CN118708389A - A lightweight dump configuration method - Google Patents

Abstract

The invention relates to the field of dump configuration, and discloses a lightweight dump configuration method, comprising: S1, obtaining software exception information, and obtaining an abnormal business scenario based on the software exception information; S2, triggering the dump lightweight configuration based on the abnormal business scenario; S3, optimizing a dump process information list based on a gray wolf optimization algorithm, obtaining optimal dump process information, and performing data reading according to the optimal dump process information; the S3 further comprises: S31, after the dump lightweight configuration is triggered, creating a gray wolf optimization algorithm model of a dump process information list based on the abnormal business scenario; S32, simulating a gray wolf hunting behavior based on the gray wolf optimization algorithm model of the dump process information list to iteratively update a gray wolf position, obtaining a final alpha wolf position, and parsing the final alpha wolf position to obtain the optimal dump process information; the application can perform optimal dump configuration to improve the flexibility and efficiency of the dump configuration.

Description

A lightweight dump configuration method

Technical Field

The present invention relates to the field of dump configuration, and in particular to a lightweight dump configuration method.

Background Art

When the software freezes, crashes, or does not run as expected, dump is one of the important problem analysis and location methods after the software is launched. Dump can assist developers in accurately finding the cause of the problem by recording and analyzing memory and stack-related information. Compared with other analysis and location methods, dump is accurate and has complete on-site restoration.

However, existing dump methods have the following shortcomings: (1) The dump efficiency is low or the information is insufficient. Using fulldump takes a long time, is inefficient, and has a great impact on the program. Although using minidump can improve efficiency, it does not provide enough information and is not conducive to troubleshooting. (2) It is not flexible enough. The dump trigger conditions of the program are diverse, and there are differences in the focus of different scenarios. Although fulldump is convenient for troubleshooting, the dump takes up a lot of resources and affects the user experience, especially in scenarios where dumps are generated frequently. Using minidump to troubleshoot problems often does not provide enough information and requires contacting the customer for on-site debugging.

Summary of the invention

The present invention aims to provide a lightweight dump configuration method, so as to perform optimal dump configuration when software exception occurs, so as to improve the flexibility and efficiency of dump configuration.

In order to achieve the above object, the present invention adopts the following technical scheme:

A lightweight dump configuration method includes:

S1, obtain software exception information, and obtain abnormal business scenarios based on the software exception information;

S2, triggers dump lightweight configuration based on abnormal business scenarios;

S3, optimizing the dump process information list based on the gray wolf optimization algorithm to obtain optimal dump process information, and reading data according to the optimal dump process information; S3 also includes:

S31, after the dump lightweight configuration is triggered, the Gray Wolf optimization algorithm model for creating a dump process information list based on abnormal business scenarios;

S32, the gray wolf optimization algorithm model based on the dump process information list simulates the gray wolf hunting behavior to iteratively update the gray wolf position, obtain the final α wolf position, and parse the final α wolf position to obtain the optimal dump process information.

The principles and advantages of this solution are: in actual application, software exception information is obtained, and abnormal business scenarios are obtained based on the software exception information. For specific business scenarios, the focus is different, and the dumped process information also needs to be extracted differently. Obtaining business scenarios is convenient for providing an analysis basis for dump lightweight configuration; dump lightweight configuration is triggered based on abnormal business scenarios to avoid software exceptions that are not adaptive, thereby improving the effectiveness and accuracy of dump lightweight configuration; optimization of dump process information is the key to achieving lightweight configuration. When there is too much dump process information, there are problems such as high resource usage and low efficiency. When there is too little information, although the efficiency is improved, there is still a problem that the information is not enough to be conducive to troubleshooting; based on the gray wolf optimization algorithm, the dump process information list is optimized. After the dump lightweight configuration is triggered, a gray wolf optimization algorithm model of the dump process information list is created based on the abnormal business scenario. The gray wolf optimization algorithm model based on the dump process information list simulates the gray wolf hunting behavior to iteratively update the gray wolf position, obtain the final α wolf position, and parse the final α wolf position to obtain the optimal dump process information. Data is read according to the optimal dump process information, thereby realizing the optimization of the dump configuration and improving the flexibility and efficiency of the dump configuration.

Preferably, as an improvement, the S2 further includes:

S21, retrieving and parsing the access control list to verify whether the user requesting access is included in the user whitelist;

S22, when the verification is yes, triggers the dump lightweight configuration.

Technical effect: By setting up a trigger mechanism, it is convenient to strengthen software security protection.

Preferably, as an improvement, the dump process information list includes system information, thread stack information, thread name, handle, module list, uninstall module list and memory.

Technical effect: The dump process information list is the dump process information configured by fulldump. Based on the combination of the above dump process information, it is easy to adapt to different business scenarios.

Preferably, as an improvement, the gray wolf optimization algorithm model for creating a dump process information list based on abnormal business scenarios includes:

S311, initializing the individual positions of the gray wolf population as a dump process information combination vector;

S312, determining a fitness function expression according to the dump process information combination speed, completeness and redundancy;

S313, determining the constraint conditions of the grey wolf optimization algorithm.

Technical effect: Through the above steps, the Gray Wolf optimization algorithm can be integrated with the dump process information of the dump lightweight configuration.

Preferably, as an improvement, the fitness function expression is:

Among them, ρ ₁ , ρ ₂ , ρ ₃ are optimization weight values, v _x is the dump process information combination speed; w _x is the completeness of the dump process information combination; r _x is the redundancy of the dump process information combination; μ is the fitness adjustment constant value, and f is the completeness judgment constant value.

Technical effect: Through the above formula, the combination completeness is used as the main influencing factor of the fitness function to ensure that the completeness of the optimal solution meets the standard.

Preferably, as an improvement, the completeness expression of the dump process information combination is:

Among them, _M0 is the effective amount of dump process information combination, _θ1 and _θ2 are the adjustment constant coefficients of usage and suspension respectively, _Mq is the amount of dump process information used by software exception problem q, and _Gq is the amount of dump process information suspended by software exception problem q.

Technical effect: By combining the suspension item with the completeness of the dump process information combination through the above formula, it is convenient to increase the software exception handling capability.

Preferably, as an improvement, the constraint condition is:

(t _m ＜T _m )and(minf _h ≥h)

Among them, _tm is the optimization solution time, _Tm is the time threshold constant, _fh is the correlation between any two items in the dump process information combination obtained in the current abnormal business scenario, and h is the correlation threshold constant.

Technical effect: The number of classification models in the optimization results is limited based on constraints to avoid a large number of useless operations.

Preferably, as an improvement, in the process of iteratively updating the position of the gray wolf, a gray wolf individual is randomly selected from the group to perform a differential search with the current individual, and the historical optimal solution of the gray wolf individual is memorized and saved, and the position update formula of the wolf group is obtained as follows:

Among them, _b1 is the individual memory coefficient, _b1∈ [0,1]; _b2 is the communication coefficient, _b2∈ [0,1]; _r3 and _r4 are random numbers in [0,1]; Pi _,best is the best position experienced by the i-th gray wolf individual; _Xj is a gray wolf individual randomly selected from the group, j≠i; Xi _,1 , Xi _,2 , Xi _,3 represent the updated positions of α, β and δ wolves respectively, and X is the current position of the gray wolf individual.

Technical effect: Through the above algorithm, the memory function of individual gray wolves can be improved, thereby enhancing the exploration ability and convergence speed of the gray wolf optimization algorithm.

Preferably, as an improvement, the calculation formula for the updated positions of the α, β and δ wolves is:

Xi _,1 = _Xα - _{A1Dα ;}

Xi _,2 = _Xβ - _{A2Dβ ;}

Xi _,3 = _Xδ - _{A3Dδ ;}

Among them, A ₁ , A ₂ and A ₃ are distance coefficient vectors; D _α , D _β , D _δ represent the distances between other ω wolves and the three best gray wolves α, β and δ; X _α , X _β and X _δ represent the positions of α, β and δ wolves respectively.

Technical effect: Based on the above algorithm, it is possible to simulate the gray wolf's encirclement hunting mechanism, update the position of the wolf pack, and thus find the optimal solution.

Preferably, as an improvement, the convergence factor update formula in the process of iteratively updating the gray wolf position is:

a _t =a _i -(a _i -a _f )*rand()+ρ*randn()

Where a _i and a _f represent the initial value and final value of the convergence factor, respectively; t is the current number of iterations; rand() is a random number from [0,1] that follows a uniform distribution; randn() is a random number from a normal distribution; and ρ is the variance that measures the degree of deviation between the convergence factor and its mathematical expectation.

Technical effect: Through the above algorithm, there is a chance to obtain a larger or smaller value in the early stage of iteration, and a chance to obtain a smaller or larger value in the later stage of iteration, which makes it easier for iteration to jump out of the local optimum, thereby accelerating the convergence speed.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a lightweight dump configuration method.

DETAILED DESCRIPTION

The following is further described in detail through specific implementation methods:

The embodiment is basically as shown in Figure 1:

A lightweight dump configuration method includes:

S1, obtain software exception information, and obtain abnormal business scenarios based on the software exception information; in this embodiment, the software exception information is identified based on the collaborative work of the software monitoring module; abnormal business scenarios such as audio and video freezes, UI freezes, database access crashes, database access freezes, etc., specific business scenarios have different focus points, and the dumped process information also needs to be extracted differently.

S2, triggering dump lightweight configuration based on abnormal business scenarios; in order to ensure software security, S2 also includes: S21, retrieving and parsing the access control list to verify whether the user requesting access is included in the user whitelist; S22, when the verification is yes, triggering dump lightweight configuration.

S3, optimizes the dump process information list based on the Gray Wolf optimization algorithm to obtain the optimal dump process information, and reads data according to the optimal dump process information; the dump process information list includes system information, thread stack information, thread name, handle, module list, uninstall module list and memory. Take Windows as an example: system information obtains the process creation time, exit time, process running time in kernel state and user state by using the system apiGetProcessTimes; thread stack information obtains the process information SYSTEM_PROCESS_INFORMATION by using NtQuerySystemInformation to specify the parameter SystemProcessInformation, access the SYSTEM_PROCESS_INFORMATION structure to obtain the number of threads and the SYSTEM_THREAD_INFORMATION of each thread, obtain the thread id, use the system apiNtQueryInformationThread to query the tib of the thread id, and obtain the starting address and size of the thread stack from the tib; the thread name is obtained through the api GetThreadDescription; the handle is obtained by specifying the SystemExtendedHandleInformation parameter using NtQuerySystemInformation; the module list information is stored in the process PEB structure, and the address of the PEB is obtained through NtQueryInformationProcess. Since the structure of the PEB is different in different Windows versions, the structure of the PEB is determined according to the target operating system version. The obtained PEB address is used to convert it into a pointer to the PEB structure to access its members. In the PEB structure, a member named Ldr is obtained. The PEB_LDR_DATA structure contains the loaded module list information. By accessing the InMemoryOrderModuleList member in the PEB_LDR_DATA structure, a bidirectional linked list is obtained, which contains the loaded module list information. The linked list traversal method is used to access and obtain the information of each module one by one, such as the module base address, module name, etc.; the unloaded module list uses the system api RtlGetUnloadEventTraceEx to obtain the name, base address, and size information of the unloaded module; the memory uses the system api ReadProcessMemory to read the process memory. When fulldump is used, it is necessary to obtain all dump process information in the dump process information list, which takes a long time and is inefficient. When minidump is used based on experience, due to the uncertainty of the problem, there is a situation where the dump process information is insufficient and the flexibility is poor. Therefore, the optimization of the dump process information is the key to achieving lightweight configuration. The S3 also includes:

S31, after the dump lightweight configuration is triggered, the Gray Wolf optimization algorithm model for creating a dump process information list based on abnormal business scenarios, specifically, including:

S311, initialize the individual positions of the gray wolf population as the dump process information combination vector; also includes the initialization of the population size, the maximum number of iterations, and the parameters a, A and C, where a is the convergence factor, A and C are the synergy coefficient vectors; in this application, the individual positions of the gray wolves and the parameters a, A and C are randomly generated.

S312, determining a fitness function expression according to the dump process information combination speed, completeness and redundancy; the fitness expression is:

Among them, ρ ₁ , ρ ₂ , and ρ ₃ are optimization weight values, v _x is the speed of dump process information combination; w _x is the completeness of dump process information combination; r _x is the redundancy of dump process information combination, and w _x and r _x are both obtained based on the dump process information combination of historical abnormal business scenarios; μ is the fitness adjustment constant value, and f is the completeness judgment constant value. When the completeness of the dump process information combination is insufficient, the μ value can be adjusted to quickly eliminate the difference solution and promote rapid convergence.

Since the causes of software anomalies are different even if they show the same or similar situations, the required dump process information is also different when dump analysis is used. In this embodiment, in order to increase the software anomaly handling capability, a suspension item is added during the completeness acquisition process of the dump process information combination, specifically:

Among them, _M0 is the effective amount of dump process information combination, _θ1 and _θ2 are the adjustment constant coefficients of usage and suspension respectively, _Mq is the amount of dump process information used by software exception problem q, and _Gq is the amount of dump process information suspended by software exception problem q; suspension means that although it is configured, the usage rate is lower than the threshold.

S313, determining the constraint conditions of the gray wolf optimization algorithm, wherein the constraint conditions are:

(t _m <T _m ) and (minfh _h ≥ h)

Among them, _tm is the optimization solution time, _Tm is the time threshold constant, _fh is the correlation between any two items in the dump process information combination obtained in the current abnormal business scenario, and this correlation is obtained based on historical data clustering, and h is the correlation threshold constant.

S32, the gray wolf optimization algorithm model based on the dump process information list simulates the gray wolf hunting behavior to iteratively update the gray wolf position, obtain the final α wolf position, and parse the final α wolf position to obtain the optimal dump process information.

In the process of iteratively updating the position of the gray wolf, a gray wolf individual is randomly selected from the group to perform differential search with the current individual, and the historical optimal solution of the gray wolf individual is memorized and saved, and the memory function of the gray wolf individual is improved, thereby enhancing the exploration ability and convergence speed of the gray wolf optimization algorithm, and the position update formula of the wolf group is obtained as follows:

Among them, _b1 is the individual memory coefficient, _b1∈ [0,1]; _b2 is the communication coefficient, _b2∈ [0,1]; _r3 and _r4 are random numbers in [0,1]; Pi _,best is the best position experienced by the i-th gray wolf individual; _Xj is a gray wolf individual randomly selected from the group, j≠i; Xi _,1 , Xi _,2 , Xi _,3 represent the updated positions of α, β and δ wolves respectively, and X is the current position of the gray wolf individual.

The calculation formula for the updated position of the α, β and δ wolves is:

Xi _,1 = _Xα - _{A1Dα ;}

Xi _,2 = _Xβ - _{A2Dβ ;}

Xi _,3 = _Xδ - _{A3Dδ ;}

Among them, A ₁ , A ₂ and A ₃ are distance coefficient vectors; the calculation formula of A (A ₁ , A ₂ or A ₃ ) is:

A＝2ar ₁ -a

Where a is the convergence factor, r ₁ is a random vector in [0,1];

D _α , D _β , D _δ represent the distances between other ω wolves and the three best gray wolves α, β, and δ; X _α , X _{β ,} and X _δ represent the positions of α, β, and δ wolves, respectively.

D _α＝ |C ₁ *X _α −X|;

D _β＝ |C ₂ *X _β -X|;

D _δ＝ |C ₃ *X _δ -X|;

C ₁ , C ₂ and C ₃ are position coefficient vectors, and C ₁ , C ₂ and C ₃ are all random numbers between 0 and 2. By adjusting (C>1) or (C<1) to emphasize or weaken the influence of prey in the distance calculation formula, it is helpful to avoid local optimality. C is C ₁ , C ₂ or C ₃ .

In order to better jump out of the local optimal solution and speed up the convergence speed, there is a chance to obtain a larger or smaller value in the early stage of iteration, and a chance to obtain a smaller or larger value in the later stage of iteration. The convergence factor update formula in the process of iteratively updating the gray wolf position is:

a _t =a _i -(a _i -a _f )*rand()+ρ*randn()

Where a _i and a _f represent the initial value and final value of the convergence factor, respectively; t is the current number of iterations; rand() is a random number from [0,1] that follows a uniform distribution; randn() is a random number from a normal distribution; and ρ is the variance that measures the degree of deviation between the convergence factor and its mathematical expectation.

After obtaining the optimal dump process information, the dump configuration and reading can be performed. In this embodiment, it is read through a local configuration file or database, or pulled from a server. Take pulling from a server as an example: the administrator sets up an Httpserver server in the background, modifies the configuration file, and configures the pulling module to send an http request to the server to obtain the configuration file; the configuration file format adopts any format, such as ini, json, xml, yaml format; minidump is lightweight and cross-platform, and is supported by multiple debuggers, such as windbg and lldb.

The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and/or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the technical solution of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims (10)

1. A lightweight dump configuration method, characterized by comprising: S1, obtain software exception information, and obtain abnormal business scenarios based on the software exception information; S2, triggers dump lightweight configuration based on abnormal business scenarios; S3, optimizing the dump process information list based on the gray wolf optimization algorithm to obtain optimal dump process information, and reading data according to the optimal dump process information; S3 also includes: S31, after the dump lightweight configuration is triggered, the Gray Wolf optimization algorithm model for creating a dump process information list based on abnormal business scenarios; S32, the gray wolf optimization algorithm model based on the dump process information list simulates the gray wolf hunting behavior to iteratively update the gray wolf position, obtain the final α wolf position, and parse the final α wolf position to obtain the optimal dump process information. 2. According to the method for configuring a lightweight dump in claim 1, it is characterized in that said S2 further comprises: S21, retrieving and parsing the access control list to verify whether the user requesting access is included in the user whitelist; S22, when the verification is yes, triggers the dump lightweight configuration. 3. A lightweight dump configuration method according to claim 1, characterized in that: the dump process information list includes system information, thread stack information, thread name, handle, module list, uninstall module list and memory. 4. A lightweight dump configuration method according to claim 1, characterized in that the gray wolf optimization algorithm model for creating a dump process information list based on abnormal business scenarios includes: S311, initializing the individual positions of the gray wolf population as a dump process information combination vector; S312, determining a fitness function expression according to the dump process information combination speed, completeness and redundancy; S313, determining the constraint conditions of the grey wolf optimization algorithm. 5. A lightweight dump configuration method according to claim 4, characterized in that the fitness function expression is: Among them, ρ ₁ , ρ ₂ , ρ ₃ are optimization weight values, v _x is the dump process information combination speed; w _x is the completeness of the dump process information combination; r _x is the redundancy of the dump process information combination; μ is the fitness adjustment constant value, and f is the completeness judgment constant value. 6. A lightweight dump configuration method according to claim 5, characterized in that the completeness expression of the dump process information combination is: Among them, _M0 is the effective amount of dump process information combination, _θ1 and _θ2 are the adjustment constant coefficients of usage and suspension respectively, _Mq is the amount of dump process information used by software exception problem q, and _Gq is the amount of dump process information suspended by software exception problem q. 7. A lightweight dump configuration method according to claim 4, characterized in that the constraint condition is: (t _m ＜T _m )and(minf _h ≥h) Among them, _tm is the optimization solution time, _Tm is the time threshold constant, _fh is the correlation between any two items in the dump process information combination obtained in the current abnormal business scenario, and h is the correlation threshold constant. 8. A lightweight dump configuration method according to claim 1, characterized in that, in the process of iteratively updating the position of the gray wolf, a gray wolf individual is randomly selected from the group and the current individual is searched for difference, and the historical optimal solution of the gray wolf individual is memorized and saved, and the position update formula of the wolf group is obtained as follows: Wherein, _b1 is the individual memory coefficient, _b1∈ [0,1]; _b2 is the communication coefficient, _b2∈ [0,1]; _r3 and _r4 are random numbers in [0,1]; Pi _,est is the best position experienced by the i-th gray wolf individual; _Xj is a randomly selected gray wolf individual in the group, j≠i; Xi _,, Xi _,, Xi _, respectively represent the updated positions of α, β and δ wolves, and X is the current position of the gray wolf individual. 9. A lightweight dump configuration method according to claim 8, characterized in that the calculation formula for the update position of the α, β and δ wolves is: Xi _,1 = _Xα - _{A1Dα ;} Xi _,2 = _Xβ - _{A2Dβ ;} Xi _,3 = _Xδ - _{A3Dδ ;} Among them, A ₁ , A ₂ and A ₃ are distance coefficient vectors; D _α , D _β , D _δ represent the distances between other ω wolves and the three best gray wolves α, β and δ; X _α , X _β and X _δ represent the positions of α, β and δ wolves respectively. 10. A lightweight dump configuration method according to claim 1, characterized in that the convergence factor update formula in the process of iteratively updating the gray wolf position is: a _t =a _i -(a _i -a _f )*rand()+ρ*randn() Where a _i and a _f represent the initial value and final value of the convergence factor, respectively; t is the current number of iterations; rand() is a random number from [0,1] that follows a uniform distribution; randn() is a random number from a normal distribution; and ρ is the variance that measures the degree of deviation between the convergence factor and its mathematical expectation.