coordination method as described in any one of claims 1 to 7.

Images