JP2007249490A - Automatic failure log selection/collection method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic failure log selection and collection method and apparatus for surely collecting log information relating to the factor of a failure by using limited resources. <P>SOLUTION: The automatic failure log selection/collection method comprises: a failure position specifying procedure for specifying a position where a failure module which has been executed at a point of time when the failure is exposed occupies in a configuration table; a selection procedure for estimating a path reaching the failure module as a sequence of nodes on the configuration table by retrieving the configuration table by tracing back to the node corresponding to the failure module, and for selecting the node of the configuration table positioned on the path as the log collection object node; a log collection procedure for collecting the log information relating to the module according to the execution of each module comprising software in executing a reproduction test; a determination procedure for determining whether or not the log information collected by the log collection procedure corresponds to the log collection object node; and a log discard procedure for discarding the log information collected on the path reaching the log information according to the determination result indicating that the collected log information does not correspond to the log collection object node. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a failure log automatic selection collection method and apparatus for collecting log information related to a failure of computer software.

Regardless of the size of the software, in order to resolve a failure that occurred during the execution of the software, it is necessary to trace the process leading to the failure and identify the cause of the failure.
For this purpose, for example, when performing a reproduction test to find the cause of a failure, a history is shown according to the passage of each node of the call tree (or configuration table) indicating the relationship between each module constituting the program. A method of collecting log information is used. Also, when returning from the processing of the called processing program to the processing of the calling source processing program, the error code in the called processing program is notified, so that the highest level can be obtained regardless of the hierarchy to which the program in which the error occurred belongs. A technique has also been proposed in which the error contents can be grasped in the processing program of the above hierarchy (see Patent Document 1).

In addition, in an information processing apparatus such as a channel control apparatus in which a plurality of equivalent hardware devices operate in parallel, only history information related to input / output operations related to arbitrary hardware (for example, a subchannel control device) is obtained. A technique of restricting history information to be analyzed by selectively recording has also been proposed (see Patent Document 2).
Conventionally, such log information is generally collected by circularly writing the collected log information using a buffer having a certain capacity as a ring buffer.

When log information is collected using such a ring buffer, the amount of log information corresponding to the process from the occurrence of the event that caused the software failure until the failure is clearly expressed is the ring buffer. As long as the storage capacity allocated to is not exceeded, the cause of the failure can be identified by tracing the log information sequentially from the occurrence of the failure.
JP-A-2-66644 JP-A-57-189229

  By the way, in a complicated and large-scale software, there may be a huge process from the occurrence of an event causing a failure to the occurrence of a visible failure. For example, in memory access exception handling, a very large number of processes occur after the memory contents that caused the failure have been destroyed, until the failure appears in response to a reference to this memory by another module. May have gone through.

  In such a case, if all the log information is collected comprehensively as each node of the call tree passes, the failure is clearly expressed from the occurrence of the event that caused the software failure. There is a possibility that the amount of log information corresponding to the above processes exceeds the storage capacity allocated to the ring buffer. That is, there is a high possibility that the log information indicating the cause of the failure is overwritten by newly collected log information and the target log information is missed.

If the cause of the failure cannot be identified from the log information, the cause is narrowed down by manually tracing the configuration table that represents the structure of the large-scale software, and a reproduction environment for reproducing the phenomenon is created. Measures such as carrying out reproduction experiments after construction are necessary, and a great deal of labor is required.
However, since resources such as a memory provided in the information processing apparatus are limited, it is clearly not practical to provide a large-capacity buffer that can store all collected log information.

  It is an object of the present invention to provide a failure log automatic selection and collection method and apparatus capable of reliably collecting log information related to the cause of a failure using limited resources.

The failure log automatic selection and collection method according to the present invention includes a failure location specifying procedure, a selection procedure, a log collection procedure, a determination procedure, and a log discard procedure.
The principle of the failure log automatic selection and collection method according to the present invention is as follows.
In the failure log automatic selection and collection method for collecting failure log information indicating the process related to the failure in the reproduction test for identifying the cause of the failure related to the software executed by the information processing system, the failure location specifying procedure is a software execution The configuration table showing the relationship between the plurality of program entries that constitute the software and the plurality of modules that constitute these program entries, according to the indication of the failure related to Specifies the position occupied by. The selection procedure searches the configuration table retroactively from the node corresponding to the failure module, estimates the route to the failure module as a sequence of nodes on the configuration table, and collects the nodes of the configuration table located on the route Select as target node. The log collection procedure collects log information related to a module according to the execution of each module constituting the software during a reproduction test. In the determination procedure, it is determined whether or not the log information collected in the log collection procedure corresponds to a node of the log collection target. The log discarding procedure discards the log information collected for the route leading to the log information according to the determination result that the collected log information does not correspond to the node of the log collection target.

The operation of the fault log automatic selection and collection method configured as described above is as follows.
Depending on the occurrence of a fault during software execution, the fault location procedure and selection procedure will select the nodes that constitute the path to the fault module in the software configuration table as log collection targets in this software configuration table. The
In the reproduction test, each time new log information is collected by the log collecting means, it is determined whether or not the new log information corresponds to a node to be collected by the determination procedure. If it is determined that the node does not correspond to the target node, the log information collected along the corresponding path prior to the new log information is discarded by the log discard procedure.

  In this way, log information related to the flow of processing that has transitioned to a module corresponding to a node that has deviated from the path described above before reaching the module in which the failure has appeared is traced back to, for example, the program entry of the module call source described above. Accordingly, only the log information collected by the log collection procedure is finally left with respect to the flow of processing that has transitioned along the path including the nodes of the log collection target selected by the selection procedure described above.

The first failure log automatic selection / collection apparatus according to the present invention comprises a configuration table, a failure location specifying means, a selection means, a log collection means, a determination means, and a log discarding means.
The principle of the first failure log automatic selection and collection apparatus according to the present invention is as follows.
In a failure log automatic selection and collection apparatus that collects failure log information representing a process related to a failure in a reproduction test for identifying the cause of the failure related to software executed by the information processing system, the configuration table includes a plurality of components constituting the software The program entries and the relationship among the modules constituting these program entries are shown. The failure location specifying means acquires information for specifying the location occupied by the failure module being executed at the time of the occurrence of the failure in the configuration table in response to the occurrence of the failure related to the execution of the software. The selection means searches the configuration table retroactively from the node corresponding to the failure module, estimates the route to the failure module as a series of nodes on the configuration table, and collects the nodes of the configuration table located on the route Select as target node. The selection information holding unit holds selection information indicating whether or not each node constituting the configuration table has been selected as a log collection target by the selection unit. The log collecting means collects log information related to a module according to execution of each module constituting the software and reproduces it in a buffer during a reproduction test. The determination unit determines whether or not the log information collected by the log collection unit corresponds to a node of a log collection target, based on the selection information held in the selection information holding unit. The log discarding unit discards the log information accumulated in the buffer for the route leading to the log information according to the determination result that the collected log information does not correspond to the node of the log collection target.

The operation of the first failure log automatic selection / collection apparatus configured as described above is as follows.
If any failure occurs during the execution of software, the failure location identifying means identifies the position occupied by the failed failure module in the configuration table, and the selection means identifies the log collection target on the path to this failure module. Nodes are selected, and selection information indicating these nodes is held in the selection information holding means.

  Then, every time new log information is collected by the log collection means during the reproduction test, a path from the newly collected log information to the above-described failure module by the determination means based on the selection information described above When it is determined whether or not the log information is related to the upper node, and the log information related to the node deviating from the above-described path is collected, the log discarding unit stores a series of logs stored in the buffer prior to the log information. For example, the log information is discarded retroactively to the node corresponding to the program entry that is the caller of the corresponding module.

Therefore, only log information when the transition is made without following the path leading to the failure module is selectively left in the buffer.
The principle of the second failure log automatic selection / collection apparatus according to the present invention is as follows.
In the first failure log automatic selection and collection apparatus described above, the selection information holding means is configured using a part of the setting register.

The operation of the second failure log automatic selection / collection apparatus configured as described above is as follows.
For example, each bit of the setting register is associated with each node of the configuration table, and the node of the log collection target is indicated by setting the bit of the setting register corresponding to each node according to the selection result by the selection unit. .
A third failure log automatic selection / collection device according to the present invention is the above-described first failure log automatic selection / collection device, comprising error information collection means and type determination means, and element information collection means in the log collection means. It is prepared for.

The principle of the third failure log automatic selection / collection apparatus according to the present invention is as follows.
In the first failure log automatic selection / collection device described above, the error information collection means collects error information related to the failure according to the occurrence of the failure. The type determining unit determines an appropriate log information set from a plurality of types of log information sets having different combinations of element information based on the error information collected by the error information collecting unit, and indicates the determined log information set Is notified to the log collecting means. The element information collecting means collects each element information included in the log information set indicated by the information designated by the notification from the type determining means.

The operation of the third failure log automatic selection / collection apparatus configured as described above is as follows.
Based on the error information collected by the error information collecting means in response to the occurrence of the failure, the type determining means determines a log information set indicating a combination of element information corresponding to the error that has occurred, and notifies the log collecting means. The At least one element information indicated by this log information set is collected as a part of the log information by the corresponding element information collecting means and accumulated in the buffer.

According to the failure log automatic selection and collection method and apparatus according to the present invention, only log information when the process transitions along the path to the module being executed at the time when the failure appears is displayed according to the configuration table. Can be selectively left in the buffer.
Therefore, by assigning a realistic capacity to the log collection buffer, it is possible to reliably collect log information indicating the event that caused the failure, greatly reducing the effort required to investigate the cause of the failure. And measures can be taken quickly.

Further, in the failure log automatic selection / collection device according to the present invention, information indicating the nodes on the configuration table to be collected by the log is held in the setting register, so that a wide range including, for example, a hardware failure of the auxiliary storage device can be obtained. It is possible to deal with the situation and collect the failure log reliably.
Furthermore, by collecting log information consisting of a combination of appropriate element information according to the failure that has occurred, it is possible to selectively collect information useful for failure analysis. The capacity can be used more effectively.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(First embodiment)
FIG. 1 shows a first embodiment of a failure log automatic selection / collection apparatus according to the present invention.
The failure log automatic selection / collection apparatus 210 shown in FIG. 1 is related to a program module related to error notification as described later based on the error notification received from the module processing unit 201 and the module call passed to the module processing unit 201. Collect log information selectively.

  In the failure log automatic selection and collection apparatus 210 shown in FIG. 1, the module identification unit 211 identifies a program module in which an error has occurred based on the error notification from the module processing unit 201, and routes information indicating the program module. The tracking processing unit 212 is notified. In addition, the configuration table holding unit 213 shown in FIG. 1 holds information representing a configuration table (see FIG. 2A) indicating the relationship between the program modules in the program entry.

Based on this configuration table, the route tracking processing unit 212 shown in FIG. 1 tracks a route from which the program module notified from the module specifying unit 211 is called, and selects a flag corresponding to each program module on this route. A selection flag provided in the holding unit 214 is set to indicate that these program modules have been selected for log collection.
At this time, as shown in FIG. 2A, the route tracking processing unit 212 goes back from the program module notified from the module specifying unit 211 (for example, the function B22 indicated by the symbol FB22 in FIG. 2A). The tree structure of the configuration table is traced up to the caller program entry (indicated by the symbol PE in FIG. 2A), and the program module corresponding to the node that has passed at this time (the black circle in FIG. 2A) A selection flag corresponding to FB2, FB) indicated by is set.

  As a result, as shown in FIG. 2B, among the selection flags provided in the selection flag holding unit 214 corresponding to each program module, the selection flag corresponding to the program module located on the above-described path is selected. Is set automatically. By realizing the selection flag holding unit 214 using, for example, a part of the setting register, information indicating the selection flag corresponding to each program module can be protected from various hardware failures. This ensures reliable execution of log collection processing.

In this way, after the selection flag indicating the route for collecting log information is set in the selection flag holding unit 214, the reproduction test for the software to be tested is started.
In response to the module call sent to the module processing unit 201, log information regarding the program module specified by this module call is acquired by the log acquisition unit 215 shown in FIG. Are accumulated in the buffer 218.

  In the log selection / collection unit 216 shown in FIG. 1, the log determination unit 217 indicates that the corresponding program module is a log collection target by the selection flag corresponding to the program module specified by the module call described above. The log information passed from the log acquisition unit 215 is passed to the buffer 218 and accumulated. On the other hand, when the selection flag indicates that the log is not to be collected, the log determination unit 217 instructs the log discard unit 219 to perform log discard processing, which will be described later, and in response to this, the log discard unit 219 A part of the log accumulated in 218 is discarded.

FIG. 3 is a flowchart showing the selective collection operation of log information.
In response to the module call passed to the module processing unit 201 shown in FIG. 1, corresponding log information is acquired by the log acquisition unit 215 (see steps 301 and 302 in FIG. 3) and transferred to the log selection collection unit 216. .
The log determination unit 217 shown in FIG. 1 refers to the selection flag held in the selection flag holding unit 214 corresponding to the program module specified by the module call described above, and the passed log information is the log information collection target. It is determined whether it corresponds to the program module (see step 303 in FIG. 3).

  For example, when the program module corresponding to the function B is designated by the module call, the log determination unit 217 selects the selection flag (FIG. 2B) held in the selection flag holding unit 214 corresponding to the corresponding program module. The log information acquired in step 302 is accumulated in the buffer 218 (step 304).

Thereafter, the program module is executed (step 305), and if an error does not occur (negative determination at step 306), the process returns to step 301 to continue the processing related to the new module call.
For example, when a module call specifying the program module corresponding to the function B2 is input following the module call specifying the program module corresponding to the function B described above, the log acquisition shown in FIG. The log information acquired by the unit 215 is accumulated in the buffer 218 following the log information FB corresponding to the function B, as shown in FIG.

As described above, as long as the program module is called along the route obtained by the route tracking processing unit 212 described above, log information (for example, FB, FB2) acquired by the log acquisition unit 215 is sequentially Accumulated in the buffer 218.
On the other hand, when log information related to the program module corresponding to the function B and the function B2 is accumulated as described above and a module call specifying the program module corresponding to the function B21 is input, Based on the corresponding selection flag, the log determination unit 217 shown in FIG. 1 determines that this log information is not a collection target (No determination at step 303).

  In response to this, the log discarding unit 219 refers to the configuration table holding unit 213, and the program module (for example, the function B21) corresponding to the log information that is determined not to be collected by the log determination unit 217 is located in the configuration table. Log information corresponding to all nodes (for example, FB21 and FB2 shown in FIG. 2 (a)) on the path from the node to be traced up the tree structure to reach the caller program entry. (Step 307), the log information is discarded (see Step 318, FIG. 4A).

  At this time, the log discarding unit 219 refers to the stack that sequentially holds the module calls instead of referring to the configuration table, and calls the program entry of the program module corresponding to the log information that is not to be collected Log information corresponding to these module calls can be discarded in a lump by obtaining the number of module calls issued from and returning the position of the write pointer in the buffer 218 by the number of module calls. .

  In this way, unnecessary log information can be discarded including log information corresponding to the flow of processing that follows a path similar to the path to the program module in which the error appears. When the log collection processing is completed when the determination in step 306 is affirmative, only log information corresponding to the flow of processing following the path to the program module in which the failure appears is stored in the buffer, as shown in FIG. 4B. 218 is left behind.

Therefore, by allocating a realistic capacity to the buffer 218, regardless of the scale and complexity of the software, the process from the occurrence of the event causing the failure to the manifestation of the failure is reliably shown. Log information can be obtained from this buffer 218.
(Second Embodiment)
FIG. 5 shows a second embodiment of the failure log automatic selection / collection apparatus according to the present invention.

5 that are the same as those shown in FIG. 1 are denoted by the same reference numerals as those shown in FIG. 1 and description thereof is omitted.
The error information collection unit 221 illustrated in FIG. 5 collects error information related to the notified error in response to the error notification from the module processing unit 201 and provides it to the processing of the log type determination unit 222.

The log type determination unit 222 prepares a plurality of log types in advance for various combinations of history information to be noticed when analyzing a failure, and based on the error information passed from the error information collection unit 221, The optimum log type is determined from these log types, and a type flag indicating this log type is set in the type flag holding unit 223.
For example, the log type determination unit 222 prepares a log type that focuses on the history of the address of the access destination, a log type that focuses on the history of write data, and a type flag that indicates element information to be collected for each of these log types. The type flag corresponding to the combination closest to the required combination of element information is stored in the type flag holding unit 223 based on the error information.

Further, the log organizing unit 224 illustrated in FIG. 5 extracts and extracts the element information collected by the above-described type flag from the element information included in the log information acquired by the log acquiring unit 215. Log information including only element information is generated and passed to the log selection / collection unit 216.
In this way, by passing log information consisting only of element information required for failure analysis to the log selection collection unit 216, the log information accumulated in the buffer 218 provided in the log selection collection unit 216 is stored. The amount of information can be further compressed.

  As a result, even when performing complicated error analysis in extremely large-scale software, the capacity of the buffer 218 can be effectively used to reliably obtain the necessary log information. It can be greatly improved.

  As described above, according to the failure log automatic selection and collection method and apparatus according to the present invention, it is possible to selectively collect only log information related to the occurrence of a failure. Regardless of this, it is possible to reliably acquire information required for failure analysis, and to support the cooperation for software development and countermeasures against failures.

  Such a feature is extremely useful in various industrial fields that develop large-scale software and maintain such software.

It is a figure which shows 1st Embodiment of the failure log automatic selection collection apparatus concerning this invention. It is explanatory drawing of a selection flag setting process. It is a flowchart showing selection collection operation | movement of log information. It is a figure explaining the selective collection process of log information. It is a figure which shows 2nd Embodiment of the failure log automatic selection collection apparatus concerning this invention.

Explanation of symbols

201 Module processing unit 210 Fault log automatic selection collection device 211 Module identification unit 212 Route tracking processing unit 213 Configuration table holding unit 214 Selection flag holding unit 215 Log acquisition unit 216 Log selection collection unit 217 Log determination unit 218 Buffer 219 Log discarding unit 221 Error information collection unit 222 Log type determination unit 223 Type flag holding unit 224 Log organization unit

Claims (4)

In the failure log automatic selection and collection method for collecting failure log information representing the process related to the failure in the reproduction test for identifying the cause of the failure related to the software executed by the information processing system,
In response to the occurrence of a failure related to the execution of the software, the failure module being executed at the time of the occurrence of the failure has a plurality of program entries constituting the software and a plurality of modules constituting these program entries. A fault location identifying procedure for identifying the position occupied in the configuration table showing the relationship of
By searching the configuration table retroactively from the node corresponding to the failure module, the route to the failure module is estimated as a series of nodes on the configuration table, and the node of the configuration table located on the route is determined. A selection procedure to select as nodes for log collection,
A log collection procedure for collecting log information related to the module according to the execution of each module constituting the software during a reproduction test;
A determination procedure for determining whether log information collected in the log collection procedure corresponds to a node of the log collection target;
A log discarding procedure for discarding log information collected for a route leading to the log information in accordance with a determination result that the collected log information does not correspond to the node of the log collection target. Fault log automatic selection collection method. In a failure log automatic selection and collection device that collects failure log information representing a process related to a failure in a reproduction test for identifying the cause of the failure related to software executed by the information processing system,
A configuration table showing a plurality of program entries constituting the software and a relationship among a plurality of modules constituting the program entries;
A fault location specifying means for acquiring information for specifying a location occupied by the fault module being executed at the time when the fault is expressed in the configuration table in response to an indication of a fault related to execution of the software;
By searching the configuration table retroactively from the node corresponding to the failure module, the route to the failure module is estimated as a series of nodes on the configuration table, and the node of the configuration table located on the route is determined. A selection means for selecting a node for log collection;
Selection information holding means for holding selection information indicating whether or not each node constituting the configuration table is selected as a log collection target by the selection means;
Log reproduction means for collecting log information related to the module in accordance with the execution of each module constituting the software during a reproduction test, and storing the log information in a buffer;
A determination unit that determines whether log information collected by the log collection unit corresponds to a node of the log collection target based on selection information held by the selection information holding unit;
Log discarding means for discarding log information accumulated in the buffer for a route to the log information in response to a determination result that the collected log information does not correspond to the node of the log collection target A failure log automatic selection and collection device characterized by the above. In the failure log automatic selection collection device according to claim 2,
The failure log automatic selection / collection device, wherein the selection information holding means is configured using a part of a setting register. In the failure log automatic selection collection device according to claim 2,
Error information collecting means for collecting error information about the failure according to the occurrence of the failure;
Based on the error information collected by the error information collection means, an appropriate log information set is determined from a plurality of types of log information sets each having a different combination of element information, and information indicating the determined log information set is collected in the log A type determining means for notifying the means,
The log collecting means comprises element information collecting means for collecting each element information included in the log information set indicated by the information designated by the notification from the type determining means, Collection device.

Images