JP4822024B2 - Fault-tolerant server, full backup method, and full backup program - Google Patents

Description

  The present invention relates to a fault tolerant server, a full backup method, and a full backup program.

  In recent years, computer systems have come to be used in various fields, so that even when a failure occurs, the computer system is required to continue processing that has been performed so far.

  As a technology that meets this demand, a fault tolerant (FT) technology for duplicating a system has been attracting attention.

  A fault tolerant server (FT server) using FT technology can operate a duplicated system synchronously, and can continue processing in the other system even if a failure occurs in one system.

  As a method for operating the duplicated systems in synchronization, the CPU (Central Processing Unit) and the CPU sub-system with the memory that each system has have the same operation with the clocks synchronized. It has been known.

  As a prior art of an FT server that employs this lockstep method, Patent Document 1 describes a technology related to an FT controller that operates a duplexed system in lockstep synchronization by the lockstep method.

In the technique described in Patent Document 1, the FT controller operates the CPU subsystem of the duplex system with lock step synchronization, and stops the lock step synchronization when a mismatch occurs between the CPU subsystems due to a failure or the like. Then, the other system is disconnected from the own system, and the process that was performed before the lock step synchronization was stopped is continued in the own CU subsystem. In addition, after the other system recovers from the failure, the FT controller causes the other CPU subsystem to resume lock step synchronization with the own CPU subsystem.
JP 2006-178616 A

  As described above, in the FT server, fault tolerance is improved by duplicating the system. However, in the event of a failure such as a failure in both systems duplicated due to a fire, backup of information including the operating system (OS: Operating System) stored in the hard disk (HDD: Hard Disc Drive) (Full backup) must be created.

  In general, in order to create a full backup, the system needs to stop all processing in order to prevent any changes such as writing from being performed on information stored in the HDD.

  However, since the FT server is required to always operate and continue processing, a general full backup method on which the system is supposed to stop processing may create a full backup. Can not.

  If the technique described in Patent Document 1 is used, it is possible to continue the processing that has been performed in one system of the duplexed system and to stop the separated system. However, with this technology, the stopped system cannot be restarted independently, so to create a full backup, a full backup of the information stored in the other system stopped by the external system Need to be created.

  Therefore, in the FT server, there is a problem of creating a full backup without using an external system by independently restarting one of the duplexed systems.

  An object of the present invention is to provide a fault tolerant server, a full backup method, and a full backup program that can solve the above-described problems.

In order to achieve the above object, the fault tolerant server of the present invention provides:
A system including a CPU subsystem is configured to be duplicated, and the CPU subsystem is a fault tolerant server that performs the same operation in lockstep synchronization,
Each of the systems
At the time of full backup, the local CPU subsystem stops the lock step synchronization and then stops the local system. When the other system stops, an FT that sends a start signal to the other system A controller,
When the CPU subsystem receives the activation signal from the other system, the CPU subsystem executes the backup operating system stored at the first specified location on the LAN, and is stored in the local system. A full backup of the stored information is stored in a second prescribed location on the LAN.

In order to achieve the above object, the full backup method of the present invention includes:
A full backup method performed by a fault tolerant server in which a system including a CPU subsystem is configured in a duplex manner and the CPU subsystem performs the same operation in lockstep synchronization,
If one system configured for duplexing creates a full backup,
The one system causing the CPU subsystem of the one system to stop the lockstep synchronization and then stopping the one system;
The other system sends a start signal to the one system when the one system stops;
When the one system receives the activation signal from the other system, it is stored in the one system by executing the backup operating system stored in the first specified location on the LAN. Storing a full backup of the stored information in a second prescribed location on the LAN.

In order to achieve the above object, the full backup program of the present invention
In a fault tolerant server in which a system including a CPU subsystem is configured to be duplicated, and the CPU subsystem performs the same operation in lockstep synchronization,
If one system configured for duplexing creates a full backup,
The one system causing the CPU subsystem of the one system to stop the lockstep synchronization and then stopping the one system;
A procedure in which the other system sends an activation signal to the one system when the one system stops;
When the one system receives the activation signal from the other system, it is stored in the one system by executing the backup operating system stored in the first specified location on the LAN. Storing a full backup of the stored information in a second prescribed location on the LAN.

  According to the fault-tolerant server of the present invention, when one system stops to create a full backup, the other system sends a start signal to the one stopped system. In addition, when one of the stopped systems receives the start signal, the full backup is stored on the LAN by executing the backup operating system.

  As a result, one of the duplexed systems can be restarted independently, so that an effect that a full backup can be created without using an external system is obtained.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 shows the configuration of an FT server according to an embodiment of the present invention.

As shown in FIG. 1, the FT server 10 of this embodiment has a configuration in which a 0-system 100 ₀ and a 1-system 100 ₁ are duplicated. The 0-system 100 ₀ and the 1-system 100 ₁ are each connected to a LAN (Local Area Network) 300, and a network HDD 200 exists on the LAN 300.

The system _{0 0} system 100 ₀ has a CPU subsystem 110 ₀ and an FT controller 130 ₀ .

Similarly, the 1-system 100 ₁ includes a CPU subsystem 110 ₁ and an FT controller 130 ₁ .

Here, the operation of the components in the 0-system 100 ₀ and the 1-system 100 ₁ will be described. In the following description, the components in the _0- system 100 ₁ are described, but the components in the 1-system 100 ₁ perform the same operation.

When the CPU subsystem 110 ₀ receives the activation signal from the FT controller 130 ₁ of the other system, the CPU subsystem 110 ₀ executes the backup OS stored in the network HDD 200 on the LAN 300 and is stored in the local system 100 _0. A full backup of the stored information is stored in the network HDD 200 on the LAN 300.

  Here, it is assumed that the backup OS is stored in the network HDD 200. However, it is only necessary that the backup OS is stored in a specified position (first specified position) on the LAN 300. In this example, the full backup is stored in the network HDD 200. However, the full backup may be stored in a specified position (second specified position) on the LAN 300.

Further, when the CPU subsystem 110 ₀ stores the full backup in the network HDD 200 on the LAN 300, the CPU subsystem 110 ₀ notifies the other system 100 ₁ to that effect.

The FT controller 130 ₀ stops the lock system synchronization (hereinafter simply referred to as “synchronization”) in the own CPU subsystem 110 ₀ and then stops the own system 100 ₀ . Even after the synchronization stopped, CPU subsystem 110 ₁ of the other system is assumed to continue processing which has been performed before stopping the synchronization.

Further, when the other system 100 ₁ stops, the FT controller 130 ₀ transmits an activation signal to the other system 100 ₁ .

When the FT controller 130 ₀ receives a notification from the other CPU subsystem 110 ₁ that the full backup has been stored, the FT controller 130 ₀ notifies the other CPU subsystem 110 ₁ to the own CPU subsystem 110 ₀ . Resume synchronization.

Further, when the FT controller 130 ₀ receives the activation signal from the other system 100 ₁ , the FT controller 130 ₀ determines whether or not the other CPU subsystem 110 ₁ is continuing the processing that was being performed before stopping the synchronization. check shifts the system 100 ₀ of its own system if you continue processing mode without device (not shown) defined. Here, the prescribed device refers to a device that is connected to the FT server 10 only one by one, and corresponds to a display, a keyboard, a mouse, and the like.

  Next, the overall operation when the FT server 10 of the present embodiment creates a full backup will be described with reference to the flowchart of FIG.

In this case, it is assumed that the system ₀ 100 continues processing, and the system ₁ 100 ₁ creates a full backup.

As shown in FIG. 2, during normal operation, 0-based CPU subsystem 110 ₀ the CPU subsystem 110 ₁ of 1 system operates synchronously (step 1000 and step 2000).

In this state, the FT controller 130 ₁ stops the synchronization of the own system CPU subsystem 110 ₀ and notifies the other system 100 ₀ that a full backup will be created (step 2010).

Next, the FT controller 130 ₁ stops the own system 100 ₁ (step 2020).

Next, the FT controller 130 ₀ receives a notification from the other FT controller 130 ₁ to create a full backup (step 1010). As a result, the system 100 ₀ detects that the synchronization with the other system 100 ₁ is stopped in order to create a full backup.

Next, FT controller 130 _0, system 100 ₁ of the other system confirms whether or not stopped (step 1020), after the system 100 ₁ of the other system has stopped, a start signal to the system 100 ₁ of the other system Transmit (step 1030).

Next, when the FT controller 130 ₁ receives the activation signal from the FT controller 130 _{0 of} the other system (step 2030), the FT controller 130 ₁ continues the processing performed before the CPU subsystem 110 ₀ of the other system stops synchronization. It is confirmed whether or not (step 2040).

FT controller 130 _1, when the CPU subsystem 110 ₀ of the other system continues the processing shifts the system 100 ₁ of its own system to a mode that does not use the device of the specified (step 2050).

When the CPU subsystem 110 ₁ receives the activation signal from the FT controller 130 ₀ of the other system, the CPU subsystem 110 ₁ executes the backup OS stored in the network HDD 200 on the LAN 300 (step 2060), thereby A full backup of the information stored in 100 ₁ is stored in the network HDD 200 on the LAN 300 (step 2070).

Next, after storing the full backup, the CPU subsystem 110 ₁ notifies the FT controller 130 ₀ to that effect (step 2080).

Thereafter, the FT controller 130 ₀ receives a notification that the full backup has been stored from the other system CPU subsystem 110 ₁ (step 1040), and the other system CPU subsystem 110 ₁ receives the own system CPU subsystem 110. Synchronization with ₀ is resumed (step 1050).

As a result, the 0-system CPU subsystem 110 ₀ and the 1-system CPU subsystem 110 ₁ have returned to their original operating states (steps 1200 and 2200).

On the other hand, FT controller 130 _1, when the CPU subsystem 110 ₀ of the other system does not continue the process has been performed before stopping the synchronization, to the mode not using the self-system device of the system 100 ₁ of the provisions of Do not migrate.

At this time, the CPU subsystem 110 ₁ executes the backup OS stored in the network HDD 200 (step 2090), thereby performing a full backup of the information stored in the local system 100 _1. (Step 2100), and the operation is terminated.

As described above, in the FT server 10 of the present embodiment, when the system 100 ₁ stops to create a full backup, the system 100 ₀ transmits an activation signal to the stopped system 100 ₁ . When the stopped system 100 ₁ receives the activation signal, the system 100 ₁ stores the full backup on the LAN 300 by executing the backup OS.

  As a result, one of the duplexed systems can be restarted independently, so that an effect that a full backup can be created without using an external system is obtained.

  Next, the present invention will be described in more detail using specific examples.

  This embodiment is a specific implementation of the FT server 10 of the embodiment shown in FIG.

  The configuration of the FT server 10 of this embodiment is shown in FIG.

As shown in FIG. 3, the FT server 10 of this embodiment includes a _0- system 100 including a CPU subsystem 110 ₀ , an IO (Input / Output) subsystem 120 ₀ , an FT controller 130 _0, and a cross link 140 _{0. 0} , a CPU subsystem 110 ₁ , an IO subsystem 120 ₁ , a 1-system system 100 _{1 including an} FT controller 130 ₁ and a cross link 140 ₁ are duplicated.

The 0-system CPU subsystem 110 ₀ includes a CPU 111 ₀ , a memory 112 _0, and a memory controller 113 ₀ that manages them. Furthermore, IO subsystem 120 ₀ 0 system includes a LAN controller 121 _0, and a HDD 123 _0, a SCSI (Small Computer System Interface) controller 122 ₀ to manage HDD 123 _0, the.

Similarly, the 1-system CPU subsystem 110 ₁ includes a CPU 111 ₁ , a memory 112 _1, and a memory controller 113 ₁ . The 1-system IO subsystem 120 ₁ includes a LAN controller 121 ₁ , an HDD 123 _1, and a SCSI controller 122 ₁ .

First, in the FT server 10 of the present embodiment, the operation of the component when the 0-system 100 ₀ and the 1-system 100 ₁ operate in synchronization will be described. In the following description, the components in the _0- system 100 ₁ are described, but the components in the 1-system 100 ₁ perform the same operation.

The FT controller 130 ₀ causes the other CPU subsystem 110 ₁ to start synchronization with the own CPU subsystem 110 ₀ via the cross link 140 ₀ .

CPU subsystem 110 ₀ which operates synchronously with the CPU subsystem 110 ₁ of the other system is in the memory 112 ₀ of its own system to the OS for controlling the FT server 10 from HDD 123 ₀ of the IO subsystem 120 ₀ of its own system read allowed to execute an OS that has read into CPU 111 ₀ of its own system.

The CPU 111 _{0 that} has executed the OS creates a virtual logical device in order to perform input / output from the FT server 10, and passes through this virtual logical device to the local IO subsystem 120 ₀ and the other IO subsystem 120 ₁ . to access.

  Next, the overall operation when the FT server 10 of this embodiment creates a full backup will be described with reference to the flowchart of FIG. 2 and FIGS.

As shown in FIG. 2, during normal operation, 0-based CPU subsystem 110 ₀ the CPU subsystem 110 ₁ of 1 system operates synchronously (step 1000 and step 2000: see FIG. 4).

In this state, the FT controller 130 ₁ stops the synchronization of the own CPU subsystem 110 ₁ , and notifies the other system 100 ₀ via the own system cross link 140 ₁ that a full backup will be created. (Step 2010: See FIG. 5).

Next, the FT controller 130 ₀ stops the own system 100 ₁ (step 2020).

Next, when the FT controller 130 ₀ receives a notification to create a full backup from the other FT controller 130 ₁ (step 1010), the other system 100 ₁ receives the notification via the own system cross link 140 _0. It is confirmed whether or not it has been stopped (step 1020).

FT controller 130 _0, after system 100 ₁ of the other system is stopped, using a LAN controller 121 ₀ of its own system, via LAN 300, it transmits an activation signal to the system 100 ₁ of the other system (step 1030: 6 reference).

Next, when the FT controller 130 ₁ receives the activation signal from the FT controller 130 ₀ of the other system (step 2030), the CPU subsystem 110 ₀ of the other system synchronizes via the cross link 140 ₁ of the own system. It is confirmed whether or not the processing performed before the stop is continued (step 2040).

FT controller 130 _1, when the CPU subsystem 110 ₀ of the other system continues the processing shifts the system 100 ₁ of its own system to a mode that does not use the specified device (step 2050).

Next, when receiving the activation signal, the CPU subsystem 110 ₁ uses the local LAN controller 121 ₁ to read the backup OS stored in the network HDD 200 on the LAN 300 into the local memory 112 ₁ , to execute the OS backup that was read into the CPU 111 ₁ of its own system (step 2060: see FIG. 7).

The CPU 111 _{1 that} has executed the backup OS uses the local LAN controller 121 ₁ to store a full backup of information stored in the local HDD 123 ₁ in the network HDD 200 (step 2070: see FIG. 8).

Next, the CPU 111 ₁ that executed the backup OS notifies the other system 100 ₀ of this via the LAN 300 using the local LAN controller 121 ₁ (step 2080: see FIG. 9).

Thereafter, the FT controller 130 ₀ receives a notification that the full backup has been stored from the other system CPU 111 ₁ (step 1040), and notifies the other system CPU subsystem 110 ₁ to the own system CPU subsystem 110 ₀ . Synchronization is resumed (step 1050).

As a result, the 0-system CPU subsystem 110 ₀ and the 1-system CPU subsystem 110 ₁ have returned to their original operating states (steps 1200 and 2200: see FIG. 3).

  The FT server 10 according to the above-described embodiment is an example, and the configuration and operation thereof can be changed as appropriate without departing from the spirit of the invention.

For example, in the above-described embodiment, the FT server 10 is configured by two independent systems, the 0-system 100 ₀ and the 1-system 100 ₁ , but the FT server 10 includes two or more independent systems. You may comprise by a system.

  In the embodiment described above, the location where the backup OS is stored and the location where the full backup is created are both the network HDD 200 on the LAN 300, but may be different locations.

  The FT server 10 of the present invention records a program for realizing its function on a recording medium readable by the FT server 10 in addition to the one realized by dedicated hardware as described above. The program recorded on this recording medium may be read by the FT server 10 and executed. The recording medium readable by the FT server 10 refers to a recording medium such as a floppy disk, a magneto-optical disk, or a CD-ROM, and a storage device such as a hard disk device built in the FT server 10. Furthermore, the recording medium that can be read by the FT server 10 is dynamically programmed for a certain period of time like a volatile memory inside the FT server 10 for a short time, as in the case of transmitting a program via the Internet. Includes what is being held.

It is a block diagram which shows the structure of the fault tolerant server of one Embodiment of this invention. It is a flowchart explaining the operation | movement of full backup creation in the fault tolerant server shown in FIG. It is a block diagram which shows the structure of the specific Example of the fault tolerant server of this invention. It is a figure explaining the process of step 1000 and step 2000 shown in FIG. It is a figure explaining the process of step 2010 shown in FIG. It is a figure explaining the process of step 1030 shown in FIG. It is a figure explaining the process of step 2060 shown in FIG. It is a figure explaining the process of step 2070 shown in FIG. It is a figure explaining the process of step 2080 shown in FIG.

Explanation of symbols

10 FT server 100 ₀ , 100 ₁ system 110 ₀ , 110 ₁ CPU subsystem 111 ₀ , 111 ₁ CPU
112 ₀ , 112 ₁ memory 113 ₀ , 113 ₁ memory controller 120 ₀ , 120 ₁ IO subsystem 121 ₀ , 121 ₁ LAN controller 122 ₀ , 122 ₁ SCSI controller 123 ₀ , 123 ₁ HDD
130 ₀ , 130 ₁ FT controller 140 ₀ , 140 ₁ Cross link 200 Network HDD
300 LAN

Claims (9)

A system including a CPU subsystem is configured to be duplicated, and the CPU subsystem is a fault tolerant server that performs the same operation in lockstep synchronization,
Each of the systems
At the time of full backup, the local CPU subsystem stops the lock step synchronization and then stops the local system. When the other system stops, an FT that sends a start signal to the other system A controller,
When the CPU subsystem receives the activation signal from the other system, the CPU subsystem executes the backup operating system stored at the first specified location on the LAN, and is stored in the local system. A fault tolerant server characterized in that a full backup of stored information is stored in a second prescribed location on the LAN. When the CPU subsystem stores the full backup on the LAN, it notifies the other system to that effect,
2. The fault tolerant server according to claim 1, wherein the FT controller causes the CPU subsystem of another system to resume the lock step synchronization with the CPU subsystem of the own system when receiving the notification from the system of another system.   When the FT controller receives the activation signal from the other system and continues the processing performed before the lock subsystem synchronization is stopped by the other CPU subsystem, the FT controller The fault-tolerant server according to claim 1, wherein the fault-tolerant server is shifted to a mode in which a specified device is not used. A full backup method performed by a fault tolerant server in which a system including a CPU subsystem is configured in a duplex manner and the CPU subsystem performs the same operation in lockstep synchronization,
If one system configured for duplexing creates a full backup,
The one system causing the CPU subsystem of the one system to stop the lockstep synchronization and then stopping the one system;
The other system sends a start signal to the one system when the one system stops;
When the one system receives the activation signal from the other system, it is stored in the one system by executing the backup operating system stored in the first specified location on the LAN. Storing a full backup of information stored in a second prescribed location on the LAN. When the one system stores the full backup on the LAN, notifying the other system to that effect;
When the other system receives the notification from the one system, further causing the CPU subsystem of the one system to resume lock step synchronization with the CPU subsystem of the other system; The full backup method according to claim 4.   When the one system receives the activation signal from the other system and the other CPU subsystem continues the processing performed before stopping the lockstep synchronization, the one system The full backup method according to claim 4, further comprising a step of transitioning to a mode in which the specified device is not used. In a fault tolerant server in which a system including a CPU subsystem is configured to be duplicated, and the CPU subsystem performs the same operation in lockstep synchronization,
If one system configured for duplexing creates a full backup,
The one system causing the CPU subsystem of the one system to stop the lockstep synchronization and then stopping the one system;
A procedure in which the other system sends an activation signal to the one system when the one system stops;
When the one system receives the activation signal from the other system, it is stored in the one system by executing the backup operating system stored in the first specified location on the LAN. And a procedure for storing a full backup of existing information in a second prescribed location on the LAN. In the fault tolerant server,
When the one system stores the full backup on the LAN, a procedure for notifying the other system to that effect;
When the other system receives the notification from the one system, the CPU subsystem of the one system further executes a procedure for resuming lock step synchronization with the CPU subsystem of the other system. The full backup program according to claim 7. In the fault tolerant server,
When the one system receives the activation signal from the other system and the other CPU subsystem continues the processing performed before stopping the lockstep synchronization, the one system The full backup program according to claim 7 or 8, further comprising executing a procedure for shifting to a mode in which a specified device is not used.

Images