A power communication network micro-service state aggregation observation method based on distributed link tracking

Abstract

The present application relates to the field of microservice monitoring, and more particularly to a power communication network microservice state aggregation observation method based on distributed link tracking, in response to a business request initiation, a state information carrier propagating with a call chain is initialized; in the call chain propagation process, each microservice node obtains upstream reference state information from the carrier, compares the state information with the node state information to determine state difference information; the preset causal judgment logic is called to determine the source of the state difference information, and the causal attribution mark of self-caused or transmitted is written into the carrier and transmitted after labeling; the carrier is collected at the predetermined stage of the call chain, and the state difference information of the self-caused source is screened out to generate an aggregation observation output. The present application can complete the causal decoupling and root cause positioning synchronously in the call chain propagation process, and improve the troubleshooting efficiency of the power communication network microservice.

Description

Technical Field

[0001] This application relates to the field of microservice monitoring technology, and in particular to a method for aggregated observation of the state of microservices in power communication networks based on distributed link tracing. Background Technology

[0002] In power communication networks, with the widespread deployment of microservice architectures, a single business request often requires chained calls through multiple microservice instances. To ensure the reliability of power communication network operations, maintenance personnel need to observe the operational status of each microservice node in the call chain and locate anomalies. Distributed tracing technology is currently widely used in call relationship analysis of microservice architectures, capable of reconstructing the complete call path of a business request and recording the call time and error codes of each node. However, in the operational practice of power communication networks, relying solely on call time and error codes is often insufficient to accurately pinpoint the root cause of a fault. A typical dilemma is that when a node in the call chain experiences an anomaly, the anomaly may be caused by the node itself or by an anomaly input transmitted from an upstream node. The corresponding handling strategies for these two situations are completely different. However, existing tracing systems lack a mechanism to identify the source of the anomaly during the call chain propagation process, making it difficult for maintenance personnel to make accurate judgments.

[0003] Secondly, when an upstream node has already experienced a severe state shift, the downstream node, operating under distorted input, may also experience a state shift that mirrors the upstream. In this case, the change in the downstream state relative to the upstream is actually very small, rendering the previous method of determining the source of the anomaly based on the differences between upstream and downstream nodes ineffective, and the true root cause node of the initial shift easily overlooked. Summary of the Invention

[0004] To address one or more problems in the existing technology, the main objective of this application is to provide a method for aggregated observation of the state of microservices in power communication networks based on distributed link tracing.

[0005] To achieve the aforementioned objectives, this application proposes a method for microservice state aggregation and observation in power communication networks based on distributed link tracing, the method comprising: In response to the initiation of a business request, a state information carrier that propagates along the call chain is initialized. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. During the call chain propagation process, upstream reference state information is obtained from the state information carrier; After obtaining the upstream reference status information, the current node status information generated by the current microservice node's execution of processing is obtained, and the current node status information is compared with the upstream reference status information. Based on the comparison result, the status difference information is determined. The preset causal determination logic is invoked to determine the source of the state difference information, and the corresponding causal attribution identifier is marked on the state difference information according to the determination result. Based on the annotation results, the annotated state difference information is written into the state information carrier and transmitted along the call chain or triggered to collect information. During a predetermined phase of the call chain, the state information carrier is collected, and state difference information originating from the state information carrier is filtered out based on the causal attribution identifier. Based on the filtering results, aggregated observation output is generated.

[0006] This application also provides a power communication network microservice state aggregation and observation system based on distributed link tracing, including: The response module is used to respond to the initiation of a business request and initialize a state information carrier that propagates along the call chain. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. The calling module is used to obtain upstream reference state information from the state information carrier during the call chain propagation process; The acquisition module is used to acquire the current node status information generated by the current microservice node after acquiring the upstream reference status information, and compare the current node status information with the upstream reference status information to determine the status difference information based on the comparison result. The judgment module is used to call the preset causal judgment logic to judge the source of the state difference information, and to mark the corresponding causal attribution identifier for the state difference information according to the judgment result. The input module is used to write the annotated state difference information into the state information carrier based on the annotation results, and to pass it along the call chain or trigger information collection. The output module is used to collect the state information carrier at a predetermined stage of the call chain, filter out the state difference information from the state information carrier based on the causal attribution identifier, and generate aggregated observation output based on the filtering results.

[0007] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.

[0008] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0009] This application's embodiment of the power communication network microservice state aggregation observation method, based on distributed link tracing, achieves the binding of microservice state information with call chain paths within the framework of distributed link tracing. This eliminates the reliance on post-hoc correlation between independent monitoring systems and call chain systems for state observation, thus resolving issues of timing misalignment and data alignment. More importantly, this method introduces a causal attribution determination mechanism during call chain propagation, enabling the analysis and labeling of the source type of each node's state change at the moment it occurs. This transforms the root cause localization problem, which originally required offline inference, into a synchronous decoupling problem during propagation. Finally, by filtering out state difference information originating from its own source, non-root cause alarms generated by propagation effects on the call chain are excluded. This ensures that the output aggregation observation results directly point to the truly faulty node requiring attention, rather than a set of all nodes exhibiting abnormal behavior across the entire chain, thereby improving the efficiency and accuracy of power communication network microservice fault diagnosis. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a method for monitoring the state aggregation of microservices in a power communication network based on distributed link tracing, according to an embodiment of this application. Figure 2 This is a flowchart illustrating another embodiment of the power communication network microservice state aggregation observation method based on distributed link tracing. Figure 3 This is a flowchart illustrating another embodiment of the power communication network microservice state aggregation observation method based on distributed link tracing in this application; Figure 4 This is a schematic block diagram of a power communication network microservice state aggregation and observation system based on distributed link tracing, according to an embodiment of this application. Figure 5 This is a schematic block diagram of the structure of a computer device according to an embodiment of this application.

[0011] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0013] Reference Figure 1 This application provides a method for microservice state aggregation and observation of power communication networks based on distributed link tracing, the method comprising: S1. In response to the initiation of a business request, initialize a state information carrier that propagates along the call chain. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. S2. During the call chain propagation process, upstream reference state information is obtained from the state information carrier; S3. After obtaining the upstream reference status information, obtain the current node status information generated by the current microservice node's execution processing, compare the current node status information with the upstream reference status information, and determine the status difference information based on the comparison result. S4. Invoke the preset causal determination logic to determine the source of the state difference information, and mark the corresponding causal attribution identifier for the state difference information according to the determination result. S5. Based on the annotation results, write the annotated state difference information into the state information carrier, and pass it along the call chain or trigger information collection. S6. In the predetermined stage of the call chain, the state information carrier is collected, and the state difference information with self-caused origin is filtered out from the state information carrier according to the causal attribution identifier, and the aggregated observation output is generated based on the filtering results.

[0014] As described in steps S1-S3 above, this method begins with the initiation of a service request. In a power communication network, a typical service request can be a remote control operation command, the transmission of relay protection data, or a telemetry data retrieval initiated by the power grid dispatching system. Regardless of the specific service type, when the request enters the first node of the microservice call chain, the method first initializes a state information carrier that propagates along the call chain. This state information carrier can be understood as a data structure or object that accompanies the request hop-by-hop throughout the call chain. It has two functions: first, to collect and carry the state information of each microservice node in the call chain, such as the CPU utilization, memory usage, request processing time, and communication protocol parsing results of each node; second, to attach a causal attribution identifier to each piece of state information to mark the source type of the state information. The existence of this carrier allows the state information to be naturally bound to the call chain path, avoiding the cumbersome post-event correlation operations required by traditional solutions through additional time-series databases or log systems. After the carrier initialization is complete, when a request enters a microservice node in the call chain, that node reads the upstream reference state information from the transmitted state information carrier. The upstream reference state information refers to the state data left by nodes in the call chain that have processed the request before the current node; more specifically, it refers to the state snapshot written to the carrier by the current node's direct upstream node after processing the request. This upstream reference state information forms the benchmark for subsequent judgments. After obtaining the upstream reference, the current node needs to obtain its own node state information generated during its processing. This node state information includes various operational status indicators of the node after completing the processing of this business request. Subsequently, the node state information is compared dimension by dimension with the upstream reference state information, and the differences between the two are calculated. This difference reflects how the current node's state has changed relative to the upstream during the request processing process.

[0015] As described in steps S4-S6 above, after the difference calculation is completed, the pre-set causal determination logic is invoked to analyze the aforementioned state difference information. The causal determination logic is essentially a set of rules or a determination strategy used to answer a core question: is the state change observed by the current node caused by the node's own behavior, or by some adverse factor transmitted from upstream? For example, if the CPU utilization of the current node increases significantly, but the amount of requested input data does not increase, then this change is more likely to be determined as self-caused; conversely, if the amount of input data increases synchronously and the message format is abnormal, then this change is more likely to be determined as transmitted. After the determination is completed, the corresponding causal attribution identifier is marked on the state difference information according to the determination result, that is, labeled as self-caused or transmitted. After the labeling is completed, the state difference information with the causal attribution identifier is written into the state information carrier. At this time, the state information carrier has accumulated the state difference of the current node and its causal attribution, and then the carrier continues to pass it down the call chain to downstream nodes. If the current node is already the end node of the call chain, it will no longer pass, but will trigger the information collection process. Once all nodes in the call chain have completed the above processing, or even if the call chain is not yet complete but a certain reporting condition is met, the method collects all records carried in the state information carrier. Then, based on the causal attribution identifier on each record, it filters out only those state difference information marked as originating from the root cause, discarding records indicating the source of the fault. The filtered originating information is further processed and presented to generate aggregated observation output for operations personnel. This output can be a service topology diagram marked with root cause nodes, or a report containing details of anomaly dimensions, used to guide operations personnel in handling the true source of the fault.

[0016] As described above, this method, within the framework of distributed tracing, binds microservice state information to the call chain path, eliminating the reliance on post-hoc correlation between independent monitoring and call chain systems for state observation, thus resolving timing misalignment and data alignment issues. More importantly, by introducing a causal attribution mechanism during call chain propagation, this method allows the state changes of each node to be analyzed and their source type labeled simultaneously, transforming the root cause localization problem, which previously required offline inference, into a synchronous decoupling problem during propagation. Finally, by filtering out state difference information originating from their own sources, non-root cause alarms caused by propagation effects on the call chain are eliminated, ensuring that the aggregated observation results directly point to the truly faulty node requiring attention, rather than a set of all nodes exhibiting abnormal behavior across the entire chain. This improves the efficiency and accuracy of microservice fault diagnosis in power communication networks.

[0017] Reference Figure 2In one embodiment, the method of initializing a state information carrier that propagates along the call chain in response to the initiation of a business request includes: S11. In response to the initiation of the service request, generate a unique tracking identifier corresponding to the service request; S12. Create an empty status information carrier object based on the unique tracking identifier; S13. In the state information carrier object, a first storage area and a second storage area are divided, wherein the first storage area is used to record the state information of each microservice node on the call chain, and the second storage area is used to record the causal attribution identifier corresponding to each state information. S14. Write the status information of the starting microservice node of the business request as the initial status information into the first storage area, and preset the causal attribution identifier corresponding to the initial status information as the transmission source. S15. Bind the status information carrier object to the call chain context of the business request, so that the status information carrier can be propagated between microservice nodes along with the call chain context of the business request.

[0018] As described above, when a business request enters the first node of the microservice call chain, this method first generates a unique trace identifier for the request. This identifier is a globally unique string or number, which, in the microservice architecture of the power communication network, can uniquely identify this business request throughout the entire distributed environment. This trace identifier acts like a pass for each business request; all subsequent call records, logs, and status information related to this request can be associated with this identifier. After generating the unique trace identifier, the method creates an empty status information carrier object based on this identifier. This carrier object can be understood as a container or a data packet; it is empty at creation, and as the request is passed hop by hop in the call chain, each microservice node continuously writes information into it. After the carrier object is created, two independent storage areas are partitioned within it. The first storage area records the status information of each microservice node in the call chain. The second storage area records the causal attribution identifier corresponding to each piece of status information. These two storage areas correspond to each other, ensuring a one-to-one mapping between status information and its causal attribution identifier. Once the carrier object is obtained, the state information of the starting microservice node of the business request before or after processing is written into the first storage area mentioned above as the initial state information.

[0019] Meanwhile, since the starting node is the first node in the entire call chain, its state changes can only originate from the initiating end of the request, and there is no abnormal input passed from upstream nodes. Therefore, the method presets the causal attribution identifier corresponding to the starting node's state information as the transmission source. This preset is not for the attribute judgment of the starting node itself, but to provide a unified reference benchmark for subsequent state comparisons among various nodes. After completing the initial information filling of the carrier object, the entire carrier object is bound to the call chain context of the business request. This binding operation ensures that the carrier object can automatically propagate among the microservice nodes through the call chain context transmission mechanism, so that each node can access the same carrier object when processing the request, without sharing data through additional transmission paths or storage media. Through the cooperation of the above steps, the initialization phase completes the construction of a fully functional carrier object. This carrier object not only has a unique identifier, but its internal structure also clearly distinguishes between the state information storage area and the causal attribution identifier storage area, and pre-writes the starting node's state information as the reference benchmark for the entire call chain. More importantly, the binding of the carrier object to the call chain context ensures that it can be propagated with the request without the recipient being aware of it. Each node only needs to follow the normal call chain context processing logic to obtain access to the carrier object, without the need for additional transmission mechanism development.

[0020] refer to Figure 3 In one embodiment, the method of comparing the current node's state information with the upstream reference state information and determining the state difference information based on the comparison result includes: S31. Parse the status information of this node, and extract a first set of status values ​​including status dimensions based on the parsing results. The status dimensions include at least one of resource utilization rate, request processing latency, business processing result code, and communication protocol exception count. S32. Parse the upstream reference state information and extract a second state value set with the same state dimension as the first state value set; S33. For each state dimension, calculate the difference between the state value of the dimension in the first state value set and the state value of the corresponding dimension in the second state value set; S34. Based on the settlement results, the differences of all dimensions calculated are combined into a state change vector, which serves as the state difference information.

[0021] As mentioned above, after obtaining the current node's state information and the upstream reference state information, scientifically calculating the difference between the two is the data foundation for subsequent causal attribution determination. This difference is not a simple qualitative judgment of "good" or "bad," but rather a quantitative comparison based on specific state dimensions. First, let's discuss the processing of the current node's state information. The current node's state information includes various operational state data of the microservice node after completing business processing. This data can cover multiple aspects such as resource utilization, time consumption, business execution results, and communication layer interactions. To effectively compare with the upstream, the current node's state information needs to be parsed to extract specific state values, which constitute the first set of state values. The selection of state dimensions can be set according to the actual operational characteristics of the power communication network microservices. A common approach includes the following dimensions: resource utilization reflects the node's load at the computing resource level; request processing latency reflects the time consumed by the node in processing business requests; business processing result code reflects the success or failure of the node's business logic execution; and communication protocol anomaly count reflects the frequency of anomalies that occur when the node interacts with power equipment or other microservices via protocol. After parsing and extracting the current node's state information, the next step is to process the upstream reference state information. This upstream reference state information contains the state data written by nodes in the call chain that processed the request before the current node. To ensure the validity of the comparison, state values ​​with the same dimensions as those in the first state value set need to be extracted from the upstream reference state information; these values ​​constitute the second state value set. The key to this step is dimension alignment; only comparisons between identical state dimensions are meaningful. Otherwise, comparisons between different dimensions have neither physical meaning nor quantification. After obtaining the two state value sets, for each state dimension, the difference between the value of that dimension in the first state value set and the corresponding value in the second state value set is calculated. This difference reflects the extent to which the state dimension has changed relative to the upstream node during the current microservice node's request processing. Finally, the differences calculated for each dimension are combined to form a state change vector. This vector serves as a structured representation of state difference information. A vector form is used because it can simultaneously carry change information across multiple dimensions, and in subsequent steps, the causal judgment logic can directly perform comprehensive analysis and judgment on this vector without needing to process the information of each dimension separately before summarizing. This structured difference information provides clear, complete, and quantifiable data input for subsequent causal attribution determination.

[0022] For example, suppose a remote control operation is being performed in a power communication network. This operation sequentially passes through the authorization verification service, the device status check service, and the command issuance service. When the call chain reaches the device status check service, this node is the current microservice node, and its preceding authorization verification service is the upstream node.

[0023] After processing the remote control request, the current node generates its own state information and reads the upstream reference state information from the state information carrier. A comparison is then performed item by item on the data of the two nodes in the same state dimension. The comparison results for each state dimension are shown in Table 1 below: Table 1

[0024] The differences in the three dimensions are combined to form a three-dimensional state change vector, which represents the state difference information of the current node. For comparison, Table 2 below shows the differences between the state difference information determined by this method and the information provided by existing typical link tracing systems in the same scenario: Table 2

[0025] The above comparison shows that existing link tracing systems can only present the call times and possible error codes of the three nodes individually under the same scenario. They lack the ability to collect state dimensions such as CPU utilization, request processing latency, and communication protocol anomaly counts, and cannot quantify changes in these dimensions into comparable difference data. This method, however, extracts multi-dimensional state values, calculates differences dimension by dimension, and combines them into a state change vector. This transforms the originally scattered and incomparable state information into structured data in a unified format, providing clear input for subsequent causal attribution determination. Operations personnel can directly assess the degree of state change of each node based on the specific differences in each dimension of the state change vector, without having to manually compare multiple monitoring systems.

[0026] In one embodiment, the method of filtering out self-caused state difference information from the state information carrier based on the causal attribution identifier and generating aggregated observation output based on the filtering results includes: During the predetermined stage of the call chain, all records written in the state information carrier are read. Each record includes state difference information of a microservice node and the causal attribution identifier corresponding to the state difference information. Iterate through all the records, and for each record, check whether the causal attribution identifier of each record is a self-caused source; Based on the results of the inspection, records with causal attribution identified as self-caused are retained as valid records, while records with causal attribution identified as transitive are marked as records to be filtered. Extract the state difference information of each microservice node from the valid records, and arrange the extracted state difference information in the order of nodes according to the topological order of the call chain to form a root cause candidate list. Based on the root cause candidate list, an aggregated observation view of the power communication network is generated. The aggregated observation view includes at least one of the following: a service topology graph labeled with self-caused source nodes, a causal path highlighting along the call chain, and abnormal state dimension details for each self-caused source node.

[0027] As described above, during the pre-defined phase of the call chain, whether the entire call chain has been executed or a certain triggering condition has been met, the system first reads all records written to the state information carrier. These records are written by each microservice node in the call chain during hop-by-hop propagation, and each record contains the state difference information of that node and the corresponding causal attribution identifier. In other words, at this point, the system has the state change of each node in the entire call chain relative to its upstream node, and the label indicating whether this change is a self-caused or transitive source. After reading all records, these records are traversed one by one, checking whether the causal attribution identifier in each record is a self-caused source. This traversal process is essentially a filtering and judgment of each node in the entire call chain. Based on the results of the traversal and checking, the records are processed separately. Those records with causal attribution identifiers indicating self-caused sources are retained as valid records because the nodes corresponding to these records are nodes that may contain real sources of failure. Records with causal attribution indicating a source of transmission are marked as pending filtering. While they also reflect the appearance of anomalies, these anomalies are propagated from upstream and are not the root cause of the fault themselves. After filtering the records, the state difference information of each microservice node is extracted from the valid records. Then, according to the topological order of the call chain, this state difference information is arranged according to the order of the nodes in the call chain, forming a root cause candidate list. The value of this list lies in the fact that it not only tells the operations personnel which nodes are likely root causes, but also reveals the distribution of these potential root causes in the call chain through the order of the nodes, which is of reference significance for determining the direction of fault propagation. Finally, an aggregated observation view of the power communication network is generated based on the root cause candidate list. This aggregated observation view can be presented in several forms. One form is to mark the self-originating nodes on the service topology map, allowing the operations personnel to intuitively see which nodes are identified as potential fault sources. Another form is to highlight the causal path propagating along the call chain, that is, to show the path of state anomalies propagating from upstream to downstream along the call chain on the topology map in a highlighted manner, helping the operations personnel understand the fault propagation process. Another approach is to list the details of the abnormal status dimension for each self-supplied source node, specifying which status dimension of a particular node is experiencing a problem, such as excessive CPU utilization or an increase in the number of communication protocol anomalies. These three presentation methods can be used independently or in combination, depending on actual operational needs.

[0028] In this embodiment, each microservice node completes its own causal attribution determination and assigns a label during the propagation process. However, these labels are scattered and remain locally on each node. If operations personnel were to check the labels node by node, the efficiency would be very low, and it would be difficult to form an overall judgment. This filtering and output method makes key arrangements in several aspects. First, it collects the causal attribution identifiers of all nodes in the entire call chain at a predetermined stage, converging the scattered information in one place. Second, it filters records with self-caused origins through traversal checks, while marking records with other origins as pending filtering rather than discarding them directly. This design preserves the integrity of the information and leaves room for possible subsequent traceability analysis. Third, it does not simply list the nodes with self-caused origins, but arranges these nodes into a list according to the topological order of the call chain. This sorting action is meaningful because in the call chain of the power communication network, the failure of an upstream node often propagates to the downstream. Arranging them according to the topological order helps operations personnel determine whether there is an upstream-downstream causal relationship between multiple self-caused origin nodes. Fourth, he presents the screening results in a variety of visualization formats, including topology icon annotations, causal path highlighting, and anomaly dimension details. Different types of presentation serve different troubleshooting needs, and operations and maintenance personnel can choose the most suitable information view according to the actual situation.

[0029] In one specific embodiment, assume a call chain contains three microservice nodes: a data acquisition service, a protocol parsing service, and a data forwarding service. During the processing of a business request, the data acquisition service, due to an improperly configured cache eviction policy, causes its CPU utilization to rise from the normal 40% to 85%. Because the causal determination logic of this node has some imperfect boundary conditions, this anomaly is incorrectly marked as a transit source rather than a self-caused source and written into the state information carrier. Subsequently, the request enters the protocol parsing service. When this node reads the upstream reference state information from the state information carrier, it obtains distorted state data from the data acquisition service, i.e., a CPU utilization of 85%. After processing the request, the protocol parsing service's own state is a CPU utilization of 88%. The state difference calculated by this node is only 3 percentage points, far below the anomaly detection threshold. Based on this, the causal determination logic believes that it has not introduced a new anomaly and marks its own state as a transit source. On the surface, the causal determination logic of the protocol parsing service is operating normally, but the problem lies in the fact that the reference information it relies on is itself distorted. The request continues to enter the data forwarding service. The node had already read 88% of the upstream references, and its own processed state was 90%, with a first-order difference of only 2 percentage points. It was also labeled as a source of propagation. After the entire call chain propagated, the actual root cause node data collection service, incorrectly labeled as a source, was filtered out in the final screening process. The subsequent two nodes, although correctly labeled, were also filtered out because they were considered sources of propagation. The final aggregated observation output was empty, and the operations personnel received no alerts; a fault was completely missed. This is the serious consequence of causal determination errors accumulating and propagating along the call chain.

[0030] To address the above issues, this embodiment introduces an additional judgment and compensation mechanism based on the first-order difference-based causal determination framework. The core idea of ​​this mechanism is as follows: after a node completes the comparison between its current state information and the upstream reference state information and obtains the state difference information, it does not directly send this first-order difference result into the causal determination logic. Instead, it first evaluates the state difference information itself to determine whether the upstream reference state information has been severely distorted to the point that the first-order difference has lost its reference value. If the evaluation result indicates that the reference has become invalid, the upstream reference is abandoned, and a fixed reference unaffected by the upstream state—namely, the initial baseline state information at the time the business request was initiated—is used to calculate the current node's cumulative offset. This cumulative offset accurately reflects the absolute offset of the current node relative to the initial state of the call chain and is unaffected by the distortion of the upstream node's state. Subsequent steps can make more accurate judgments based on this cumulative offset, thereby preventing causal misjudgments from propagating along the call chain.

[0031] Specifically, the method of comparing the current node's state information with the upstream reference state information and determining the state difference information based on the comparison result further includes: Determine whether the state difference information meets the preset reference failure condition; When the reference failure condition is met, the initial reference state information is obtained, which is the state information at the time the service request is initiated. The current node status information is compared with the initial reference status information to determine the cumulative offset of the current node status information relative to the initial reference status information; The cumulative offset is written into the status information carrier as additional information of the current node's status information. When the reference failure condition is not met, the state information carrier remains unchanged and continues to be passed along the call chain.

[0032] As described above, in the preceding steps, the current microservice node has completed the comparison between its own state information and the upstream reference state information, obtaining state difference information, i.e., the first-order difference result. This first-order difference result is then sent to the causal determination logic for source determination. However, as shown in the aforementioned scenario, if the upstream node itself has undergone severe distortion, the downstream node's own state may shift synchronously after receiving the distorted input. In this case, the state difference between the downstream node and the upstream node is actually very small, and the first-order difference result may be lower than the anomaly determination threshold, causing the causal determination logic to run correctly but draw an incorrect conclusion, and the true root cause node may be missed. To address the above scenario, this embodiment first evaluates the state difference information before sending it to the causal determination logic to determine whether it meets the preset reference failure condition. The reference failure condition means that the upstream reference state information itself has deviated from the normal range, causing the first-order difference calculated based on this reference to fail to accurately reflect the current node's state change. This assessment can be achieved in several ways, such as checking whether the state values ​​of each dimension in the upstream reference state information have exceeded the preset normal operating range, or checking whether the deviation of the upstream reference state information from the initial baseline state information exceeds a certain standard. If the state difference information meets the reference failure condition, it indicates that the current node is in a special scenario where the upstream has been distorted.

[0033] At this point, the method no longer relies on upstream reference state information, but instead obtains initial baseline state information. This initial baseline state information is recorded at the moment the business request is initiated; it represents the state at the start of the call chain and remains unchanged throughout the entire call chain propagation. After obtaining the initial baseline state information, this method compares the current node's local state information with the initial baseline state information to determine the cumulative offset of the current node's state information relative to the initial baseline state information. Unlike the first-order difference, which reflects the change of the current node relative to the upstream node, the cumulative offset reflects the change of the current node relative to the start of the call chain. This cumulative offset is not affected by whether the upstream node's state is distorted, because its reference baseline is always the fixed initial state at the start of the call chain. After determining the cumulative offset, this cumulative offset is written into the state information carrier as additional information of the current node's state information. In this way, in addition to recording state difference information and causal attribution identifiers, the state information carrier also records the cumulative offset. This cumulative offset can be regarded as an absolute offset record accumulated node by node along the entire call chain, providing an alternative data source independent of the causal attribution identifier for subsequent root cause determination. Conversely, if the state difference information does not meet the reference failure condition, it means that the upstream reference state information is still reliable and the first-order difference is still valid. In this case, the method does not need additional processing; it simply maintains the state information carrier unchanged and continues to propagate downstream along the call chain, allowing the original causal determination process to execute normally. Through this mechanism, even when causal misjudgment has begun to propagate along the chain, the accumulated offset can still accurately record the absolute offset of each node relative to the initial state, preserving an independent criterion for subsequent steps to accurately pinpoint the true root cause node, thus preventing the situation described above where all anomalies in the entire chain are missed.

[0034] In one embodiment, the method of comparing the current node state information with the initial reference state information to determine the cumulative offset of the current node state information relative to the initial reference state information includes: Extract the current node state value corresponding to each state dimension from the current node state information, and read the reference state value corresponding to each state dimension from the initial reference state information from the initial reference area; For each state dimension, calculate the absolute value of the difference between the current node's state value and the baseline state value, and use the calculated result as the dimension offset of the state dimension; The preset weight mapping table is invoked to assign corresponding weight coefficients to each state dimension based on the business type of the current microservice node. Call the preset truncation function, input the dimension offset of each state dimension into the truncation function, and obtain the truncated offset of each state dimension. The truncation function is used to set the dimension offset less than the preset truncation threshold to zero, and retain the dimension offset greater than or equal to the preset truncation threshold as the dimension offset. For each state dimension, the truncated offset is multiplied by the corresponding weight coefficient to obtain the weighted contribution value of each state dimension. The weighted contribution values ​​of each state dimension are summed, and the sum is used as the cumulative offset.

[0035] As mentioned above, in the scenario described earlier, in the call chain consisting of the data acquisition service, the specification parsing service, and the data forwarding service, due to the distortion of the upstream reference, the state difference information calculated by the specification parsing service is only 3 percentage points, and that of the data forwarding service is only 2 percentage points. Both values ​​are too small to trigger a valid causal determination. However, if we shift our perspective from changes relative to the upstream to changes relative to the initial state of the call chain, we find a different situation: the CPU utilization of the specification parsing service is 88%, while the initial state of the data acquisition service at the start of the call chain is 40%, with a cumulative offset of 48 percentage points; the CPU utilization of the data forwarding service is 90%, with a cumulative offset of 50 percentage points. These two cumulative offsets clearly exceed the normal range and accurately reflect that the entire call chain has undergone a severe state shift. This comparison reveals a key issue: how to transform the above approach into an operational calculation method so that each node can calculate such a reliable cumulative offset.

[0036] Specifically, after determining that the cumulative offset needs to be calculated, the method first extracts the corresponding state values ​​of the current node for each state dimension from the current node's state information. These state dimensions are consistent with those used in the preceding steps, such as CPU utilization, request processing latency, and communication protocol exception count. Simultaneously, the method reads the baseline state values ​​for each dimension from the initial baseline state information in the state information carrier. The initial baseline state information is set at the moment the business request is initiated, storing the state values ​​of each dimension at the start of the call chain, and remains unchanged throughout the entire call chain propagation. This step is essentially data preparation and alignment, ensuring that the two sets of data upon which subsequent calculations depend are dimensionally one-to-one. After data preparation, the method calculates the absolute value of the difference between the current node's state value and the baseline state value for each state dimension, using the result as the dimension offset for that dimension. The reason for using an absolute value is that the cumulative offset focuses on the degree of deviation itself, not the direction of deviation. Whether the deviation is high or low, as long as it is large enough, it means that the node's state has changed significantly relative to the initial state, and it deserves to be recorded and monitored. Through this step, each state dimension obtains an independent dimension offset. With the dimension offsets for each dimension, a pre-defined weight mapping table is invoked to assign a corresponding weight coefficient to each state dimension based on the business type of the current microservice node.

[0037] The weight mapping table is a pre-configured data table that defines the importance level of each state dimension according to different service types. For example, for relay protection services, the weight coefficient for communication protocol anomaly count might be set relatively high because this dimension directly relates to whether the protection signal can be transmitted correctly; while for ordinary data acquisition services, the weight coefficient for CPU utilization might be relatively higher because this dimension reflects the data acquisition throughput capacity. By introducing weight coefficients, the calculation of cumulative offset no longer treats all dimensions equally, but rather allows key dimensions to occupy a larger proportion in the final cumulative offset based on the actual focus of different services in the power communication network. After weight allocation, a preset truncation function is called, inputting the dimension offsets of each state dimension to obtain the truncated offsets of each state dimension. The truncation function filters out small dimension offsets. Microservices commonly experience periodic, minor state fluctuations during daily operation, such as timed heartbeat reporting and periodic state updates. These operations can cause brief, small increases in CPU utilization or memory usage, but these increases are normal and do not indicate a genuine state anomaly. Without a truncation function to filter these fluctuations, they would continuously accumulate in the cumulative offset, causing it to become artificially high at the end of the call chain and interfering with subsequent root cause determination. The truncation function works by setting dimension offsets below a preset threshold to zero, effectively ignoring these minor fluctuations; and retaining the original values ​​of dimension offsets greater than or equal to the preset threshold. Through this non-linear processing, the cumulative offset does not continuously increase due to minor periodic fluctuations; it is only included in the cumulative offset when the deviation of a dimension truly reaches a noteworthy level. After truncation, for each state dimension, the truncated offset is multiplied by the corresponding weight coefficient to obtain the weighted contribution value of that state dimension. This step combines truncation filtering and weight allocation mechanisms. The final contribution of each dimension is filtered for minor fluctuations and weighted by business semantics. Finally, the weighted contribution values ​​of all state dimensions are summed, and the sum is used as the final cumulative offset. This cumulative offset is a scalar value that integrates multi-dimensional information, undergoes truncation filtering and differentiated weighting, and represents the overall deviation of the current microservice node from the initial state of the call chain.

[0038] In one embodiment, the method further includes collecting the state information carrier at a predetermined stage of the call chain, filtering out self-caused state difference information from the state information carrier based on the causal attribution identifier, and generating aggregated observation output based on the filtering results. During a predetermined phase of the call chain, the state information carrier is collected, and the causal attribution identifier and cumulative offset of each microservice node are read from the state information carrier. Traverse each microservice node to determine if there is a target node. The target node is the node whose causal attribution identifier is the source of the transmission and whose cumulative offset exceeds a preset offset threshold. When the target node exists, for each target node, trace upstream along the call chain to obtain the node with the lowest cumulative offset among the upstream nodes of the target node as the reference node, and compare the cumulative offset of the target node with the cumulative offset of the reference node. Based on the comparison results, when the cumulative offset of the target node exceeds the cumulative offset of the reference node and the excess reaches a preset magnitude threshold, the causal attribution identifier of the target node is corrected to hide the self-caused source. Write the corrected causal attribution identifier into the position corresponding to the target node in the state information carrier; When filtering state difference information with self-caused sources, state difference information with causal attribution identifiers that are self-caused or have hidden self-caused sources is also filtered out. Based on the filtering results, aggregated observation output is generated, and the hidden self-caused sources are presented in the aggregated observation output in an identifier that is different from that of self-caused sources. When the target node does not exist, the state information carrier remains unchanged and aggregated observation output is generated according to the original filtering method.

[0039] As described above, after the preceding steps, the state information carrier has accumulated two different types of information. One is the causal attribution identifier for each node, generated by first-order difference through causal determination logic, indicating whether each node is a self-caused source or a transitive source. The other is the cumulative offset of each node, generated by comparing with the initial baseline, recording the absolute deviation of each node from the initial state of the call chain. These two types of information reflect the abnormal state of each node in the call chain from different perspectives, but they may not always be consistent. When a node is marked as a transitive source by the causal determination logic, but its cumulative offset is abnormally high, a contradictory signal occurs, requiring a dedicated mechanism to handle this situation.

[0040] Specifically, during the pre-defined phase of the call chain, this method collects state information carriers and reads the causal attribution identifiers and cumulative offsets of each microservice node from these carriers. At this point, the system possesses two sets of information for each node in the entire call chain: one set based on the causal attribution judgment obtained from first-order difference, and the other set based on the absolute offset degree obtained from the initial baseline. After reading all the information, the system traverses each microservice node to determine if a target node exists. A target node must simultaneously meet two conditions: its causal attribution identifier indicates a propagation source, but its cumulative offset exceeds a preset offset threshold. The first condition indicates that the node is not considered the root cause by the causal determination logic, and its anomaly is propagated from upstream. The second condition indicates that, from the perspective of absolute offset, the node's state has significantly deviated from its initial state. When both conditions are met, the node is a special node requiring further analysis. If no such node exists, it means there is no significant contradiction between the first-order difference and the cumulative offset, and the original filtering method can be used directly. When the aforementioned target nodes exist, for each target node, the method traces upstream along the call chain to obtain the node with the lowest cumulative offset among the upstream nodes of that target node, and uses this node as the reference node. Then, the cumulative offset of the target node is compared with the cumulative offset of the reference node. The reason for choosing the node with the lowest cumulative offset among the upstream nodes as the reference is that this reference node represents the state baseline with the smallest state offset in the upstream call chain.

[0041] If the cumulative offset of the target node significantly exceeds this benchmark, it indicates that the target node itself has indeed introduced additional offset, rather than simply following the upstream offset. Specifically, when the cumulative offset of the target node exceeds the cumulative offset of the reference node, and the exceedance reaches a preset threshold, the causal attribution identifier of the target node is corrected to a hidden self-caused source. The logic behind this correction is that if the target node is merely dragged down by upstream distortion, its cumulative offset should not significantly exceed that of the upstream node with the lowest offset. However, if its cumulative offset is significantly higher than the lowest upstream level, it indicates that it itself is the true source of offset, but is misjudged as a propagation source due to the failure of first-order difference. This correction action actually uses the independent information dimension of cumulative offset to reverse the possible causal misjudgment. After the correction is completed, the corrected causal attribution identifier is written into the state information carrier at the corresponding position of the target node, replacing the original propagation source identifier. In the final filtering stage, the method simultaneously filters out state difference information where the causal attribution identifier is a self-caused source or a hidden self-caused source, and generates aggregated observation output based on the filtering results. These two types of identifiers are presented differently in the output. For example, nodes that are originally self-caused are marked in the usual way, while nodes that have been corrected to hide their self-caused origins are marked differently to remind operations personnel that the node is the root cause node that has been corrected after verification by the accumulated offset, and its original causal attribution identifier may have been over-biased due to the failure of the first-order difference. If no target node is found after traversing all nodes, it means that there is no obvious contradiction signal between the first-order difference and the accumulated offset on the entire call chain. At this time, the state information carrier remains unchanged, and the aggregated observation output can be generated according to the original filtering method.

[0042] Reference Figure 4 This application also provides a power communication network microservice state aggregation and observation system based on distributed link tracing, comprising: Response module 1 is used to respond to the initiation of a business request and initialize a state information carrier that propagates along the call chain. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. Call module 2 is used to obtain upstream reference state information from the state information carrier during the call chain propagation process; The acquisition module 3 is used to acquire the current node status information generated by the current microservice node after acquiring the upstream reference status information, and compare the current node status information with the upstream reference status information to determine the status difference information based on the comparison result. The judgment module 4 is used to call the preset causal judgment logic to judge the source of the state difference information, and to mark the corresponding causal attribution identifier for the state difference information according to the judgment result. Input module 5 is used to write the annotated state difference information into the state information carrier based on the annotation results, and to pass it along the call chain or trigger information collection. Output module 6 is used to collect the state information carrier at a predetermined stage of the call chain, and filter out the state difference information with self-caused origin from the state information carrier according to the causal attribution identifier, and generate aggregated observation output based on the filtering results.

[0043] As described above, it can be understood that each component of the power communication network microservice state aggregation observation system based on distributed link tracing proposed in this application can realize the function of any of the power communication network microservice state aggregation observation methods based on distributed link tracing as described above, and the specific structure will not be repeated.

[0044] Reference Figure 5 This application also provides a computer device, which may be a server, and its internal structure may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores monitoring data and other data. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a microservice state aggregation observation method for power communication networks based on distributed link tracing.

[0045] The processor described above executes the power communication network microservice state aggregation and observation method based on distributed link tracing, comprising: responding to the initiation of a service request, initializing a state information carrier that propagates along the call chain, wherein the state information carrier carries the state information of each microservice node on the call chain and the corresponding causal attribution identifier; during the call chain propagation, obtaining upstream reference state information from the state information carrier; after obtaining the upstream reference state information, obtaining the current node state information generated by the current microservice node's processing, comparing the current node state information with the upstream reference state information, and determining state difference information based on the comparison result; invoking a preset causal determination logic to determine the source of the state difference information, and labeling the state difference information with the corresponding causal attribution identifier based on the determination result; based on the labeling result, writing the labeled state difference information into the state information carrier, and propagating it along the call chain or triggering information collection; at a predetermined stage of the call chain, collecting the state information carrier, and filtering out the self-originating state difference information from the state information carrier according to the causal attribution identifier, and generating aggregated observation output based on the filtering result.

[0046] One embodiment of this application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements a method for aggregating and observing the state of microservices in a power communication network based on distributed link tracing. The method includes the following steps: in response to a service request, initializing a state information carrier that propagates along the call chain. The state information carrier carries the state information of each microservice node in the call chain and the corresponding causal attribution identifier; during the call chain propagation, obtaining upstream reference state information from the state information carrier; and after obtaining the upstream reference state information, obtaining the current node state generated by the processing performed by the current microservice node. The system collects and compares the current node's state information with the upstream reference state information, and determines the state difference information based on the comparison result. It then invokes a preset causal determination logic to determine the source of the state difference information and labels the state difference information with a corresponding causal attribution identifier based on the determination result. Based on the labeling result, the labeled state difference information is written into the state information carrier and transmitted along the call chain or triggered information collection. At a predetermined stage of the call chain, the system collects the state information carrier and filters out the self-originating state difference information from the state information carrier based on the causal attribution identifier, generating an aggregated observation output based on the filtering result.

[0047] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0048] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for microservice state aggregation and observation in power communication networks based on distributed link tracing, characterized in that, The method includes: In response to the initiation of a business request, a state information carrier that propagates along the call chain is initialized. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. During the call chain propagation process, upstream reference state information is obtained from the state information carrier; After obtaining the upstream reference status information, the current node status information generated by the current microservice node's execution of processing is obtained, and the current node status information is compared with the upstream reference status information. Based on the comparison result, the status difference information is determined. The preset causal determination logic is invoked to determine the source of the state difference information, and the corresponding causal attribution identifier is marked on the state difference information according to the determination result. Based on the annotation results, the annotated state difference information is written into the state information carrier and transmitted along the call chain or triggered to collect information. During a predetermined phase of the call chain, the state information carrier is collected, and state difference information originating from the state information carrier is filtered out based on the causal attribution identifier. An aggregated observation output is then generated based on the filtering results, specifically including: During the predetermined stage of the call chain, all records written in the state information carrier are read. Each record includes state difference information of a microservice node and the causal attribution identifier corresponding to the state difference information. Iterate through all the records, and for each record, check whether the causal attribution identifier of each record is a self-caused source; Based on the results of the inspection, records with causal attribution identified as self-caused are retained as valid records, while records with causal attribution identified as transitive are marked as records to be filtered. Extract the state difference information of each microservice node from the valid records, and arrange the extracted state difference information in the order of nodes according to the topological order of the call chain to form a root cause candidate list. Based on the root cause candidate list, an aggregated observation view of the power communication network is generated. The aggregated observation view includes at least one of the following: a service topology graph labeled with self-caused source nodes, a causal path highlighting along the call chain, and abnormal state dimension details for each self-caused source node.

2. The method for microservice state aggregation and observation of power communication networks based on distributed link tracing according to claim 1, characterized in that, The method for initializing a state information carrier that propagates along the call chain in response to the initiation of a business request includes: In response to the initiation of the service request, a unique tracking identifier corresponding to the service request is generated; Based on the unique tracking identifier, create an empty status information carrier object; In the state information carrier object, a first storage area and a second storage area are divided, wherein the first storage area is used to record the state information of each microservice node in the call chain, and the second storage area is used to record the causal attribution identifier corresponding to each state information. Write the state information of the starting microservice node of the business request as the initial state information into the first storage area, and preset the causal attribution identifier corresponding to the initial state information as the transmission source; The state information carrier object is bound to the call chain context of the business request, so that the state information carrier can be propagated between microservice nodes along with the call chain context of the business request.

3. The method for microservice state aggregation and observation of power communication networks based on distributed link tracing according to claim 1, characterized in that, The method of comparing the current node's state information with the upstream reference state information and determining the state difference information based on the comparison result includes: The node status information is parsed, and based on the parsing results, a first set of status values ​​including status dimensions is extracted. The status dimensions include at least one of resource utilization, request processing latency, business processing result code, and communication protocol exception count. Parse the upstream reference state information and extract a second state value set with the same state dimension as the first state value set; For each state dimension, calculate the difference between the state value of the dimension in the first state value set and the state value of the corresponding dimension in the second state value set; Based on the calculation results, the differences in all dimensions are combined into a state change vector, which serves as the state difference information.

4. The method for microservice state aggregation and observation of power communication networks based on distributed link tracing according to claim 1, characterized in that, The method of comparing the current node's state information with the upstream reference state information and determining the state difference information based on the comparison result further includes: Determine whether the state difference information meets the preset reference failure condition; When the reference failure condition is met, the initial reference state information is obtained, which is the state information at the time the service request is initiated. The current node status information is compared with the initial reference status information to determine the cumulative offset of the current node status information relative to the initial reference status information; The cumulative offset is written into the status information carrier as additional information of the current node's status information. When the reference failure condition is not met, the state information carrier remains unchanged and continues to be passed along the call chain.

5. The method for microservice state aggregation and observation of power communication networks based on distributed link tracing according to claim 4, characterized in that, The method of comparing the current node's state information with the initial baseline state information to determine the cumulative offset of the current node's state information relative to the initial baseline state information includes: Extract the current node state value corresponding to each state dimension from the current node state information, and read the reference state value corresponding to each state dimension from the initial reference state information from the initial reference area; For each state dimension, calculate the absolute value of the difference between the current node's state value and the baseline state value, and use the calculated result as the dimension offset of the state dimension; The preset weight mapping table is invoked to assign corresponding weight coefficients to each state dimension based on the business type of the current microservice node. Call the preset truncation function, input the dimension offset of each state dimension into the truncation function, and obtain the truncated offset of each state dimension. The truncation function is used to set the dimension offset less than the preset truncation threshold to zero, and retain the dimension offset greater than or equal to the preset truncation threshold as the dimension offset. For each state dimension, the truncated offset is multiplied by the corresponding weight coefficient to obtain the weighted contribution value of each state dimension. The weighted contribution values ​​of each state dimension are summed, and the sum is used as the cumulative offset.

6. The method for microservice state aggregation and observation of power communication networks based on distributed link tracing according to claim 5, characterized in that, The method further includes, in a predetermined stage of the call chain, collecting the state information carrier, filtering out self-originating state difference information from the state information carrier based on the causal attribution identifier, and generating aggregated observation output based on the filtering results. During a predetermined phase of the call chain, the state information carrier is collected, and the causal attribution identifier and cumulative offset of each microservice node are read from the state information carrier. Traverse each microservice node to determine if there is a target node. The target node is the node whose causal attribution identifier is the source of the transmission and whose cumulative offset exceeds a preset offset threshold. When the target node exists, for each target node, trace upstream along the call chain to obtain the node with the lowest cumulative offset among the upstream nodes of the target node as the reference node, and compare the cumulative offset of the target node with the cumulative offset of the reference node. Based on the comparison results, when the cumulative offset of the target node exceeds the cumulative offset of the reference node and the excess reaches a preset magnitude threshold, the causal attribution identifier of the target node is corrected to hide the self-caused source. Write the corrected causal attribution identifier into the position corresponding to the target node in the state information carrier; When filtering state difference information with self-caused sources, state difference information with causal attribution identifiers that are self-caused or have hidden self-caused sources is also filtered out. Based on the filtering results, aggregated observation output is generated, and the hidden self-caused sources are presented in the aggregated observation output in an identifier that is different from that of self-caused sources. When the target node does not exist, the state information carrier remains unchanged and aggregated observation output is generated according to the original filtering method.

7. A power communication network microservice state aggregation and observation system based on distributed link tracing, used to execute the steps of the power communication network microservice state aggregation and observation method based on distributed link tracing as described in any one of claims 1-6, characterized in that, include: The response module is used to respond to the initiation of a business request and initialize a state information carrier that propagates along the call chain. The state information carrier is used to carry the state information of each microservice node in the call chain and the causal attribution identifier corresponding to the state information. The calling module is used to obtain upstream reference state information from the state information carrier during the call chain propagation process; The acquisition module is used to acquire the current node status information generated by the current microservice node after acquiring the upstream reference status information, and compare the current node status information with the upstream reference status information to determine the status difference information based on the comparison result. The judgment module is used to call the preset causal judgment logic to judge the source of the state difference information, and to mark the corresponding causal attribution identifier for the state difference information according to the judgment result. The input module is used to write the annotated state difference information into the state information carrier based on the annotation results, and to pass it along the call chain or trigger information collection. The output module is used to collect the state information carrier at a predetermined stage of the call chain, and filter out the state difference information with self-caused origin from the state information carrier according to the causal attribution identifier, and generate aggregated observation output based on the filtering results.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images