Monitoring and warning method based on java application

Abstract

The invention discloses a monitoring and warning method based on java application, which belongs to the technical field of data processing, and comprises the following steps: after a monitoring center is started, heartbeat signals are generated in a rated period; acquiring state values of the entrance monitoring burying point and the JVM monitoring burying point when the heartbeat signal is generated each time; taking the state value with high severity in the state value of the entrance monitoring burying point and the state value of the JVM monitoring burying point as the state value of the monitoring center; if the state value of the monitoring center is smaller than the warning state value, a report trigger is triggered; if the state value of the monitoring center is greater than or equal to the warning state value, triggering a warning trigger; when the accumulated triggering times of the report trigger exceed a report threshold value, a brief report is formed, and when the accumulated triggering times of the warning trigger exceed a warning threshold value, warning detailed information is formed. According to the monitoring and warning method based on the java application, the problems that all working threads work and external requests are completely blocked in an existing monitoring mode are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a monitoring and warning method based on Java application. Background Art

[0002] Currently, the main Java application monitoring methods available on the market include system load monitoring, JVM heap size monitoring, CPU usage monitoring, QPS monitoring, and average response time monitoring. However, these monitoring methods only capture the current system status and therefore cannot accurately assess system health. For example, assuming the JVM heap size is currently large, it may return to normal after garbage collection the next second. Another example is CPU usage. With many applications on a server, even if CPU usage is temporarily high, it can be difficult to determine which Java application node is causing the problem.

[0003] Therefore, it's currently difficult to monitor Java applications specifically for threads currently processing tasks. Java applications have a limited number of worker threads, with each request handled by a worker thread. Each worker thread is released only after completing the current request. Therefore, without monitoring of threads currently processing tasks, all worker threads will be busy, completely blocking external requests. This can cause users to experience a service outage. Summary of the Invention

[0004] In order to overcome the defects of the prior art, the present invention provides a monitoring and warning method based on Java application to solve the above problems.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a monitoring and warning method based on Java application, comprising the following steps:

[0006] S1: After the Java application is started, the monitoring center generates a heartbeat signal at a specified period; in this solution, the period is 1 second.

[0007] S2: Each time a heartbeat signal is generated, the monitoring center sequentially obtains the status values ​​of the entry monitoring point and the JVM monitoring point after status checks; the status values ​​include normal status value, warning status value, warning status value and forced protection status value arranged in ascending order;

[0008] S3: The status value of the entry monitoring tracking point and the status value of the JVM monitoring tracking point with a higher severity is used as the monitoring center status value corresponding to the current heartbeat signal;

[0009] S4: If the monitoring center status value corresponding to the current heartbeat signal is less than the warning status value, the report trigger is triggered and the cumulative number of triggers of the report trigger is recorded; if the monitoring center status value corresponding to the current heartbeat signal is greater than or equal to the warning status value, the warning trigger is triggered and the cumulative number of triggers of the warning trigger is recorded;

[0010] When the cumulative number of triggers of the report trigger exceeds the preset report threshold, the cumulative number of triggers of the report trigger is cleared and a brief report is generated and sent to the monitoring center. When the cumulative number of triggers of the warning trigger exceeds the preset warning threshold, the cumulative number of triggers of the warning trigger is cleared and detailed warning information is generated and sent to the monitoring center.

[0011] It is worth noting that in step S2, for the status check of the entry monitoring point, each time a heartbeat signal is generated, the entry monitoring point obtains the value of the number of threads inPointCount;

[0012] The initial status value of the entrance monitoring point is the normal status value;

[0013] When inPointCount>A, the status value changes from the normal status value to the warning status value;

[0014] In the case where the state value is the warning state value, when inPointCount>A in B consecutive heartbeat signals, the state value is converted from the warning state value to the warning state value;

[0015] When inPointCount>D in C consecutive heartbeat signals, the status value changes from the warning status value to the forced protection status value;

[0016] When the state value is an early warning state value, a warning state value, or a forced protection state value, and inPointCount≤A in E consecutive heartbeat signals, the state value is converted to a normal state value.

[0017] Preferably, in step S2, when the state value is a forced protection state value, the entry monitoring embedding point calls the thread record table, and adds a current limiting strategy for all entry methods in the thread record table;

[0018] When the status value is converted from the mandatory protection status value to other status values, the current limiting policy added to the entry method is removed.

[0019] Optionally, in step S2, when a request passes through a slice, the in method of the entry monitoring tracking point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is increased by 1;

[0020] When the business processing corresponding to the request is completed and returned to the aspect, the out method of the entry monitoring point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is reduced by 1.

[0021] Specifically, in step S2, the entry monitoring tracking point obtains the entry method name and entry method parameters obtained by the web interceptor or the method name and parameters obtained by the dubbo interceptor through the in(String method,Object args) method, where String method represents the entry method name and Object args represents the entry method parameters;

[0022] Then call the thread record table to record the thread information corresponding to the entry method name and entry method parameters.

[0023] It is worth noting that, in step S2, the web interceptor obtains the entry URL path and parameters as the entry method name and entry method parameters through the HttpServletRequest request parameters in the interface method in the general interface of java web.

[0024] Preferably, in step S2, the dubbo interceptor obtains the entry method name and entry method parameters through the Invoker<?>invoker,Invocation invocation parameter in the interface method in the Filter interface.

[0025] Optionally, in step S2, the entry monitoring point calls the thread record table through the out() method to clear the corresponding thread information in the thread record table.

[0026] Specifically, in step S2, for the status check of the JVM monitoring tracking point, the initial status value of the JVM monitoring tracking point is a normal status value;

[0027] When the heap usage exceeds G in F consecutive heartbeat signals, the status value changes from normal to warning.

[0028] In the case where the state value is a warning state value, when the heap usage rate in H consecutive heartbeat signals is lower than 1, the state value is converted from the warning state value to the normal state value.

[0029] Preferably, the JVM monitoring embedding point obtains the current heap size and heap upper limit by calling the Java language tool ManagementFactory.getMemoryMXBean(), and the heap usage rate = current heap size / heap upper limit.

[0030] The beneficial effects of the present invention are as follows: in the monitoring and warning method based on Java application, a report trigger is configured to send a report threshold to the monitoring center once, the cumulative trigger count of the report trigger is initially 0, and when the monitoring center status value is less than the warning status value, each time a heartbeat signal is generated, the cumulative trigger count is increased by 1, and when the upper limit, that is, when the reporting threshold is reached, the count is set to 0; whenever the cumulative trigger count of the report trigger is 0, the report trigger obtains the value of the thread count inPointCount of the entry monitoring tracking point, adds the heap status obtained from the JVM monitoring tracking point as a report, and then calls a message sender to send the report to the monitoring center. The warning trigger is configured to send a warning threshold to the monitoring center once, the cumulative trigger count of the warning trigger is initially 0, and when the monitoring center status value is greater than or equal to the warning status value, each time a heartbeat signal is generated, the cumulative trigger count is increased by 1, and when the upper limit, that is, when the warning threshold is reached, the count is set to 0; whenever the cumulative trigger count of the warning trigger is 0, the information in the thread record table and the heap status obtained from the JVM monitoring tracking point are obtained as warning details, and then the detailed information is sent to the monitoring center using a message sender. The thread count inPointCount, tracked through entry monitoring, reflects the Java application's worker thread usage. The thread log table shows which entry points are called most frequently and which entry points take too long to call. The heap usage rate, obtained through JVM monitoring, reveals heap usage. When the heap is nearly full, the thread log table identifies the entry point causing the heap to be full. This links memory usage with call activity, making it easier to monitor threads processing tasks and avoid situations where all worker threads are busy. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A monitoring and warning method based on Java application in one embodiment of the present invention;

[0032] Figure 2 Schematic diagram of a thread record table in one embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following is a further description of specific embodiments of the present invention in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is intended to facilitate understanding of the present invention and does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0034] like Figure 1 and 2As shown, a monitoring and warning method based on Java application includes the following steps:

[0035] S1: After the Java application is started, the monitoring center generates a heartbeat signal at a specified period.

[0036] S2: Each time a heartbeat signal is generated, the monitoring center sequentially obtains the status values ​​of the entry monitoring point and the JVM monitoring point after status checks; the status values ​​include normal status value, warning status value, warning status value and forced protection status value arranged in ascending order;

[0037] S3: The status value of the entry monitoring tracking point and the status value of the JVM monitoring tracking point with a higher severity is used as the monitoring center status value corresponding to the current heartbeat signal;

[0038] S4: If the monitoring center status value corresponding to the current heartbeat signal is less than the warning status value, the report trigger is triggered and the cumulative number of triggers of the report trigger is recorded; if the monitoring center status value corresponding to the current heartbeat signal is greater than or equal to the warning status value, the warning trigger is triggered and the cumulative number of triggers of the warning trigger is recorded;

[0039] When the cumulative number of triggers of the report trigger exceeds the preset report threshold, the cumulative number of triggers of the report trigger is cleared and a brief report is generated and sent to the monitoring center. When the cumulative number of triggers of the warning trigger exceeds the preset warning threshold, the cumulative number of triggers of the warning trigger is cleared and detailed warning information is generated and sent to the monitoring center.

[0040] In the monitoring and warning method based on Java application, a reporting threshold for sending a briefing to the monitoring center is configured for the reporting trigger, that is, a briefing is sent to the outside world every certain number of heartbeat signals. In this embodiment, the reporting threshold corresponding to the reporting trigger is 60 times, and each heartbeat signal is 1 second, so the interval is 60 times, which is one minute; the cumulative triggering number of the reporting trigger is initially 0. When the status value of the monitoring center is less than the warning status value, the cumulative triggering number is increased by 1 each time a heartbeat signal is generated. When the upper limit is reached, that is, when the reporting threshold of 60 is reached, the number is set to 0; whenever the cumulative triggering number of the reporting trigger is 0, the reporting trigger will obtain the value of the number of threads inPointCount of the entry monitoring point, plus the heap status (heap size and heap usage) obtained from the JVM monitoring point as a briefing, and then call the message sender to send the briefing to the monitoring center. The warning trigger is configured to send a warning threshold for sending detailed warning information to the monitoring center. In this embodiment, the warning threshold corresponding to the warning trigger is configured to 30 times. The initial value of the cumulative trigger count of the warning trigger is 0. When the monitoring center status value is greater than or equal to the warning status value, the cumulative trigger count is increased by 1 each time a heartbeat signal is generated. When the upper limit is reached, that is, when the warning threshold of 60 is reached, the count is set to 0. Whenever the cumulative trigger count of the warning trigger is 0, the information in the thread record table (thread id, thread name, entry method name, entry method parameters, entry method call time, SQL being executed and SQL start time) and the heap status obtained by the JVM monitoring point are obtained as warning details, and then the detailed information is sent to the monitoring center using a message sender. The number of threads inPointCount through the entry monitoring point reflects the situation of the Java application's working threads being occupied. At this time, it can be seen from the thread record table which entries are called the most and which entry calls take too long. The heap usage rate obtained through JVM monitoring tracking points can be used to show the heap usage. When the heap is almost full, you can also check from the thread record table which entry point call caused the heap to be almost full. This connects the memory situation with the call situation, making it easier to monitor the threads that are processing tasks to avoid the situation where all worker threads are working.

[0041] In this embodiment, there is a counter inside the report trigger and the warning trigger. Each time a heartbeat signal is generated and the trigger condition is met, the corresponding counter will be +1. When the count reaches the upper limit value (that is, the corresponding preset report threshold or preset warning threshold), the count will be reset to 0. When the corresponding counter is 0, the report or warning will be triggered.

[0042] The traditional approach is to detect health conditions and send reports at the same time. For example, if you set the application status to be detected every 60 seconds, the detection and report will be sent every 60 seconds. This solution detects once a second. When the application is in an unhealthy state, a warning report will be sent for the first time (because the cumulative number of triggers reaches 0). This is much more sensitive than detecting every 60 seconds.

[0043] It is worth noting that in step S2, for the status check of the entry monitoring point, each time a heartbeat signal is generated, the entry monitoring point obtains the value of the number of threads inPointCount;

[0044] The initial status value of the entrance monitoring point is the normal status value;

[0045] When inPointCount>A, the status value changes from the normal status value to the warning status value;

[0046] In the case where the state value is the warning state value, when inPointCount>A in B consecutive heartbeat signals, the state value is converted from the warning state value to the warning state value;

[0047] When inPointCount>D in C consecutive heartbeat signals, the status value changes from the warning status value to the forced protection status value;

[0048] When the state value is an early warning state value, a warning state value, or a forced protection state value, and inPointCount≤A in E consecutive heartbeat signals, the state value is converted to a normal state value.

[0049] In this embodiment, A=30, B=10, C=120, D=80, and E=30.

[0050] Preferably, in step S2, when the state value is a forced protection state value, the entry monitoring embedding point calls the thread record table, and adds a current limiting strategy for all entry methods in the thread record table;

[0051] When the status value is converted from the mandatory protection status value to other status values, the current limiting policy added to the entry method is removed.

[0052] In this solution, the current limiting strategy is a prior art technology. For reference, please refer to the Chinese invention patent application number 202211608731.5, "A Current Limiting Method and System for Protecting Applications." In this embodiment, the monitoring center has a built-in message receiving module that receives externally sent current limiting strategies. The monitoring center then calls the entry monitoring tracking point to set the current limiting strategy for the entry method.

[0053] Optionally, in step S2, when the request passes through the aspect, the in method of the entry monitoring point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is increased by 1; in this embodiment, when the request passes through the aspect, it is checked whether the call of the current entry method triggers current limiting; in this embodiment, the triggering strategies of current limiting are divided into always and monitor. When the triggering strategy of current limiting is always, the current limiting strategy is always enabled for the current entry method. When the triggering strategy of current limiting is monitor, current limiting is triggered when the current status value is greater than or equal to the warning status value; if current limiting is triggered, the current limiting strategy is called to determine whether it can pass. If it cannot pass, the request is interrupted and an exception is thrown. If it can pass, the entry method name and entry method parameters of the thread corresponding to the request are obtained through the in method. After obtaining the number of threads inPointCount, the corresponding status value is obtained according to the number of threads inPointCount; the current limiting setting in the current limiter is as follows: {method name} = {always / monitor} | {current limiting strategy};

[0054] When the business processing corresponding to the request is completed and returned to the aspect, the out method of the entry monitoring point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is reduced by 1.

[0055] Aspects are a core concept in aspect-oriented programming (AOP). They separate and centralize scattered, yet often required, cross-cutting concerns (such as logging, security, transactions, and performance monitoring). In other words, an aspect encapsulates a specific type of cross-cutting concern. Therefore, in this example, after adding an aspect, all method calls will first pass through the aspect.

[0056] In other words, an aspect is an encapsulation of a certain type of horizontal concern functionality.

[0057] Specifically, in step S2, the entry monitoring tracking point obtains the entry method name and entry method parameters obtained by the web interceptor or the method name and parameters obtained by the dubbo interceptor through the in(String method,Object args) method, where String method represents the entry method name and Object args represents the entry method parameters;

[0058] Then call the thread record table to record the thread information corresponding to the entry method name and entry method parameters.

[0059] In this embodiment, the current thread information is obtained through the method Thread.currentThread() provided by Java, such as Figure 2 As shown, the thread information includes thread ID, thread name, entry method name, entry method parameters, entry method call time, SQL being executed and SQL start time.

[0060] In this embodiment, the data source tracking point is set to monitor the SQL being executed, the SQL start time and the thread ID; the data source tracking point provides a mybatis plug-in, which can be used directly after configuration. The mybatis plug-in implements the intercept interface method. The Invocation invocation parameter in the intercept interface method can obtain the SQL currently being executed; the data source tracking point provides in(string sql) and out methods. When the request passes through the aspect, the in method of the data source tracking point is called. When the business processing is completed and returns to the aspect, the out method of the data source tracking point is called to complete the monitoring of the data source tracking point; specifically, the in(string sql) method: searches the thread record table according to the current thread id. If a record can be found, the "sql being executed" and "sql start time" are added to the record; the out method: searches the thread record table according to the current thread id. If a record can be found, the "sql being executed" and "sql start time" information on the record are cleared. The data source tracking point is to provide more sufficient information for entry monitoring tracking points.

[0061] It is worth noting that, in step S2, the web interceptor obtains the entry URL path and parameters as the entry method name and entry method parameters through the HttpServletRequest request parameters in the interface method in the general interface of java web.

[0062] Preferably, in step S2, the dubbo interceptor obtains the entry method name and entry method parameters through the Invoker<?>invoker,Invocation invocation parameter in the interface method in the Filter interface.

[0063] Optionally, in step S2, the entry monitoring point calls the thread record table through the out() method (this method has no parameters) to clear the corresponding thread information in the thread record table.

[0064] It is worth noting that in step S2, for the status check of the JVM monitoring tracking point, the initial status value of the JVM monitoring tracking point is a normal status value;

[0065] When the heap usage exceeds G in F consecutive heartbeat signals, the status value changes from normal to warning.

[0066] In the case where the state value is a warning state value, when the heap usage rate in H consecutive heartbeat signals is lower than 1, the state value is converted from the warning state value to the normal state value.

[0067] In this embodiment, F=3, G=98%, H=10, and I=95%.

[0068] Specifically, the JVM monitoring point obtains the current heap size and heap upper limit by calling the Java language tool ManagementFactory.getMemoryMXBean(), and the heap usage = current heap size / heap upper limit.

[0069] In this embodiment, the JVM monitoring point is built into the monitoring center.

[0070] In this solution, three indicators of Java applications are monitored: JVM heap size, entry tracking monitoring, and data source tracking monitoring. Unlike traditional methods, this solution will link the information of these three. When the status value of the monitoring center is above the warning level, the information of these three will be reported together so that users can quickly find the cause of the warning. The information of the JVM monitoring tracking point includes: the maximum heap value and the used heap value; the information of the entry monitoring tracking point includes: the full name of the method, parameters, the request start time, and the name of the thread processing the request; the information of the data source tracking point includes: sql, sql execution start time, and the current thread name; the monitoring center status value has four states: normal, warning, warning, and forced protection. In the normal and warning states, the device only reports simple monitoring information. Starting from the warning state, detailed warning information will be reported. In the forced protection state, all currently accessed entries will be analyzed, and the preset current limiting strategy will be forcibly added to the entry that may cause the cause.

[0071] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. It is apparent to those skilled in the art that various changes, modifications, substitutions, and variations to these embodiments may be made without departing from the principles and spirit of the present invention, and these changes and modifications still fall within the scope of protection of the present invention.

Claims

1. A monitoring and warning method based on Java application, characterized in that: The following steps are involved: S1: After the Java application is started, the monitoring center generates a heartbeat signal at a specified period. S2: Each time a heartbeat signal is generated, the monitoring center sequentially obtains the status values ​​of the entry monitoring point and the JVM monitoring point after status checks; the status values ​​include normal status value, warning status value, warning status value and forced protection status value arranged in ascending order; S3: The status value of the entry monitoring tracking point and the status value of the JVM monitoring tracking point with a higher severity is used as the monitoring center status value corresponding to the current heartbeat signal; S4: If the monitoring center status value corresponding to the current heartbeat signal is less than the warning status value, a report trigger is triggered, and the cumulative number of triggering times of the report trigger is recorded; If the monitoring center status value corresponding to the current heartbeat signal is greater than or equal to the warning status value, the warning trigger is triggered and the cumulative number of triggering of the warning trigger is recorded; When the cumulative number of triggers of the report trigger exceeds the preset report threshold, the cumulative number of triggers of the report trigger is cleared and a brief report is generated and sent to the monitoring center. When the cumulative number of triggers of the warning trigger exceeds the preset warning threshold, the cumulative number of triggers of the warning trigger is cleared and detailed warning information is generated and sent to the monitoring center.

2. The Java-based monitoring and warning method according to claim 1, wherein: In step S2, for the status check of the entry monitoring point, each time a heartbeat signal is generated, the entry monitoring point obtains the value of the thread number inPointCount; The initial status value of the entrance monitoring point is the normal status value; When inPointCount>A, the status value changes from the normal status value to the warning status value; In the case where the state value is the warning state value, when inPointCount>A in B consecutive heartbeat signals, the state value is converted from the warning state value to the warning state value; When inPointCount>D in C consecutive heartbeat signals, the status value changes from the warning status value to the forced protection status value; When the state value is an early warning state value, a warning state value, or a forced protection state value, and inPointCount≤A in E consecutive heartbeat signals, the state value is converted to a normal state value.

3. The Java-based monitoring and warning method according to claim 2, wherein: In step S2, when the state value is a forced protection state value, the entry monitoring point calls the thread record table and adds a current limiting strategy for all entry methods in the thread record table; When the status value is converted from the mandatory protection status value to other status values, the current limiting policy added to the entry method is removed.

4. The Java application-based monitoring and warning method according to claim 3, characterized in that: In step S2, when a request passes through a slice, the in method of the entry monitoring point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is increased by 1; When the business processing corresponding to the request is completed and returned to the aspect, the out method of the entry monitoring point is called to obtain the thread corresponding to the request, and the number of threads inPointCount is reduced by 1.

5. The Java application-based monitoring and warning method according to claim 4, characterized in that: In step S2, the entry monitoring point obtains the entry method name and entry method parameters obtained by the web interceptor or the method name and parameters obtained by the dubbo interceptor through the in(String method,Object args) method, where Stringmethod represents the entry method name and Object args represents the entry method parameters; Then call the thread record table to record the thread information corresponding to the entry method name and entry method parameters.

6. The Java application-based monitoring and warning method according to claim 4, characterized in that: In the step S2, the web interceptor obtains the entry URL path and parameters as the entry method name and entry method parameters through the HttpServletRequestrequest parameters in the interface method in the general interface of the java web.

7. The Java application-based monitoring and warning method according to claim 4, characterized in that: In step S2, the dubbo interceptor obtains the entry method name and entry method parameters through the Invoker<?>invoker,Invocation invocation parameter in the interface method in the Filter interface.

8. The Java application-based monitoring and warning method according to claim 3, characterized in that: In step S2, the entry monitoring point calls the thread record table through the out() method to clear the corresponding thread information in the thread record table.

9. The Java application-based monitoring and warning method according to claim 1, characterized in that: In step S2, the status of the JVM monitoring tracking point is checked, and the initial status value of the JVM monitoring tracking point is a normal status value; When the heap usage exceeds G in F consecutive heartbeat signals, the status value changes from normal to warning. In the case where the state value is a warning state value, when the heap usage rate in H consecutive heartbeat signals is lower than 1, the state value is converted from the warning state value to the normal state value.

10. The Java application-based monitoring and warning method according to claim 9, characterized in that: The JVM monitoring point obtains the current heap size and heap upper limit by calling the Java language tool ManagementFactory.getMemoryMXBean(), and the heap usage = current heap size / heap upper limit.

Images