JPH10333944A - Memory dump sample system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the time until computer system restart and to sample memory contents required for the investigation of the cause of a fault without omission by successively dumping and outputting a memory area not dumped or outputted yet after an operating system by the fault is reloaded. SOLUTION: An operating system core part area dump sample means 101 dumps only the memory area where the operating system is placed before reloading the core part of the operating system in the case that a computer system is stopped by the fault. An area allocation time dump sample means 102 outputs the image of the memory area when the memory area is not the dumped and outputted area before using a certain memory area after the operating system is reloaded. A dump non-outputted area dump sample means 103 successively dumps and outputs the memory area not dumped or outputted yet after reloading the operating system.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to the collection of information when a computer system fails, and more particularly to the collection of a memory dump when a computer system having a large-scale memory fails.

[0002]

2. Description of the Related Art Conventionally, there is a memory dump collecting method or a collecting device for collecting information when a computer system fails.

For example, the dump output device disclosed in Japanese Patent Application Laid-Open No. H5-151034 reduces the time required to restart the system by selecting and dumping only a memory image of a system control area necessary for analyzing a failure. It is described that the capacity of the secondary storage device required for the above is reduced.

In the memory dump collection system disclosed in Japanese Patent Application Laid-Open No. 5-53882, only a memory image of a process related to a process that caused a system down and existing in a main storage device at the time of system down is selected. It is described that the time required to restart the system is shortened by dumping the data and the capacity of the secondary storage device required for dumping is reduced.

Further, in the memory dump collection control method of the computer system disclosed in Japanese Patent Application Laid-Open No. 2-186451, information required for failure analysis is selected and output to a dump file to reduce the system restart time. It describes shortening and reducing the capacity of a secondary storage device required for dumping.

Further, the memory dump output method disclosed in Japanese Patent Application Laid-Open No. 3-255555 reduces the memory dump processing time by dumping only the memory area that is referenced and updated in the abnormal termination phase. Is described.

Conventionally, when a failure occurs in a computer system and the system stops, saving a memory image (collecting a memory dump) at the time of failure, which is effective in investigating the cause, is implemented. Collect the entire memory, or select and collect only a specific area as shown in the conventional examples of the above publications, and after the memory dump collection is completed, restart the system. I was going.

[0008]

In the prior art, there is a problem that the system cannot be restarted until the memory dump has been collected, and that the business using the computer system cannot be performed during that time. (Decrease in system operation rate). In a computer system having a large-scale memory, which has increased in recent years, when dumping the entire memory, the time required for collecting the memory dump becomes enormous, and this problem becomes more remarkable. Even when only a specific area is selected and collected by using the technique disclosed in the above-mentioned publication, depending on the state of the failure, there is a possibility that much time is required to collect a memory dump.

In the prior art, when the system is restarted or after the system is restarted, it is possible to load a new program or save other programs without saving the previous memory image. This is because the memory area was updated for the purpose, and it was necessary to complete the memory dump collection before restarting the system. Also, in the technology disclosed in the above publication, if a process or module operating at the time of detection of a failure and an area referred or updated by a process or module related thereto are large, a memory dump is performed. This requires a lot of sampling time, and the system downtime becomes longer.

When a memory dump of only a specific area is collected by using the technique disclosed in the above publication, a memory image of an area necessary for investigating a cause of a failure is leaked from a target, and There was a problem that it might not be possible to conduct an investigation.

In the technique disclosed in the above publication, the process or module that was running when the system failure was detected, and the area referenced or updated by the related process or module, or the system failure was determined. Although only the area determined from the detected location or situation is selected and collected, the cause of the failure is caused by a process or module that is apparently unrelated to the process or module that was operating when the failure was detected. This is because there are many cases. In such a case, the area that has been referenced or updated by the process or module that caused the failure is not targeted for dumping, so that the cause cannot be investigated.

It is an object of the present invention to reduce the time from the stoppage due to a failure in a computer system to the restart of the system, and to collect the contents of a memory area required for investigating the cause of the failure without omission. The purpose of the present invention is to provide a memory dump collection method that can perform the above.

[0013]

According to the memory dump collecting method of the present invention, when the computer system is stopped due to a failure, the memory in which the core of the operating system is located before reloading the core of the operating system. An operating system core area dumping means for dumping only the area, and before using the memory area after the operating system is reloaded, if the memory area is not an area for which dump output has been performed, An area allocation dump collecting means for dumping an image; and a dump non-output area dump collecting means for sequentially dumping a memory area which has not been dumped after the operating system is reloaded. Operating system outage From, characterized in that to shorten the time until the operating restarted.

Further, the memory dump collecting method according to the present invention further comprises a memory dump collecting area management table for recording a dumped state of the memory area in units of pages, wherein the core area dump collecting means comprises: The flag corresponding to the page in the core area of the collection area management table is set to a value other than the reset value, the flags corresponding to all other pages are set to the reset value, the area allocation dump collecting means and the dump non-output area The dump collecting means sets the flag corresponding to the page of the dumped memory area of the memory dump collecting area management table to a value other than a reset value after the dump output, sets the area allocation time dump collecting means and the dump non-output area dump If the flag is a reset value, the corresponding memory area is dumped. Characterized in that it determined not to be a force already.

Further, in the memory dump collecting method according to the present invention, the dump non-output area dump collecting means is executed with the lowest priority.

Further, the memory dump collecting method according to the present invention is characterized in that the area-collecting dump collecting means and the dump non-output area dump collecting means are executed in parallel.

A recording medium according to the present invention is characterized by recording a program for causing a computer to execute the above-described memory dump collecting method.

[Operation] After the system is restarted, when allocating a memory area for loading a new program or for other purposes, if an image of the area to be allocated has not been dumped yet, the time of area allocation is Since the dump is output by the dump collecting means, if the computer system stops due to a failure, only the memory area where the core of the operating system is located is dumped by the OS core area dump collecting means, and then the memory dump is collected. The system can be restarted without waiting for completion.

[0019] Further, since the dump non-output area dump collecting means sequentially collects dumps even in areas where no dump has been output, after the system is restarted, dumps are newly collected in areas which will not be newly used. It becomes possible.

[0020]

Embodiments of the present invention will be described with reference to the drawings.

In this embodiment, as shown in FIG. 1, when the computer system is stopped due to a failure (S110), the contents of the operating system are rewritten by the system loader when the system is restarted, as shown in FIG. OS that collects memory dumps only in the memory area where the core part (memory resident part) is located
When the memory area is allocated for the purpose of loading a new program into the memory after the system is restarted with the core area dump collecting means 101, the area to be allocated is a non-dumped area. It is determined whether or not the area has not been collected. If the area has not been collected, the area allocation-time dump collecting means 102 outputs an image before allocation to a memory dump file. The output area dump collecting means 103 is provided.

Next, the operation of this embodiment will be described with reference to FIG. 2 in addition to FIG.

When the computer system detects the fault,
When the operation is stopped (S110), first, the OS core area dumping means 101 stores the image of the memory area where the core 202 (memory resident) of the operating system is located in the memory dump file 204 (FIG. 2). Output.

Thereafter, the core part of the operating system is stored in a dedicated location 2 on the memory 201 by the system loader.
02, and restart the system (S1
11). After system restart, new program 2
When a memory area is required for loading the 06 or other purposes (S112), the dump collection means 1 at the time of area allocation
In step S113, it is checked whether the image of the memory area 203 to be allocated has already been output to the memory dump file 204 (S113). If not, the memory image is output to the memory dump file 204 (S114). .

After the system is restarted, the dump non-output area dump collecting means 103 is activated, and the images of the memory areas whose images have not been output to the memory dump file 204 are sequentially stored in the memory dump file 204.
(S116, S117). The dump non-output area dump collecting means 103 and the area allocation-time dump collecting means 102 operate in parallel.

As can be seen from the above, when the computer system is stopped due to the failure according to the embodiment of the present invention, the system can be restarted without waiting for the output of all the memory dumps to be completed. Can be shortened. In addition, since the entire memory is dumped, the image of the memory area required for investigating the cause of the failure will not be overlooked.
The likelihood of preventing a recurrence of the disorder increases.

[0027]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described in detail with reference to the drawings.

In this embodiment, the memory dump collection method is based on the core part of the operating system whose contents are rewritten by the system loader when the computer system is stopped due to a failure as shown in FIG. OS core area dump collecting means 101 for collecting a memory dump only in the memory area where the (memory resident unit) is located, and a purpose of loading a new program into the memory after restarting the system. When allocating a memory area, determine whether the area to be allocated is an uncollected area, and if it is an uncollected area, output the image before allocation to a memory dump file. Means 102
And a dump non-output area dump collecting means 10 for sequentially outputting an image of the dump non-collection area to a memory dump file
3

In this embodiment, the OS core area dump collecting means 101 and the area allocation time dump collecting means 1
02 is implemented as a part of the core part of the operating system.
Implement as an independent program. When the dump non-output area dump collecting means 103 is loaded into the memory 201, the area collecting dump collecting means 102 operates.
However, the dump non-output area dump collecting unit 103 does not necessarily have to be implemented as an independent program, and may be included in the OS core unit 202. Also,
The image of each memory area is a memory dump file 204
Is stored in the memory dump collection area management table 301 (FIG. 3) used for checking.

Memory dump collection area management table 301
Has an element (flag) corresponding to the allocation unit (page) of the memory area. Each element (flag) has a value other than the reset value when the image of the corresponding allocation unit of the memory area has been output to the memory dump file 204. value,
When not yet output, the reset value is held.
That is, each element (flag) serves as a memory dump collection completed flag. The memory dump collection area management table 301 stores the OS core area 20 in the memory 201 (FIG. 2).
Resident in 2

Next, the operation of this embodiment will be described.

When the computer system detects a fault,
When the operation is stopped (S110), first, the OS core area dumping means 101 outputs an image of the memory area where the operating system core 202 (memory resident) is located to the memory dump file 204 on the disk. I do. At this time, the operating system core unit 2 of the memory dump collection area management table 301
The value of the element corresponding to the memory area where 02 is placed is set to, for example, 1 and the other elements are set to, for example, 0, and the memory dump collection area management table 205 is temporarily saved on the disk. Core 202 of the operating system
Is stored at the top of the memory 201, and its size is determined by the system loader by the OS core area size storage variable 303.
Is set to The OS core part dump collection means 101 recognizes the size of the area where the core part 202 of the operating system is located by referring to the OS core part area size storage variable.

After that, the core part of the operating system is loaded on the memory by the system loader, and the memory dump collection area management table 205 saved on the disk is read again on the memory 201 to restart the system ( S111). After the system is restarted, when a memory area is required for loading a new program 206 or for another purpose (S112), the area allocation time dump collection unit 102 corresponds to the memory area 203 to be allocated. If the value of the element is 0 with reference to the element of the memory dump collection area management table 301 (S113), it is determined that the image of the area has not yet been output to the memory dump file 204, and the memory image is stored in the memory dump file 204. (S
114). The memory dump file 204 is stored in the memory 201
The image of each memory area is output to a location on the memory dump file 204 where the absolute address on the memory 201 where the area exists and the offset from the beginning of the memory dump file are equal. At the same time, the value of the corresponding element in the memory dump collection area management table 301 is updated to 1.

After the system is restarted, a program in which the dump non-output area dump collecting means 103 is mounted is started. At the time of startup, the CPU allocation priority is set to the lowest, so that the other program operates when the CPU is not using the CPU (idle state), and the influence on the operation of the other program is minimized. To However, the priority may be changed depending on the status of the system. This program uses the memory dump collection area management table 30
Each element of 1 is sequentially checked (S116), and the value is 0
If so, the image of the memory area corresponding to the element is output to the memory dump file 204, and at the same time, the value of the element is updated to 1 (S117).

When this processing is repeated and all the elements of the memory dump collection area management table 301 are not 0, that is, when the dump collection of all the memory areas is completed, this program ends (S118).

[0036]

As described above, according to the present invention, before the image of the allocation unit (page) of each memory area is first updated by the area allocation dump collection means after the system is restarted, the memory of the memory is updated. Since the image of the area is output to the memory dump file, the system can be restarted without waiting for the entire memory dump to be output. It can be shortened.

Furthermore, since dumping of the entire memory is ultimately performed, the image of the memory area required for investigating the cause of the failure does not become uncollected. Since the number of cases in which the failure becomes possible increases, the possibility of preventing the recurrence of the failure increases.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the flow of processing of a memory dump collection method according to the present invention.

FIG. 2 is a diagram showing a positional relationship of related information in a memory dump collection method according to the present invention.

FIG. 3 shows the relationship between the memory dump collection area management table and the memory area allocation unit (page) in the memory dump collection method according to the present invention, the OS core area size storage variable, and the core of the operating system (memory resident section). FIG.

[Explanation of symbols]

101 OS core part area dump collecting means 102 Dump collecting means at the time of area allocation 103 Dump non-output area dump collecting means 201 Memory 202 OS core part 203 Memory area to be allocated 204 Memory dump file saved on disk 205 Memory dump collecting area Management table 206 Program file to be newly loaded 301 Memory dump collection area management table 302 Memory 303 OS core area size storage variable

Claims (5)

[Claims] 1. An operating system core area dumping method for dumping only a memory area where the operating system core is located before reloading the operating system core when the computer system is stopped due to a failure. Means for collecting an area allocation dump for dumping an image of the memory area before using the memory area after the operating system is reloaded, if the memory area is not an area for which dump output has been performed; After the operating system has been reloaded,
A dump non-output area dump collecting means for sequentially dump-outputting memory areas which have not yet been dump output. 2. A memory dump collection area management table for recording a dumped state of the memory area in units of pages, wherein the core area dump collection means is provided in a core area of the memory dump collection area management table. The flag corresponding to a certain page is set to a value other than the reset value, the flags corresponding to all other pages are set to the reset value, and the area allocation time dump collection means and the dump non-output area dump collection means complete dump output. Later, the flag corresponding to the page of the dumped memory area in the memory dump collection area management table is set to a value other than the reset value, and the area allocation time dump collection means and the dump non-output area dump collection means reset the flag. If the value is a value, it is determined that the corresponding memory area has not been dumped. The memory dump collection method according to claim 1. 3. The memory dump collecting method according to claim 1, wherein the dump non-output area dump collecting means is executed with the lowest priority. 4. The memory dump collection method according to claim 1, wherein the area allocation dump collection unit and the dump non-output area dump collection unit are executed in parallel. . 5. A recording medium recording a program for causing a computer to execute the memory dump collection method according to claim 1. Description: