Server exception handling methods and server

Abstract

This invention relates to the field of servers, disclosing a server anomaly handling method and a server, including a main body. One end of the main body has horizontally spaced hard drive slots, and a flexible hard drive mounting component is installed on the inner side of the main body corresponding to the hard drive slots. A heat dissipation mechanism is located on the inner side of the main body next to the flexible hard drive mounting component, and a motherboard is located on the bottom inner side of the main body next to the heat dissipation mechanism. In this invention, the hard drive is fixed using a flexible mounting component, and elastic cushioning is achieved through a pressing spring, effectively mitigating vibration damage to the hard drive. Simultaneously, the L-shaped mesh pressure plate design ensures the stability of the hard drive fixation while providing ventilation channels to improve the hard drive's heat dissipation effect. The arc-shaped strip facilitates quick insertion and removal of the hard drive, reducing maintenance difficulty. The CPU component is pressed and fixed by an arc-shaped pressure arm, ensuring close contact between the heat dissipation plate, the thermal grease coating, and the CPU chip, improving heat conduction efficiency.

Description

Technical Field

[0001] This invention relates to the field of servers, and more particularly to server exception handling methods and servers. Background Technology

[0002] With the rapid development of cloud computing and big data technologies, servers, as a high-performance and highly stable dedicated computer device, possess powerful computing, storage, and data processing capabilities. They can operate continuously and stably 24 / 7 and are mainly used in network environments to provide core services such as data storage, resource scheduling, business response, computing power support, network services, and data management for client devices. They are widely used in scenarios such as internet platforms, enterprise offices, cloud computing, data centers, database operation and maintenance, and website construction, and are the core infrastructure in digital network systems that carries data interaction and business operations.

[0003] Current servers exhibit the following technical deficiencies in practical applications: The server hardware installation structure is unreasonable; hard drives are often rigidly fixed without proper cushioning, making them susceptible to loosening of interfaces and hardware damage from vibrations during transportation and operation. Furthermore, the enclosed hard drive installation structure results in poor ventilation and heat dissipation, making them prone to overheating and operational abnormalities. The CPU, as the core computing component of the server, suffers from an imperfect heat dissipation design, low thermal conductivity, and insufficiently secure fixing, leading to issues such as loose thermal paste application and detachment of heatsink fins, affecting CPU stability. Existing server anomaly handling technologies are inadequate, with anomaly monitoring dimensions being limited, often only collecting in-band data at the software level. The use of out-of-band data at the data or hardware level, without comprehensive analysis in conjunction with business-related data, leads to a high rate of missed and false positives in anomaly detection. Anomaly prediction capabilities are weak, making it difficult to identify latent and complex anomalies. Actions are often taken only after anomalies have occurred, failing to provide early warnings or mitigate risks. Anomaly root cause localization is inefficient, struggling to quickly distinguish between soft faults, hard faults, and business-related anomalies, resulting in lengthy localization times and impacting remediation efficiency. Remediation strategies lack specificity, failing to classify and handle anomalies according to their severity and scope, and lacking effective post-remediation verification and closed-loop optimization mechanisms, leading to recurring anomalies. Furthermore, the entire anomaly handling process lacks a traceable audit system, hindering subsequent strategy optimization and problem investigation. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a server anomaly handling method and server to solve the problems of low stability of existing server hardware structure, poor heat dissipation, low anomaly handling efficiency and weak prediction ability.

[0005] To address the aforementioned problems, the present invention is implemented through the following technical solution.

[0006] The server anomaly handling method and server include: a main body, one end of which has horizontally spaced hard drive slots, and a hard drive flexible mounting component installed on the inner side of the main body corresponding to the hard drive slots; a heat dissipation mechanism is provided on the inner side of the main body next to the hard drive flexible mounting component; a motherboard is provided on the bottom inner side of the main body next to the heat dissipation mechanism; multiple fastening screws are provided on the motherboard where no electrical components are located, and the motherboard is connected to the main body through the fastening screws; a CPU component is installed on one side of the upper end of the motherboard, and a memory module is provided on the upper end of the motherboard next to the CPU component; a power supply is provided on the inner side of the main body next to the motherboard; and multiple expansion slots one and two are provided on the side of the main body away from the hard drive slots.

[0007] In one embodiment, heat dissipation grooves are symmetrically provided on both sides of the main body, the upper end of the main body is open, and mounting connecting pieces are provided at equal intervals on both sides of the upper end of the main body. A cover plate is placed on the upper end of the mounting connecting piece. The shape and structure of the cover plate matches the shape and structure of the upper end opening of the main body, and a fastening screw six is ​​provided through the cover plate corresponding to the mounting connecting piece. The fastening screw six extends into the mounting connecting piece.

[0008] In one embodiment, the hard drive elastic mounting assembly includes a hard drive, arc-shaped strips, L-shaped mesh pressure plates, pressing springs, a mounting plate, and a set of fastening screws. Two arc-shaped strips are symmetrically arranged longitudinally at the front end of the hard drive. Two sets of L-shaped mesh pressure plates, pressing springs, mounting plates, and fastening screws are symmetrically arranged on the upper and lower sides of the hard drive for pressing and fixing the hard drive, and ventilation channels are reserved at the upper and lower ends of the hard drive. One end of the L-shaped mesh pressure plate is arc-shaped and raised. At least six pressing springs are arranged in a rectangular array between the L-shaped mesh pressure plate and the mounting plate. The mounting plate is connected to the main body by the fastening screws.

[0009] In one embodiment, ventilation slot one is provided at both the upper and lower ends of the hard disk slot of the main body, and ventilation slot two is provided at the end of the main body away from the hard disk slot. The ventilation slot one, the L-shaped mesh pressure plate and the ventilation slot two provide a connected ventilation channel for the main body.

[0010] In one embodiment, the heat dissipation mechanism includes a base plate, a fixing frame, a second fastening screw, a locking groove, a mounting frame, an opening groove, an inner arc-shaped mounting plate, a heat sink housing, a third fastening screw, heat dissipation fins, a motor, a rotating shaft, a connector, fan blades, and a dustproof mesh plate. The base plate is fixedly connected to the bottom inner side of the main body by the second fastening screw.

[0011] In one embodiment, the upper end of the base plate is provided with a fixing frame, and the inner side of the fixing frame is provided with not less than four locking slots at equal intervals. An installation frame is locked in the locking slot from top to bottom. The installation frame is provided with an opening slot through the ventilation direction of the main body. The inner wall of the installation frame is provided with an inner arc-shaped installation plate. The inner arc-shaped installation plate is connected to the radiator housing by fastening screws.

[0012] In one embodiment, a heat dissipation fin is provided on one end surface of the radiator housing, and a motor is provided at the middle of the heat dissipation fin at one end of the radiator housing. The output end of the motor is connected to a rotating shaft for forward transmission, and the front end of the rotating shaft is connected to a fan blade through a connector. A dustproof mesh plate is provided on the inner side of the mounting frame corresponding to the front end of the fan blade.

[0013] In one embodiment, the CPU assembly includes a CPU socket, a CPU chip, a thermal grease coating, a heat spreader, an arc-shaped pressure arm, five fastening screws, and thermal fins. The CPU socket is located at the top of the motherboard, and a CPU chip body is inserted into the CPU socket. The upper surface of the CPU chip body is coated with a thermal grease coating, and a heat spreader is placed on top of the thermal grease coating. Thermal fins are disposed on top of the heat spreader.

[0014] In one embodiment, the heat dissipation plate has arc-shaped pressure arms on both sides. The arc-shaped pressure arms are flat near the heat dissipation plate and arc-shaped away from the heat dissipation plate. The arc-shaped front end of the arc-shaped pressure arm presses downward against the upper surface of the motherboard. A fastening screw five is provided through the flat part of the arc-shaped pressure arm and extends into the motherboard, thereby realizing the heat dissipation plate pressing and fixing the thermal grease coating on the CPU chip.

[0015] Server exception handling methods include:

[0016] S1. Multi-source data acquisition and standardized processing: Construct a multi-dimensional anomaly monitoring system, simultaneously collecting in-band data, out-of-band data, and business-related data from the server. The acquisition frequency is set to 5 seconds / time to form a standardized monitoring dataset. The in-band data includes software-level data such as operating system kernel logs, CPU utilization, memory usage, and disk I / O, covering all core processes of the server with 100% coverage. The out-of-band data includes physical-level data such as server hardware health status and hardware fault alarms, collected in real time through the BMC chip with a data transmission latency ≤100ms. The business-related data includes business operation data such as request QPS, response latency, and error rate, covering all core business modules with a data acquisition completeness ≥99.8%.

[0017] S2. Anomaly Prediction and Risk Classification: An anomaly prediction model is constructed based on an improved isolated forest algorithm. The standardized monitoring dataset is preprocessed and features are extracted, resulting in 18 key features (adding three new features compared to existing models: hardware temperature gradient, process occupancy fluctuation, and peak deviation of business requests). Model parameters are updated in real-time using a sliding window (window size set to 3 minutes) to achieve early prediction of latent and complex anomalies, outputting the anomaly prediction results and risk levels (1-10). The improved isolated forest algorithm introduces a business load weight coefficient (dynamically adjusted according to business QPS, weight range 0.1-0.9) to address the high false positive rate of traditional algorithms in scenarios with fluctuating business operations. The model's prediction accuracy is ≥99.2%, recall is ≥95.5%, and latent anomaly identification rate is ≥88%, representing an improvement of over 28 percentage points compared to existing technologies.

[0018] S3. Anomaly Root Cause Locator and Report Generation: When the anomaly prediction model outputs a high-risk warning (risk level ≥ 8) or detects an actual anomaly, a multi-source data cross-validation mechanism is activated. The validation time is ≤ 30 seconds. Combining in-band, out-of-band, and business-related data, the graph neural network is used to mine anomaly correlations, locate the anomaly root cause, distinguish between soft faults, hard faults, and business-related anomalies, with a location accuracy of ≥ 98.3% and a location time of ≤ 12 minutes. A location report containing the anomaly location, anomaly type, scope of impact, and number of affected users is generated.

[0019] S4. Tiered Automated Repair and Business Assurance: Based on anomaly location reports, a pre-defined atomic repair module library is invoked. The module library contains 12 core repair modules (two new modules added compared to the existing ones: hidden fault repair and cascading anomaly blocking). Repair strategies are dynamically combined according to the anomaly type and impact scope to execute automated repair operations. For minor anomalies that can be repaired online, the repair time is ≤60 seconds and the repair success rate is ≥99%. For severe anomalies that require offline repair, the business migration process is automatically triggered, with a migration time of ≤25 seconds and a business interruption duration of ≤30 seconds, which is more than 75% shorter than existing technologies and a migration success rate of ≥99.5%, before the repair operation is executed.

[0020] S5. Repair Effect Verification and Closed-Loop Feedback: After the repair is completed, the repair effect is verified. The verification time is set to 5-10 minutes. If the verification is successful, the exception handling log and model parameters are updated to form a closed-loop feedback. If the verification fails, the repair strategy is adjusted and the repair operation is re-executed until the repair is successful. The overall repair success rate is ≥98.8%, which is more than 33 percentage points higher than the existing technology.

[0021] S6. End-to-End Audit and Strategy Optimization: Establish an end-to-end audit system for anomaly handling, record data throughout the entire process of anomaly prediction, location, repair, and verification, retain data for at least 1 year, generate traceable audit reports, and achieve 100% audit coverage; simultaneously, continuously optimize the anomaly prediction model and repair strategy based on historical data, update model parameters every 7 days, and optimize repair strategies every 15 days. After optimization, the model's prediction accuracy improves by 0.3%-0.5% per month, and the repair efficiency improves by 5%-8% per month.

[0022] This invention provides a server exception handling method and a server. Compared with the prior art, it has the following advantages:

[0023] 1. The hard drive is secured using a flexible mounting component, with a spring providing cushioning to effectively reduce vibration damage. The L-shaped mesh pressure plate design ensures stability while providing ventilation channels for improved heat dissipation. The curved strip facilitates quick hard drive insertion and removal, reducing maintenance difficulty. The CPU assembly is secured with a curved pressure arm, ensuring close contact between the heat spreader, thermal grease coating, and CPU chip, improving heat dissipation efficiency and ensuring stable CPU operation. The main body and cover are detachably connected via mounting tabs and screws, and the cooling mechanism engages with the mounting bracket via a snap-fit ​​groove, facilitating future maintenance and repair.

[0024] 2. A continuous ventilation channel is constructed through ventilation slot one, L-shaped mesh pressure plate and ventilation slot two. Together with the fan blades and heat dissipation fins of the heat dissipation mechanism, it accelerates the airflow inside the server and quickly dissipates heat. The dustproof mesh plate can effectively block dust from entering the heat dissipation mechanism and avoid the heat dissipation components from malfunctioning due to dust accumulation. The heat dissipation slots on both sides of the main body further improve the overall heat dissipation effect and effectively solve the problem of abnormal operation caused by overheating of server hardware.

[0025] 3. Construct a multi-dimensional anomaly monitoring system, simultaneously collecting in-band, out-of-band, and business-related data to ensure the comprehensiveness and completeness of data collection; build an anomaly prediction model based on an improved isolated forest algorithm, adding three key features and introducing a business load weight coefficient to significantly improve the accuracy of anomaly prediction, especially the ability to identify latent and complex anomalies, enabling early warning of anomalies; mine anomaly correlations through graph neural networks to quickly locate the root cause of anomalies, distinguish fault types, and improve the efficiency and accuracy of location; implement graded automated repair according to anomaly type and impact scope, repairing minor anomalies online and triggering business migration before repairing severe anomalies, significantly shortening business interruption time and improving repair success rate; set up effect verification and closed-loop feedback mechanisms after repair to ensure repair effectiveness and prevent anomaly recurrence; at the same time, build a full-process audit system to achieve full traceability of anomaly handling, and continuously optimize the prediction model and repair strategy based on historical data to continuously improve anomaly handling capabilities. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the overall three-dimensional structure of the server.

[0027] Figure 2 This is a schematic diagram of the server's internal structure.

[0028] Figure 3 This is a schematic diagram of the server body after separation from the heat dissipation mechanism, motherboard, and hard drive flexible mounting components.

[0029] Figure 4 This is a schematic diagram of the server body and the flexible installation component of the hard drive.

[0030] Figure 5 Exploded view of the flexible mounting component for server hard drives.

[0031] Figure 6 This is a schematic diagram of the exploded structure of a server's heat dissipation mechanism.

[0032] Figure 7 This is a schematic diagram of the exploded structure of the server motherboard and CPU components.

[0033] Figure 8 This is a flowchart illustrating the server exception handling process.

[0034] The attached figures are labeled as follows:

[0035] 1. Main body; 2. Hard drive bay; 3. Hard drive flexible mounting assembly; 301. Hard drive; 302. Arc-shaped strip; 303. L-shaped mesh pressure plate; 304. Pressing spring; 305. Mounting plate; 306. Fastening screw one; 4. Heat dissipation groove; 5. Heat dissipation mechanism; 501. Base plate; 502. Fixing bracket; 503. Fastening screw two; 504. Snap-fit ​​groove; 505. Mounting frame; 506. Opening groove; 507. Inner arc-shaped mounting plate; 508. Heat sink housing; 509. Fastening screw three; 510. Heat dissipation fins; 511. Motor; 512. 513. Rotating shaft; 514. Connector; 515. Fan blade; 516. Dustproof mesh plate; 6. Motherboard; 7. Fastening screw four; 8. CPU assembly; 801. CPU socket; 802. CPU chip; 803. Thermal grease coating; 804. Heat dissipation plate; 805. Arc-shaped pressure arm; 806. Fastening screw five; 807. Thermal fins; 9. Memory module; 10. Power supply; 11. Expansion slot one; 12. Expansion slot two; 13. Mounting connector; 14. Cover plate; 15. Fastening screw six; 16. Ventilation slot one; 17. Ventilation slot two. Detailed Implementation

[0036] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0037] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0038] Reference Figures 1-8 The server includes a server and a server with a fault handling method. The server comprises: a main body 1, a hard drive slot 2 with equal horizontal spacing at one end of the main body 1, a hard drive flexible mounting component 3 installed on the inner side of the main body 1 corresponding to the hard drive slot 2, a heat dissipation mechanism 5 on the inner side of the main body 1 next to the hard drive flexible mounting component 3, a motherboard 6 on the inner bottom of the main body 1 next to the heat dissipation mechanism 5, multiple fastening screws 7 on the motherboard 6 where there are no electrical components, and the motherboard 6 is connected to the main body 1 through the fastening screws 7, a CPU component 8 installed on the upper side of the motherboard 6, a memory module 9 on the upper side of the motherboard 6 next to the CPU component 8, a power supply 10 on the inner side of the main body 1 next to the motherboard 6, and multiple expansion slots 11 and 22 on the side of the main body 1 away from the hard drive slot 2.

[0039] The main body 1 has symmetrical heat dissipation grooves 4 on both sides. The upper end of the main body 1 is open. The upper end of the main body 1 has mounting connecting pieces 13 equidistantly arranged on both sides. The upper end of the mounting connecting pieces 13 is covered with a cover plate 14. The shape and structure of the cover plate 14 match the shape and structure of the upper end of the main body 1. The cover plate 14 is provided with a fastening screw six 15 through the mounting connecting piece 13. The fastening screw six 15 extends into the mounting connecting piece 13.

[0040] The hard drive flexible mounting assembly 3 includes a hard drive 301, an arc-shaped strip 302, an L-shaped mesh pressure plate 303, a pressing spring 304, a mounting plate 305, and a fastening screw 306. The front end of the hard drive 301 is symmetrically provided with two arc-shaped strips 302. The upper and lower sides of the hard drive 301 are symmetrically provided with two sets of L-shaped mesh pressure plates 303, pressing springs 304, mounting plates 305, and fastening screws 306 for pressing and fixing the hard drive 301, and for providing ventilation channels at the upper and lower ends of the hard drive 301. One end of the L-shaped mesh pressure plate 303 is curved upwards. There are no less than six pressing springs 304 arranged in a rectangular array between the L-shaped mesh pressure plate 303 and the mounting plate 305. The mounting plate 305 is connected to the main body 1 by the fastening screw 306.

[0041] Ventilation slots 16 are provided at both the top and bottom of the hard disk slot 2 on the main body 1, and ventilation slot 2 17 is provided at the end of the main body 1 away from the hard disk slot 2. Ventilation slots 16, L-shaped mesh pressure plate 303 and ventilation slot 2 17 provide ventilation channels for the main body 1.

[0042] The heat dissipation mechanism 5 includes a base plate 501, a fixing bracket 502, a second fastening screw 503, a snap-fit ​​groove 504, a mounting frame 505, an opening groove 506, an inner arc-shaped mounting plate 507, a heat sink housing 508, a third fastening screw 509, heat dissipation fins 510, a motor 511, a rotating shaft 512, a connector 513, fan blades 514, and a dustproof mesh plate 515. The base plate 501 is fixedly connected to the bottom inner side of the main body 1 by the second fastening screw 503.

[0043] The upper end of the base plate 501 is provided with a fixing bracket 502. The inner side of the fixing bracket 502 is provided with no less than four locking slots 504 at equal intervals. The mounting frame 505 is locked from top to bottom in the locking slots 504. The mounting frame 505 is provided with an opening slot 506 through the ventilation direction of the main body 1. The inner wall of the mounting frame 505 is provided with an inner arc-shaped mounting plate 507. The inner arc-shaped mounting plate 507 is connected to the radiator housing 508 by fastening screws 509.

[0044] The heat sink housing 508 has heat dissipation fins 510 on one end surface. A motor 511 is provided at the middle of the heat dissipation fins 510 at one end of the heat sink housing 508. The output end of the motor 511 is connected to a rotating shaft 512 for forward drive. The front end of the rotating shaft 512 is connected to a fan blade 514 through a connector 513. A dustproof mesh plate 515 is provided on the inner side of the mounting frame 505 corresponding to the front end of the fan blade 514.

[0045] CPU component 8 includes CPU socket 801, CPU chip 802, thermal grease coating 803, heat spreader 804, arc-shaped pressure arm 805, fastening screw 806, and heat-conducting fins 807. CPU socket 801 is located at the top of motherboard 6. CPU chip body 802 is inserted into CPU socket 801. The upper surface of CPU chip body 802 is coated with thermal grease coating 803. Heat spreader 804 is placed on top of thermal grease coating 803. Heat-conducting fins 807 are placed on top of heat spreader 804.

[0046] The heat dissipation plate 804 has arc-shaped pressure arms 805 on both sides. The arc-shaped pressure arms 805 are flat near the heat dissipation plate 804 and arc-shaped away from the heat dissipation plate 804. The front arc-shaped part of the arc-shaped pressure arm 805 presses downward to the upper surface of the motherboard 6. The flat part of the arc-shaped pressure arm 805 is provided with a fastening screw 806, which extends into the motherboard 6, so as to realize the heat dissipation plate 804 to press and fix the thermal grease coating 803 and the CPU chip 802.

[0047] Server exception handling methods include:

[0048] S1. Multi-source data acquisition and standardized processing: Construct a multi-dimensional anomaly monitoring system, simultaneously collecting in-band data, out-of-band data, and business-related data from the server. The acquisition frequency is set to 5 seconds / time, forming a standardized monitoring dataset. In-band data includes software-level data such as operating system kernel logs, CPU utilization, memory usage, and disk I / O, covering all core processes of the server with 100% coverage. Out-of-band data includes physical-level data such as server hardware health status and hardware fault alarms, collected in real time through the BMC chip with a data transmission latency of ≤100ms. Business-related data includes business operation data such as request QPS, response latency, and error rate, covering all core business modules with data collection completeness ≥99.8%.

[0049] S2. Anomaly Prediction and Risk Classification: An anomaly prediction model is constructed based on an improved isolated forest algorithm. The standardized monitoring dataset is preprocessed and features are extracted, resulting in 18 key features (adding three new features compared to existing models: hardware temperature gradient, process occupancy fluctuation, and peak deviation of business requests). Model parameters are updated in real-time using a sliding window (window size set to 3 minutes) to achieve early prediction of latent and complex anomalies, outputting the anomaly prediction results and risk levels (1-10). The improved isolated forest algorithm introduces a business load weight coefficient (dynamically adjusted according to business QPS, weight range 0.1-0.9) to address the high false positive rate of traditional algorithms in scenarios with fluctuating business operations. The model's prediction accuracy is ≥99.2%, recall is ≥95.5%, and latent anomaly identification rate is ≥88%, representing an improvement of over 28 percentage points compared to existing technologies.

[0050] S3. Anomaly Root Cause Locator and Report Generation: When the anomaly prediction model outputs a high-risk warning (risk level ≥ 8) or detects an actual anomaly, a multi-source data cross-validation mechanism is activated. The validation time is ≤ 30 seconds. Combining in-band, out-of-band, and business-related data, the graph neural network is used to mine anomaly correlations, locate the anomaly root cause, distinguish between soft faults, hard faults, and business-related anomalies, with a location accuracy of ≥ 98.3% and a location time of ≤ 12 minutes. A location report containing the anomaly location, anomaly type, scope of impact, and number of affected users is generated.

[0051] S4. Tiered Automated Repair and Business Assurance: Based on anomaly location reports, a pre-defined atomic repair module library is invoked. The module library contains 12 core repair modules (two new modules added compared to the existing ones: hidden fault repair and cascading anomaly blocking). Repair strategies are dynamically combined according to the anomaly type and impact scope to execute automated repair operations. For minor anomalies that can be repaired online, the repair time is ≤60 seconds and the repair success rate is ≥99%. For severe anomalies that require offline repair, the business migration process is automatically triggered, with a migration time of ≤25 seconds and a business interruption duration of ≤30 seconds, which is more than 75% shorter than existing technologies and a migration success rate of ≥99.5%, before the repair operation is executed.

[0052] S5. Repair Effect Verification and Closed-Loop Feedback: After the repair is completed, the repair effect is verified. The verification time is set to 5-10 minutes. If the verification is successful, the exception handling log and model parameters are updated to form a closed-loop feedback. If the verification fails, the repair strategy is adjusted and the repair operation is re-executed until the repair is successful. The overall repair success rate is ≥98.8%, which is more than 33 percentage points higher than the existing technology.

[0053] S6. End-to-End Audit and Strategy Optimization: Establish an end-to-end audit system for anomaly handling, record data throughout the entire process of anomaly prediction, location, repair, and verification, retain data for at least 1 year, generate traceable audit reports, and achieve 100% audit coverage; simultaneously, continuously optimize the anomaly prediction model and repair strategy based on historical data, update model parameters every 7 days, and optimize repair strategies every 15 days. After optimization, the model's prediction accuracy improves by 0.3%-0.5% per month, and the repair efficiency improves by 5%-8% per month.

[0054] During use, after the power supply 10 is connected to the external circuit, it provides a stable and continuous power supply to all components of the server, ensuring that each component starts up and runs normally. Among them, the CPU component 8 is the core computing component of the server. The CPU chip 802 is inserted into the CPU socket 801 and electrically connected. After power-on, the CPU chip 802 begins to execute various computing tasks, including data processing and instruction parsing. The motherboard 6 serves as the connection carrier for each component, realizing signal transmission and power distribution for components such as the CPU component 8, memory module 9, power supply 10, and expansion slots. The memory module 9 is directly electrically connected to the CPU component 8 and is used to temporarily store intermediate data and instructions generated during CPU operations, reducing data read latency between the CPU and hard disk 301 and significantly improving computing efficiency.

[0055] The hard drive 301 is fixed in the hard drive slot 2 by the hard drive elastic mounting component 3. During installation, the arc-shaped strip 302 at the front end of the hard drive 301 is used for quick guidance. After being successfully inserted into the slot, the L-shaped mesh pressure plate 303 is tightly attached to the upper and lower surfaces of the hard drive 301 under the elastic force of the pressing spring 304, achieving elastic buffer fixation. This can effectively reduce the damage to the interface and internal components of the hard drive 301 caused by vibration during server transportation and operation. At the same time, the mesh structure of the L-shaped mesh pressure plate 303 can reserve sufficient ventilation gaps at the upper and lower ends of the hard drive 301. Meanwhile, the ventilation slot 16 opened in the main body 1 (located at the upper and lower ends of the hard drive slot 2), the mesh gap of the L-shaped mesh pressure plate 303, and the ventilation slot 17 at the end of the main body 1 away from the hard drive slot 2 together form a continuous ventilation channel, providing a channel for air circulation inside the server, and achieving efficient heat dissipation in conjunction with the heat dissipation mechanism 5.

[0056] When the heat dissipation mechanism 5 is working, the base plate 501 is fixed to the bottom of the inner side of the main body 1 by fastening screw 2 503, which plays a role in bearing and fixing. The snap-fit ​​groove 504 on the fixing bracket 502 realizes the quick snap-fit ​​fixing of the mounting frame 505. The inner arc-shaped mounting plate 507 on the inner wall of the mounting frame 505 fixes the heat sink housing 508 by fastening screw 3 509, ensuring that the heat dissipation mechanism 5 is installed stably. After the motor 511 is powered on, it drives the rotating shaft 512 to rotate. The rotating shaft 512 drives the fan blade 514 to rotate at high speed through the connector 513, which accelerates the air flow inside the server and blows the internal heat to the heat dissipation fins 510. The heat sink housing 508 conducts heat to the heat dissipation fins 510. The heat dissipation fins 510 increase the heat dissipation area and quickly dissipate the heat to the air, realizing the rapid heat dissipation. The dustproof mesh plate 515 on the inner side of the mounting frame 505 can effectively block the external dust from entering the heat dissipation mechanism 5, avoid dust accumulation on the surface of the heat dissipation fins 510 and fan blades 514, and prevent the heat dissipation efficiency from decreasing and the components from failing.

[0057] During the heat dissipation process of CPU component 8, the heat generated by the CPU chip 802 is first conducted to the thermal grease coating 803 on its surface. The thermal grease coating 803 fills the tiny gap between the CPU chip 802 and the heat dissipation plate 804, reducing heat conduction resistance and improving heat conduction efficiency. The heat dissipation plate 804 evenly conducts the heat to the heat-conducting fins 807 above, which quickly dissipate the heat. Combined with the airflow generated by the heat dissipation mechanism 5, the heat dissipation effect of the CPU is further improved. The arc-shaped pressure arms 805 on both sides of the heat dissipation plate 804 are pressed and fixed by the fastening screws 806. The flat end of the arm is attached to the heat dissipation plate 804, and the arc end is pressed against the surface of the motherboard 6, ensuring that the heat dissipation plate 804, the thermal grease coating 803, and the CPU chip 802 are properly aligned. The fit is tight to avoid air gaps caused by loose fit, which would affect heat conduction and ensure stable operation of the CPU chip 802 at a suitable temperature. The heat dissipation slots 4 on both sides of the main body 1 further assist in ventilation and heat dissipation. The cover plate 14 is fixed to the mounting connecting piece 13 at the top of the main body 1 by fastening screws 6 15, which protects the internal components of the main body 1 and prevents dust and debris from entering. At the same time, the fastening screws 6 15 enable a detachable connection, which is convenient for future maintenance, repair and replacement of internal components. The expansion slot 11 and expansion slot 22 on the side of the main body 1 away from the hard drive slot 2 can be used to insert network cards, graphics cards, interface expansion modules, etc., according to actual business needs, so as to realize flexible expansion of server functions, adapt to different application scenarios and improve the versatility and practicality of the server.

[0058] The working principle of this server anomaly handling method is as follows: During normal server operation, a multi-dimensional anomaly monitoring system continuously operates. Through the server's built-in monitoring module, three types of core data are collected synchronously at a frequency of 5 seconds per collection, ensuring the comprehensiveness and real-time nature of anomaly monitoring. Specifically, in-band data is collected through the operating system's built-in monitoring program, covering all core server processes, including operating system kernel logs, CPU usage, memory consumption, disk I / O, and other software-level data, achieving 100% coverage and enabling real-time capture of abnormal signals during software operation. Out-of-band data is collected in real-time through the BMC chip, focusing on the server's hardware health status. The system collects physical data such as CPU temperature, hard drive temperature, power supply voltage, and fan speed, as well as hardware fault alarm information. Data transmission latency is controlled within 100ms to ensure that hardware anomalies can be detected in a timely manner. Business-related data is collected through the business monitoring module, covering all core business modules, including business operation data such as request QPS, response latency, and error rate. The data collection integrity is ≥99.8%, which can accurately reflect the impact of anomalies on business operations. After collection, all types of data are standardized to unify data format, units, and collection frequency, forming a standardized monitoring dataset to avoid affecting the accuracy of subsequent analysis due to inconsistent data formats.

[0059] The standardized monitoring dataset is input into the anomaly prediction model built based on the improved isolated forest algorithm. The model first preprocesses the data to remove outliers generated during the collection process and fill in missing values ​​to ensure the accuracy of the data. Then, 18 key features are extracted from the data. Among them, three new features are added: hardware temperature gradient, process occupancy fluctuation, and peak deviation of business requests. These features can effectively capture the characteristic signals of hidden anomalies and compound anomalies, making up for the shortcomings of existing technologies.

[0060] Meanwhile, by updating model parameters in real time through a 3-minute sliding window, the model can dynamically adapt to changes in server operating status, such as business load fluctuations and hardware operating status changes, thereby improving the adaptability of model predictions. The improved Isolation Forest algorithm addresses the problem of high misjudgment rate in traditional algorithms under business fluctuation scenarios by introducing a business load weight coefficient (dynamically adjusted according to business QPS, with a weight range of 0.1-0.9). When the business QPS is high, the weight coefficient increases, reducing the probability of normal business fluctuations being misjudged as abnormal. After the model runs, it outputs anomaly prediction results and levels 1-10. The system classifies risks into levels, with levels 1-3 representing minor risks, levels 4-7 representing medium risks, and levels 8-10 representing high risks, enabling early warning of anomalies. When the anomaly prediction model outputs a high-risk warning (risk level ≥ 8), or the monitoring module directly detects an actual anomaly (such as excessive CPU temperature, abnormal hard disk I / O, or excessively high business response latency), a multi-source data cross-validation mechanism is immediately activated. Verification is completed within 30 seconds by comparing the anomaly characteristics of in-band, out-of-band, and business-related data to eliminate false anomalies caused by factors such as data mis-collection and transmission interference, ensuring the authenticity of the anomaly signal.

[0061] After successful verification, the system uses graph neural networks to uncover abnormal correlations between the three types of data. For example, excessive CPU temperature may be related to a malfunctioning cooling system (abnormal fan speed in out-of-band data), high CPU utilization (in-band data), and excessively high peak business requests (business-related data). This allows for precise identification of the root cause of the anomaly and a clear distinction between soft faults (such as software process crashes and system vulnerabilities), hard faults (such as cooling system failure, hard drive damage, and poor CPU contact), and business-related anomalies (such as business request overload and interface anomalies). The accuracy rate of the location is ≥98.3%, and the location time is ≤12 minutes. Simultaneously, a location report is automatically generated, clearly recording key information such as the location of the anomaly, its type, its scope of impact, and the number of affected users, providing a clear basis for subsequent repair operations. Based on the location report, the system automatically calls a pre-set atomic repair module library, which contains 12 core repair modules, including two newly added modules: hidden fault repair and cascading anomaly blocking. It can cover various abnormal scenarios; based on the type and scope of the abnormality, it dynamically combines repair strategies and executes automated repair operations: for minor abnormalities that can be repaired online (such as high memory usage or single process abnormality), the system calls the corresponding repair module to automatically close the abnormal process, release memory, and restart related services. The repair time is ≤60 seconds, the repair success rate is ≥99%, and there is no need to interrupt business operations; for serious abnormalities that require offline repair (such as hard drive damage, heat dissipation mechanism failure, or CPU failure), the system first automatically triggers the business migration process to quickly migrate all services carried by the server to the backup server. The migration time is ≤25 seconds, the business interruption time is ≤30 seconds, which is more than 75% shorter than existing technologies, and the migration success rate is ≥99.5%, ensuring that business continuity is not affected. After the business migration is completed, the repair module is called again to prompt staff to perform offline repair operations, such as replacing the damaged hard drive, repairing the heat dissipation mechanism, and reinstalling the CPU.

[0062] After the repair operation is completed, the system enters the repair effect verification phase, which lasts for 5-10 minutes. The monitoring module collects the corresponding data again, comparing the changes before and after the repair to determine whether the anomaly has been completely eliminated, whether server operation has returned to stability, and whether business operations are normal. If the verification passes, the anomaly handling log is automatically updated, recording detailed anomaly information, repair process, and repair results. Simultaneously, the parameters of the anomaly prediction model are updated, and the anomaly data is included in the model training set, forming a closed-loop feedback loop to continuously improve the model's subsequent prediction accuracy. If the verification fails, the repair strategy is adjusted based on the verification results, such as replacing the repair module, adding repair steps, investigating potential related anomalies, and re-executing the repair operation until successful, ensuring an overall repair success rate of ≥98.8%. At the same time, a full-process anomaly handling audit system is built to record anomaly prediction, location, repair, and verification in real time. The entire process of verification data, including data collection time, anomaly characteristics, prediction results, location information, repair strategies, repair duration, and verification results, is retained for at least one year, generating a traceable audit report with 100% audit coverage. This facilitates subsequent problem investigation, accountability, and anomaly pattern analysis by staff. Based on historical audit data, the anomaly prediction model and repair strategies are continuously optimized. Model parameters are updated every 7 days, incorporating newly added anomaly data into the training to further improve the model's prediction accuracy. Repair strategies are optimized every 15 days, adjusting the combination logic and priority of repair modules, streamlining the repair process, and improving repair efficiency. Actual testing shows that the optimized model's prediction accuracy improves by 0.3%-0.5% per month, and repair efficiency improves by 5%-8% per month, continuously enhancing the accuracy and efficiency of server anomaly handling and ensuring long-term stable server operation.

[0063] Therefore, although the invention has been described herein with reference to specific embodiments thereof, freedom of modification, various changes and substitutions are also within the scope of the foregoing disclosure, and it should be understood that in some cases, certain features of the invention may be adopted without departing from the scope and spirit of the invention and without corresponding use of other features. Thus, many modifications can be made to adapt a particular environment or material to the essential scope and spirit of the invention. The invention is not intended to be limited to the specific terminology used in the following claims and / or the specific embodiments disclosed as the best mode for carrying out the invention, but the invention will include any and all embodiments and equivalents falling within the scope of the appended claims. Therefore, the scope of the invention will be defined only by the appended claims.

Claims

1. A server, characterized in that, include: The main body (1) has a hard disk slot (2) with equal horizontal spacing at one end. A hard disk elastic mounting component (3) is installed on the inner side of the main body (1) corresponding to the hard disk slot (2). A heat dissipation mechanism (5) is provided on the inner side of the main body (1) corresponding to the hard disk elastic mounting component (3). A motherboard (6) is provided on the bottom inner side of the main body (1) corresponding to the heat dissipation mechanism (5). A number of fastening screws (7) are provided on the motherboard (6) where there are no electrical components. The motherboard (6) is connected to the main body (1) through the fastening screws (7). A CPU component (8) is installed on one side of the upper end of the motherboard (6). A memory module (9) is provided on the upper end of the motherboard (6) corresponding to the CPU component (8). A power supply (10) is provided on the inner side of the main body (1) corresponding to the motherboard (6). A number of expansion slots (11) and expansion slots (12) are provided on the side of the main body (1) away from the hard disk slot (2).

2. The server according to claim 1, characterized in that, The main body (1) has symmetrical heat dissipation grooves (4) on both sides. The upper end of the main body (1) is open, and the upper end of the main body (1) is provided with mounting connecting pieces (13) at equal intervals on both sides. The upper end of the mounting connecting piece (13) is provided with a cover plate (14). The shape and structure of the cover plate (14) match the shape and structure of the upper end of the main body (1). The cover plate (14) is provided with a fastening screw six (15) through the mounting connecting piece (13). The fastening screw six (15) extends into the mounting connecting piece (13).

3. The server according to claim 1, characterized in that, The hard disk elastic mounting assembly (3) includes a hard disk (301), an arc-shaped strip (302), an L-shaped mesh pressure plate (303), a pressing spring (304), a mounting plate (305), and a first set of fastening screws (306). The front end of the hard disk (301) is symmetrically provided with two arc-shaped strips (302). The upper and lower sides of the hard disk (301) are symmetrically provided with two sets of L-shaped mesh pressure plates (303), pressing springs (304), mounting plates (305), and first set of fastening screws (306) for pressing and fixing the hard disk (301) and for reserving ventilation channels at the upper and lower ends of the hard disk (301). One end of the L-shaped mesh pressure plate (303) is curved upward. There are no less than six pressing springs (304) arranged in a rectangular array between the L-shaped mesh pressure plate (303) and the mounting plate (305). The mounting plate (305) is connected to the main body (1) by the first set of fastening screws (306).

4. The server according to claim 1, characterized in that, Ventilation slot 1 (16) is provided at both the top and bottom of the hard disk slot (2) of the main body (1), and ventilation slot 2 (17) is provided at the end of the main body (1) away from the hard disk slot (2). Ventilation slot 1 (16), L-shaped mesh pressure plate (303) and ventilation slot 2 (17) provide ventilation channels for the main body (1).

5. The server according to claim 1, characterized in that, The heat dissipation mechanism (5) includes a base plate (501), a fixing frame (502), a second fastening screw (503), a locking groove (504), a mounting frame (505), an opening groove (506), an inner arc-shaped mounting plate (507), a radiator shell (508), a third fastening screw (509), heat dissipation fins (510), a motor (511), a rotating shaft (512), a connector (513), a fan blade (514), and a dustproof mesh plate (515). The base plate (501) is fixedly connected to the bottom inner side of the main body (1) by the second fastening screw (503).

6. The server according to claim 5, characterized in that, The base plate (501) is provided with a fixing frame (502) at the upper end. The fixing frame (502) has at least four locking slots (504) equidistantly spaced on the inner side. The locking slots (504) are fitted with an installation frame (505) from top to bottom. The installation frame (505) has an opening slot (506) through it along the ventilation direction of the main body (1). The inner wall of the installation frame (505) is provided with an inner arc-shaped installation plate (507). The inner arc-shaped installation plate (507) is connected to the radiator housing (508) by three fastening screws (509).

7. The server according to claim 6, characterized in that, The heat sink housing (508) has heat dissipation fins (510) on one end surface. A motor (511) is provided at the middle of the heat dissipation fins (510) at one end of the heat sink housing (508). A rotating shaft (512) is connected to the output end of the motor (511) for forward transmission. A fan blade (514) is connected to the front end of the rotating shaft (512) through a connector (513). A dustproof mesh plate (515) is provided on the inner side of the mounting frame (505) corresponding to the front end of the fan blade (514).

8. The server according to claim 1, characterized in that, The CPU assembly (8) includes a CPU socket (801), a CPU chip (802), a thermal grease coating (803), a heat spreader plate (804), an arc-shaped pressure arm (805), five fastening screws (806), and heat-conducting fins (807). The CPU socket (801) is located on the upper end of the motherboard (6). The CPU chip body (802) is inserted into the CPU socket (801). The upper surface of the CPU chip body (802) is coated with a thermal grease coating (803). A heat spreader plate (804) is placed on top of the thermal grease coating (803). Heat-conducting fins (807) are placed on top of the heat spreader plate (804).

9. The server according to claim 8, characterized in that, The heat-conducting plate (804) has arc-shaped pressure arms (805) on both sides. The arc-shaped pressure arms (805) are flat near the heat-conducting plate (804) and arc-shaped away from the heat-conducting plate (804). The arc-shaped front end of the arc-shaped pressure arm (805) is pressed downward onto the upper surface of the motherboard (6). The flat part of the arc-shaped pressure arm (805) is provided with a fastening screw five (806), and the fastening screw five (806) extends into the motherboard (6) to realize the heat-conducting plate (804) pressing and fixing the thermal grease coating (803) on the CPU chip (802).

10. A server exception handling method, employing the server described in any one of claims 1-9, characterized in that, include: S1. Multi-source data acquisition and standardized processing: Construct a multi-dimensional anomaly monitoring system, simultaneously collecting in-band data, out-of-band data, and business-related data from the server. The acquisition frequency is set to 5 seconds / time to form a standardized monitoring dataset. The in-band data includes software-level data such as operating system kernel logs, CPU utilization, memory usage, and disk I / O, covering all core processes of the server with 100% coverage. The out-of-band data includes physical-level data such as server hardware health status and hardware fault alarms, collected in real time through the BMC chip with a data transmission latency ≤100ms. The business-related data includes business operation data such as request QPS, response latency, and error rate, covering all core business modules with a data acquisition completeness ≥99.8%. S2. Anomaly Prediction and Risk Classification: An anomaly prediction model is constructed based on an improved isolated forest algorithm. The standardized monitoring dataset is preprocessed and features are extracted, resulting in 18 key features (adding three new features compared to existing models: hardware temperature gradient, process occupancy fluctuation, and peak deviation of business requests). Model parameters are updated in real-time using a sliding window (window size set to 3 minutes) to achieve early prediction of latent and complex anomalies, outputting the anomaly prediction results and risk levels (1-10). The improved isolated forest algorithm introduces a business load weight coefficient (dynamically adjusted according to business QPS, weight range 0.1-0.9) to address the high false positive rate of traditional algorithms in scenarios with fluctuating business operations. The model's prediction accuracy is ≥99.2%, recall is ≥95.5%, and latent anomaly identification rate is ≥88%, representing an improvement of over 28 percentage points compared to existing technologies. S3. Anomaly Root Cause Locator and Report Generation: When the anomaly prediction model outputs a high-risk warning (risk level ≥ 8) or detects an actual anomaly, a multi-source data cross-validation mechanism is activated. The validation time is ≤ 30 seconds. Combining in-band, out-of-band, and business-related data, the graph neural network is used to mine anomaly correlations, locate the anomaly root cause, distinguish between soft faults, hard faults, and business-related anomalies, with a location accuracy of ≥ 98.3% and a location time of ≤ 12 minutes. A location report containing the anomaly location, anomaly type, scope of impact, and number of affected users is generated. S4. Tiered Automated Repair and Business Assurance: Based on anomaly location reports, a pre-defined atomic repair module library is invoked. The module library contains 12 core repair modules (two new modules added compared to the existing ones: hidden fault repair and cascading anomaly blocking). Repair strategies are dynamically combined according to the anomaly type and impact scope to execute automated repair operations. For minor anomalies that can be repaired online, the repair time is ≤60 seconds and the repair success rate is ≥99%. For severe anomalies that require offline repair, the business migration process is automatically triggered, with a migration time of ≤25 seconds and a business interruption duration of ≤30 seconds, which is more than 75% shorter than existing technologies and a migration success rate of ≥99.5%, before the repair operation is executed. S5. Repair Effect Verification and Closed-Loop Feedback: After the repair is completed, the repair effect is verified. The verification time is set to 5-10 minutes. If the verification is successful, the exception handling log and model parameters are updated to form a closed-loop feedback. If the verification fails, the repair strategy is adjusted and the repair operation is re-executed until the repair is successful. The overall repair success rate is ≥98.8%, which is more than 33 percentage points higher than the existing technology. S6. End-to-End Audit and Strategy Optimization: Establish an end-to-end audit system for anomaly handling, record data throughout the entire process of anomaly prediction, location, repair, and verification, retain data for at least 1 year, generate traceable audit reports, and achieve 100% audit coverage; simultaneously, continuously optimize the anomaly prediction model and repair strategy based on historical data, update model parameters every 7 days, and optimize repair strategies every 15 days. After optimization, the model's prediction accuracy improves by 0.3%-0.5% per month, and the repair efficiency improves by 5%-8% per month.

Images