JP2006172100A - High-speed changeover method for operating system and method therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To operate a substitutional second OS (a system 2) immediately when a failure occurs in an OS (a system 1) during operation, and to dump and collect necessary contents among memory contents of the system 1 by the OS of the system 2. <P>SOLUTION: In this OS changeover method, a memory area 701 of the first system used by a program including the first OS and a memory area 702 of the second system used by a program including the second OS are disposed, one OS is changed over to the other OS when the failure occurs in one OS. The memory area is divided into a plurality of memory management units, dump collection necessity 73 of whether a dump is collected or not and memory recovery necessity 74 determining whether to put the memory area under management of the newly changed-over OS or not when the failure occurs are set in each the memory area of the memory management unit, the OS is changed over when the failure occurs, and the area of unnecessity of the dump collection is put under the management of the OS after the changeover in the memory management units to bring the area into a usable state. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention shortens the operation stop time of a computer system (hereinafter referred to as a system) by performing a switching process of an operating system (hereinafter referred to as an OS) when a failure occurs in the computer system and collecting a dump of a memory area after the switching. It is related to the method to do.

  Conventionally, when an OS failure occurs, a memory content dump is collected in order to investigate the cause of the failure, and the system is restarted after completion of the dump collection. For this reason, the time until the completion of dump collection increases in proportion to the amount of memory installed in the system, and the system stop time increases.

As a conventional technique for collecting a memory dump at the time of an OS failure, for example, as described in Patent Document 1, a dump is not collected from all memory areas when an OS failure occurs. A method has been proposed in which only the memory is collected first, the OS is reloaded, and the remaining part is dumped, so that the contents of the memory area can be collected without omission and the time to start up the OS is shortened. ing.
JP-A-10-333944

  As described above, when a failure or the like occurs in the OS running on the computer and the OS is stopped, it is necessary to immediately start the alternative OS and resume the work. However, the memory contents are dumped for investigating the cause of the failure. When collecting, there is a problem that the OS cannot be restarted until the dump collecting process is completed. Moreover, since the time required for dump collection becomes longer in proportion to the area where dump collection is performed, there is a problem that the time until restart becomes longer.

  Further, as described in Patent Document 1 above, dump collection is performed only for a minimum necessary part, for example, the core part of the OS before restarting the OS, and the remaining memory area after the OS is restarted. Although a technique for shortening the time until restart by a technique of collecting a dump has been proposed, the time required for collecting the dump is not eliminated.

In order to solve the above problems, the present invention mainly adopts the following configuration.
A first system memory area used by a program including the first OS and a second memory area used by a program including the second OS are arranged on the memory of the computer system and are operating. An OS switching method for switching to the second OS or the first OS when a failure of the first OS or the second OS occurs.
The memory area is divided into a plurality of memory management units, and for each memory area of the memory management unit, whether or not to collect a dump is determined, and under the management of the OS that is newly switched when the failure occurs Set whether or not to collect memory to decide whether to place it,
The OS is switched when the failure occurs, and the dump collection rejection area in the memory management unit is placed under the management of the OS after switching to be in a usable state.

  According to the present invention, when the operating OS (system 1) cannot continue normal processing due to a failure or the like, the second OS is immediately maintained while retaining the memory contents used by the system 1 OS. (System 2) can be operated, and the memory contents of system 1 can be saved on a disk or the like by the OS of system 2.

  In addition, the memory area that was being used in system 1 is collected in system 2 in order from the area where dump collection is completed, and is used for work in system 2 to collect dumps in a batch (all areas in memory). Compared to the case, the memory can be used efficiently.

  The operating system switching method according to the embodiment of the present invention will be described in detail below with reference to FIGS. First, an overview of an operating system switching method according to an embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a diagram for explaining the outline of the system switching process and dump collection / collection process of the operating system according to the embodiment of the present invention. FIG. 3 is a diagram showing the flow of system switching processing and dump collection / collection processing according to this embodiment.

  In the switching method of the operating system according to the embodiment of the present invention, two types of OS areas of the system 1 and the system 2 are constructed in advance on the memory so that they are switched when a failure occurs, and the system 2 does not compress the memory of the system 1 It is configured to keep the size to the minimum necessary. Here, the system 1 and the system 2 refer to a system such as an OS, an application program under the OS, and an entire program including its data. The operating OS (for example, the system 1) and the alternative OS (for example, the system 2) may be the same OS (the same OS when executing the same business) or different OSs.

  In FIG. 1, the virtual address space 101 of the system 1 maps to the system 1 area 105 and the dump collected / collected area 106 of the physical address space, and the virtual address space 102 of the system 2 maps to the system 2 area 107 of the physical address space. is doing. Here, when the system 1 is switched to the system 2 due to a failure or the like while the system 1 is operating, the work of the system 1 is resumed on the system 2, and the memory dump collection / collection program 104 is operated on the system 2, A dump is collected from the memory of the system 1 to a dump storage area 108 in a storage device such as a disk. The area 106 for which dump collection has been completed is configured to expand the physical address space that can be used by the system 2 by mapping to the virtual space 102 of the system 2 each time.

  Next, a flow of system switching processing (switching from the system 1 to the system 2) according to the present embodiment will be described with reference to FIG. In the system switching process, the address of system 2 is first called, and control is transferred to system 2 (301). At this point, system 1 is stopped and all subsequent processing is performed by system 2. In the system 2, the memory management table 20 (described in detail in FIG. 2 to be described later) is constructed (302), and using this memory management table, the system 1 memory is first collected (303). Here, collecting the memory in the system 2 means moving this memory to the management of the OS of the system 2. Therefore, the process 303 is to place a memory area that does not require dump collection under the management of the OS of the system 2. The memory recovery process is also performed in the memory dump collection / recovery process of the next process 304, but the recovery process by the memory dump process is avoided by first performing the recovery process for the area where dump collection is unnecessary, and the system 2 immediately It becomes possible to use.

  After the memory collection process 303 of the system 1 is completed, a memory dump is collected and collected (304). That is, when it is necessary to dump in the memory management table, a dump is collected, and then this memory area is placed under the management of the OS of the system 2. Finally, system 1 is reloaded from the disk into memory and initialized to a startable state (system 1 becomes usable again by system 1) (305), and switching from system 2 to system 1 is possible The system switching process is completed. In this way, when a failure occurs in the system 2, it is possible to switch to the system 1, and the system switching process is repeatedly performed between the system 1 and the system 2, thereby minimizing the system operation stop time. It becomes possible to suppress.

  In other words, the main gist of the OS switching method according to the embodiment of the present invention is that a second OS (system 2) is loaded in advance in addition to the operating OS (system 1) on the memory of one computer system. The system is switched to the system 2 while maintaining the memory state used in the system 1 when a failure occurs. In system 2, the area that does not require dump collection of the memory contents of system 1 is immediately recovered (placed under the management of the OS of system 2) and made available to system 2, and the area that requires dump collection is dumped. The memory is sequentially collected later.

  The above description is an outline (configuration and processing flow) of the switching method of the operating system according to the embodiment of the present invention. Next, the specific configuration and the specific of the switching method of the operating system according to the present embodiment are described. A processing flow will be described.

"Memory Management Table"
The memory management table 20 is a table in which all physical memory areas are divided into a plurality of areas such as pages and managed, and dump collection / collection processing is processed for each area. Sometimes used.

  As shown in FIG. 2, the memory management table includes an “area number” (201) indicating a memory area, a “dump necessity” (202) indicating whether or not to perform dump collection of the area, and collecting the area. “Recovery Necessity” (203) indicating whether or not the system 2 is to be reused, and “State” indicating the processing state of the area (for example, a state indicating whether or not the usage in the area is determined by the system) (204).

"Memory management table construction process"
The construction process of the memory management table shown in FIG. 2 will be described with reference to FIG. FIG. 4 is a diagram showing the flow of the construction process of the memory management table used in this embodiment. For each area of the physical memory, the process (process 402 to 406) surrounded by process 401 (acquisition of the first area) and process 407 (whether there is a next area) is performed.

  First, in the determination 402 whether or not memory dump collection is necessary, it is determined whether or not dump collection of the area is unnecessary. For the area determined to require dump collection, “necessary” is displayed in the “dump necessity / unnecessary” (202) column. If it is determined that dump collection is not necessary, “No” is set in the “Dump Necessity” (202) column. Examples of areas that do not require dump collection include areas mapped to VRAM, etc., cache areas, and areas that are not writable by hardware. Increase the area that does not require dump collection according to the dump collection policy. It is also possible. For example, partial dump collection is possible by analyzing the memory area of the system 1 such as “dump collection of OS area only” and “OS area and process area that was operating at the time of failure”. Here, the method for determining the dump collection range is not mentioned.

  Next, in the memory recovery necessity determination 403, it is determined whether or not the area needs to be recovered and used in the system 2. ”Is set to“ No ”in the“ Memory recovery necessity ”(203) column when the collection is not required. The areas determined not to be collected include the reserved area of the system 1 OS and the area of the system 2 OS.

  In the case of “dump collection unnecessary” and “memory recovery unnecessary” (404), since it is not necessary to process the area, “status” (204) is set to “completed” (405). In other cases (when processing is required), “not yet” is set in “state” (204) (406). The memory management table 20 is constructed by repeating the above processing (402 to 406) for all the memory areas.

"Memory recovery"
The memory collection process will be described with reference to FIG. FIG. 5 is a diagram showing the flow of memory recovery processing according to this embodiment. For each area of the physical memory, processing (502 to 506) surrounded by processing 501 and processing 507 is performed. The memory collection process shown in FIG. 5 shows a process based on a memory management table that indicates whether or not dump is necessary and whether or not collection is necessary.

  First, in the process 502, the value of the “status” (204) in the memory management table 20 is read. If it is “completed”, the process proceeds to the end of the process repetition 507 and the process in that area is terminated. If the value is “not yet”, the process proceeds to the next recovery necessity determination 503. In the process 503, the value of “recovery necessity” (203) in the memory management table 20 is read. If “no”, the process proceeds to the end of the process repetition 507 and the process in that area is terminated. If the value of 203) is “necessary”, the process proceeds to the next dump collection necessity determination 504.

  In the process 504, the value of “dump necessity / unnecessary” (202) in the memory management table 20 is read. If “necessary”, the process proceeds to the end of the process repetition 507, the process for the area is terminated, In this case, the process proceeds to the next memory collection 505. In the process 505, the area is collected by the system 2, registered as a physical memory that can be used by the system 2, and the process proceeds to 506. The area collected by the system 2 is registered in the OS physical memory table so that it can be mapped to the virtual address space of the system 2, and when the OS or application issues a memory allocation request Then, virtual-physical mapping is performed by registering the collected memory area in the address conversion table. The address conversion table refers to a table used when the CPU converts a virtual address to a physical address.

  Finally, in process 506, the “status” (204) of the memory management table 20 is set to “complete”. Memory recovery is performed by repeating the above processing (502 to 506) for all memory areas.

"Dump collection and collection"
The dump collection / collection processing will be described with reference to FIG. FIG. 6 is a diagram showing a flow of dump collection / memory recovery processing according to the present embodiment. For each area of the physical memory, processing (602 to 607) surrounded by processing 601 and processing 608 is performed. The dump collection / collection process shown in FIG. 6 shows a process based on a memory management table that requires dumping.

  First, in the process 602, the value of the “status” (204) in the memory management table 20 is read. If it is “completed”, the process proceeds to the end of the process repetition 608, the process of the area is terminated, and the “status” (204) ) Is “not yet”, the process proceeds to the next dump collection necessity determination 603. In the process 603, the value of “dump necessity / unnecessary” (202) in the memory management table 20 is read. When the value of “Dump Necessity” (202) is “Necessary”, the dump collection of the area is performed at the dump collection 604, and then the process proceeds to the memory collection necessity determination 605.

  In the process 605, the value of “recovery necessary” (203) in the memory management table 20 is read. If “No”, the process proceeds to the “status” process 607 in the memory management table 20. When the value of “requirement required” (203) is “necessary”, the next memory recovery 606 recovers the area in the system 2 (puts under the management of the system 2), and the “status” of the memory management table 20 Go to process 607. In process 607, the “status” (204) of the memory management table 20 is set to “completed”. The dump collection / collection is performed by repeating the above processing (602 to 507) for all the memory areas.

  Next, the characteristics of the present embodiment will be described with reference to FIG. 7 illustrating the procedure of the system switching process and dump collection / collection process of the operating system according to the present embodiment. In (1) of FIG. 7, the memory address space is allocated to the system 1 area and the system 2 area, and the first OS and the second OS are loaded respectively (appropriate application programs are installed as necessary). The system 1 is up and running. In (2) and (3) of FIG. 7, when a failure occurs in the first OS of the system 1, the system is immediately switched to the system 2 and the second OS is activated. (4) indicates the management status (dump necessity / rejection necessity / recovery necessity / status) of each area managed by the memory management table shown in FIG. (5) indicates that the area that does not require dump collection is collected by the system 2 (under management).

  In FIG. 7 (6), in principle, dumping is performed on the area where dumping is necessary or not in the dump storage area of the storage device. The memory management table is not placed under the management of the system 2 (the reserved area is not collected by the system 2). The dumped area is collected by the system 2 and is made usable by the system 2. In (7) and (8), dump collection areas 761 and 762 are collected in the dump storage area to complete dump collection, and these areas are subsequently collected by the system 2. In (9), all the areas collected by the system 2 become the system 2 area. The system 2 is operated in the system 2 area, and a program such as the first OS stored in the storage device is loaded into the system 1 area. Or another third OS may be loaded.

  As described above, the OS switching method according to the embodiment of the present invention has the following features. That is, two types of OS areas, system 1 and system 2, are built in the memory in advance, and the system is switched when the OS fails. At this time, the system 2 keeps the memory of the system 1 to a minimum size so as not to compress the memory. By immediately switching from system 1 to system 2 when a failure occurs, the system operation stop time can be minimized (compared to the conventional technology in which an alternative OS is started after executing dump processing). In the system 2, the memory contents used in the system 1 are dumped together with the execution of the business, and a memory area is added for the system 2 in order from the area where the dump collection is completed, and used in the business of the system 2. When the dump collection processing is completed, the next system 1 is prepared for startup in the event of a system 2 failure (by loading a normal OS stored on a disk or the like into the reserved area of system 1), and the system 2 fails When this occurs, the system is switched to the system 1 again. In this way, the operating rate of the computer system can be improved by alternately operating the system 1 and the system 2 when any system fails. When the system 1 is switched to the system 2, a third OS is loaded from a recording medium such as a separate disk in place of the first OS that was previously operating in the memory area of the system 1, The third OS can be operated by the next system switching. Furthermore, in addition to the system 1 and the system 2, it is also possible to newly prepare the system 3 and provide each system area so that each system can be used properly.

It is a figure explaining the outline | summary of the system switching process and dump collection | collection / collection | recovery process of the operating system which concerns on embodiment of this invention. It is a figure which shows the structure of the memory management table used by this embodiment. It is a figure which shows the flow of the system switching process regarding this embodiment, and a dump collection | collection / collection | recovery process. It is a figure which shows the flow of the construction process of the memory management table used by this embodiment. It is a figure which shows the flow of the memory collection | recovery process regarding this embodiment. It is a figure which shows the flow of the dump collection and memory collection | recovery process regarding this embodiment. It is explanatory drawing illustrating the procedure of the system switching process and dump collection | collection / collection | recovery process of the operating system which concerns on this embodiment.

Explanation of symbols

101 Virtual address space of system 1 102 Virtual address space of system 2 103 Memory management table 104 Dump collection / memory collection program 105 Physical address area used by system 1 106 Dump collection / collected area 107 Physical address area used by system 2 108 dump storage area 20 memory management table 201 area number 202 dump necessity 203 collection necessity 204 status 301 to 305 system switching processing steps 401 to 407 memory management table construction processing steps 501 to 507 memory recovery processing processing steps 601 608 Steps for dump collection / collection processing 701 to 781 Procedures for system switching processing and dump collection / collection processing

Claims (6)

A first system memory area used by a program including the first OS and a second memory area used by a program including the second OS are arranged on the memory of the computer system and are operating. An OS switching method for switching to the second OS or the first OS when a failure of the first OS or the second OS occurs.
Dividing the memory area into a plurality of memory management units;
Whether to collect dumps for each memory area of the memory management unit, whether to collect dumps, and whether to collect memory that determines whether to place under the management of an OS that is newly switched when the failure occurs Set,
The OS switching method, wherein the OS is switched when the failure occurs, and the dump collection rejection area is placed under the management of the OS after switching in the memory management unit. In claim 1,
An OS switching method characterized in that the memory area for dump collection is dumped into a dump storage area of a storage device and then placed under the management of the OS after switching to be usable. In claim 2,
Selecting whether or not to place the memory area for dump collection under the management of the OS after switching based on the necessity of memory collection every time the dump is collected in the memory management unit. OS switching method. In claim 1,
When a failure occurs in the first operating OS, the second OS is switched to, and the first OS without a failure is reloaded in the memory area used by the first OS,
An OS switching method characterized in that the first and second OSs are repeatedly used when any OS failure occurs. In claim 1,
When a failure occurs in the first operating OS, the second OS is switched to, and a memory region used by the first OS is loaded with a third OS that has no failure,
An OS switching method characterized by switching to the third OS when a failure occurs in the second OS. A first system memory area used by a program including a first OS and a second memory area used by a program including a second OS are arranged on a memory of the computer system, and the first OS And an OS switching method for switching the second OS when a failure occurs,
The memory area is divided into a plurality of memory management units,
For each memory area of the memory management unit, whether or not to collect a dump, whether to collect a dump, and whether or not to collect a memory that determines whether to place under the management of an OS that is newly switched when the failure occurs Set,
Switch to the alternative second OS when a failure occurs in the first operating OS;
Place the dump collection rejection area in the memory management unit under the management of the second OS and make it usable.
A memory area that is essential for dump collection in the memory management unit is dumped in a dump storage area of a storage device and then placed under the management of the second OS to be usable.
An OS switching method, comprising: reloading a first OS that has no failure in a memory area used by the first OS.

Images