KR101130850B1 - Apparatus and method for checkpoint control - Google Patents

Abstract

The present invention relates to a checkpoint control apparatus and method. The present invention provides a checkpoint controller for storing, transferring, changing, and deleting checkpoint information to recover from a system failure, and a RAM disk configured as a shared area of a file system format accessible to all processes. At least two nodes are included that are configured to store information. The checkpoint controller sets a portion or the entirety of the main memory as the RAM disk, assigns and assigns a checkpoint region to the RAM disk, and stores checkpoint information in the allocated checkpoint region. The present invention stores the checkpoint information in a RAM disk, which is a shared area accessible to all processors, thereby improving access speed and convenience of use of the checkpoint area.

Checkpoint Information, RAM Disk, Cluster, High Availability

Description

Checkpoint control device and method {APPARATUS AND METHOD FOR CHECKPOINT CONTROL}

1 is a block diagram showing a cluster computer including a checkpoint control apparatus according to an embodiment of the present invention;

2 is a flowchart illustrating a checkpoint control method according to an exemplary embodiment of the present invention;

3 is a block diagram illustrating a high availability system including a checkpoint control apparatus according to another embodiment of the present invention.

`` Explanation of symbols for main parts of drawings ''

10, 20: node 11, 21: ramdisk

12, 22: control unit 13, 23: process address space

14a, 17, 24a: checkpoint area 15, 25: hard disk

16: Cluster File System

The present invention relates to a checkpoint control method, and more particularly, to a checkpoint control apparatus and method for diversifying a method of accessing checkpoint information and easily performing checkpoint information tracking when a failure occurs.

The checkpoint technique records the state information of the cluster application in the checkpoint area and provides a function to preserve the state information of the application in the event of a software or hardware failure. When recovering from a failure situation, the checkpoint technique retrieves data stored in the checkpoint area and restarts the application from the state recorded before the failure, thereby minimizing data loss due to the failure and reducing failure recovery time.

A conventional checkpoint controller allocates a checkpoint area to a memory area allocated to a data structure of a process address scope or to a file system of a hard disk as a storage medium.

This conventional checkpoint control device has the following problems.

First, when a checkpoint area is allocated to and managed by a main memory area within the process, the main memory area is set as a protection scope within the process, so that a process other than the corresponding process accesses the checkpoint area. access). Therefore, when a failure occurs in the corresponding process, there is a problem in that the checkpoint information cannot be used using another process.

When the checkpoint area is allocated to the file system of the hard disk and managed, the time required to assign checkpoint information is increased according to the characteristics of the mechanically driven disk, compared to the case of using the memory area. . That is, the use of the memory area is a direct assign method, whereas the use of a disk file system has an indirect allocation method through a disk input / output device controller, thereby reducing the processing speed.

Accordingly, an object of the present invention is to provide a checkpoint control apparatus and method for easily managing a checkpoint area as a file system.

Another object of the present invention is to provide a checkpoint control apparatus and method for shortening the time required for allocating / using a checkpoint area.

According to a feature of the present invention for achieving the above object, the present invention includes a checkpoint control unit for performing the storage, transfer, change and delete operation of the checkpoint information to recover the system failure; The systems are configured to comprise at least two systems comprising a checkpoint storage unit configured to set a shared area of an accessible file system format to store the checkpoint information.

Preferably, the checkpoint storage unit is a RAM disk.

The checkpoint controller preferably allocates a portion of the checkpoint storage unit to a checkpoint region.

The checkpoint information is preferably generated according to the AIS Spec of the Service Availability Forum (SAF).

Preferably, the checkpoint controller controls the access control setting and the read / write operation of the checkpoint area according to the file input / output method.

The checkpoint controller may be configured to extract a data offset, a data size, a data length, and data from the checkpoint information and store the data offset in the checkpoint area using an input / output system call function.

Preferably, the checkpoint controller selectively controls access to the checkpoint storage unit according to a preset security setting.

The security setting is preferably set for each node, for each processor, or for each user.

The present invention is preferably configured to further include a file system provided with the checkpoint storage unit for backup and storage separately in case a failure recovery through another system is impossible.

The file system is preferably a cluster file system in which the systems share stored information.

According to another aspect of the present invention, there is provided a method of generating a RAM disk for storing checkpoint information in at least two systems, allocating a checkpoint region to a predetermined region of the generated RAM disk. And storing the checkpoint information in the allocated checkpoint area, and recovering a failing system using the stored checkpoint information.

The RAM disk is preferably set as a shared area of a file system format.

The storing step may further include a security setting step of setting an access restriction of the stored checkpoint information.

In the security setting step, it is preferable to set access authority for each node, processor, or user.

The storing step may further include a step of backing up the checkpoint information to a file system set as a shared area so as to recover the system when an abnormality of the RAM disk occurs.

The checkpoint information is preferably stored by a file I / O method.

In the storing step, it is preferable to extract and store a data offset, a data size, a data length, and data from the checkpoint information.

According to the present invention having the configuration described above, a checkpoint area is allocated to a RAM disk having a file system format to store and synchronize checkpoint information. Accordingly, the present invention speeds up access to checkpoint information and improves ease of management.

Hereinafter, a checkpoint control apparatus and method according to an exemplary embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram illustrating a cluster computer including a checkpoint control apparatus according to an exemplary embodiment of the present invention.

In FIG. 1, the cluster computer comprises a plurality of nodes 10, 20 consisting of terminal units such as one or more hosts, personal computers, or portable computers. Since the configurations of the nodes are the same, the configuration of one node will be described in the present embodiment.

Although not shown, any one of the nodes 10 and 20 of FIG. 1 includes a process for performing overall operations of the node, a communication module for performing communication between the nodes 10 and 20, and the process. The main memory, which is a space required for the operation of the controller, is configured.

Each of the nodes 10 and 20 checkpoint controls to store / register / change / delete checkpoint information generated according to AIS Spec of Service Availability Forum for system recovery. The control unit 12 is provided. The checkpoint controller 12 drives the process to perform various operations of the node 10, and manages checkpoint information of the node 10 and other nodes 20. To this end, the checkpoint controller 12 sets a portion or all of the main memory to the RAM disk 11 by an initial configuration of software / hardware. In addition, the checkpoint controller 12 replaces the RAM disk 11 with the checkpoint region 14a instead of the checkpoint region 14 that is a protection scope in the conventional process internal address space 13. Reallocate and store the checkpoint information.

The checkpoint control unit 12 stores the checkpoint area 14a by synchronizing the operating states of the nodes 10 and 20 constituting the cluster computer, and when the abnormality occurs at any one node, the stored checkpoint. Provide information to the offending node. The checkpoint area 14a is set to a file system format, and is a shared area accessible to all processes of each node 10 or 20 or to the checkpoint controller 12 and 22. Is preferably.

The checkpoint control unit 12 is a kind of application, which extracts a data offset, a data size, a data length, and data from the checkpoint information. Control to store at the same address of the checkpoint areas 14a and 24a of each of the nodes 10 and 20. In addition, the checkpoint controller 12 performs an operation of restricting access to the RAM disk 11 according to a security setting based on an access control setting for each node, processor, or user.

In addition, hard disks 15 and 25 for storing various programs and data are connected to the nodes 10 and 20. In addition, a cluster file system 16 shared by all nodes 10 and 20 is provided. The cluster file system 16 checks the checkpoint areas 14a and 24a when an error occurs in the RAM disks 11 and 21 of all the nodes 10 and 20 or because of access control restrictions on the node. It is preferable to include a checkpoint area 17 for backing up and storing the checkpoint information in case of not being able to access).

The cluster computer including the checkpoint control apparatus according to the present invention configured as described above allocates a checkpoint area storing checkpoint information generated at each node to a RAM disk set in a shared area and a file system format of each node. .

Next, a checkpoint control method according to a preferred embodiment of the present invention will be described in detail with reference to FIG. 2.

FIG. 2 is a flowchart illustrating a checkpoint control method of the first node 10 in the checkpoint control apparatus shown in FIG. 1 step by step. In the present exemplary embodiment, the first node 10 will be described. However, the present invention may be equally applied to a checkpoint control method of a cluster system including a plurality of nodes 10 and 20.

In the tenth step S10 of FIG. 2, when driving is started when power is applied, the checkpoint controller 12 of the first node 10 may delete all or part of the main memory by the configuration. The RAM disk 11 is used.

In detail, the checkpoint controller 12 may include a file system in which the process may manage checkpoint information using a file input / output method in a part of a memory area of the RAM disk 11. Format with). In addition, the generated file system is configured to set a point to be loaded in the memory area so that the memory area becomes usable with the set file system.

For example, given the fact that checkpoints are used on Linux, the process of creating a RAM disk (11) in Linux,

1) '# mk2fs / dev / ram0', which formats 'ram0', the memory area provided by Linux into 'ext2 file system',

2) '#mkdir -p / mnt / ram0', which sets the point to be loaded

3) This is done via '#mount -t ext / dev / ram0 / mnt / ram', which loads the memory area into the 'ext2 file system'.

In a twelfth step S12, the checkpoint control unit 12 reallocates the checkpoint area 14a to the RAM disk 11. For example, 'struct saCkptCheckpointSection {void * sectionData; } 'Is specified as' / dev / ram0', the RAM disk 11. At this time, the checkpoint control unit 12 allocates the checkpoint area 14a to the pointer size, and when data to be synchronized is actually received, the checkpoint control unit 12 reassigns the reallocation by the section assign function of the checkpoint. By specifying the section data size and allocating application data. In this reallocation, once again, the section data area is designated as the RAM disk 11 area of '/ dev / ram0', and application control data is allocated. For example, on Linux, the section assignment function's "sectionData = realloc (ckptCheckpointSection-> sectionData, sizeRequired);" Is assigned to 'sectionData = / dev / ram0'.

 In a fourteenth step S14, the checkpoint control unit 12 controls access to the checkpoint information S16 generated as the system is driven by the process using a file I / O method. Set access control and read () / write (). In this case, the checkpoint controller 12 may generate a data offset from the checkpoint information generated at the first node 10 or received at the external second node 20 using a system call function. ), Checkpoint area 14a allocated to RAM disk 11 using data size, data length, data, and I / O system call function. Store in

The checkpoint controller 12 preferably performs the access control setting for each node, processor, or user. In detail, the checkpoint controller 12 controls an access mode of the ram disk 11 so that only an authorized process can access the ram disk 11. For example, you can set the access mode to 'rwxrwxrwx'. This is classified from the left into the account of the first node 10, the group to which the first node 10 belongs, and other process groups, and reads the checkpoint area 14 of the first node 10 for each group. (r), write (w), execute (x).

Subsequently, the checkpoint controller 12 multicasts the checkpoint information through the communication module 50 so as to synchronize checkpoint information of another node 20 (S20). In addition, the checkpoint control unit 12 may check checkpoint information stored in the checkpoint area 14a of the RAM disk 11 at the time of failure of all the nodes 10 and 20 or in accordance with the access control restriction. The checkpoint information is also stored in the checkpoint area 17 of the cluster file system 16 in case it is not available.

If a failure occurs in any one node, for example, the second node 20 in the twenty-second step S22, if the process of the second node 20 is set to be accessible to the first node 10 (S24). That is, the failure node, that is, the second node 20 may be restored using the checkpoint information stored in the checkpoint area 14a of the first node 10 (S26).

On the other hand, when the process set to disable access control in the 24th step (S24) requests access, the checkpoint control unit 12 of the first node 10 responds that the checkpoint area 14a is inaccessible. do. In this case, the checkpoint controller 22 of the second node 20 may perform a recovery operation using the checkpoint information stored in the checkpoint area 17 of the cluster file system 16 (S19).

The checkpoint control apparatus and method according to the preferred embodiment of the present invention through the above process allocates a RAM disk as a checkpoint area and stores the checkpoint information.

Next, Figure 3 is a block diagram of a checkpoint control apparatus according to another embodiment of the present invention. In FIG. 3, a checkpoint controller applied to a high availability system is configured to store checkpoint information in the checkpoint area 114 of the RAM disk 111. The configuration of FIG. 3 is similar to that of FIG. In brief, at least two nodes are configured in connection with the high availability system. Each node 110 and 120 includes a checkpoint controller 112 and 122 and a RAM disk 111 and 121 allocated to store checkpoint information as shown in FIG. 1.

In FIG. 3, the central processing unit (not shown) of the high availability system 130 may access the checkpoint area 114a to know the state of the application 110, and the application ( It performs a failover function according to the state of 110. The failover function refers to a backup mode function in which a 2-wire system takes over a function and operates when the 1-wire system is stopped due to failure or end of operation time. This failover function is used to create a non-disruptive system, and is applied in the event of a personal computer (PC) system (in case of a processor failure) and a network (path connection, storage, web server) failure. If data is not accessible due to a single server failure, multiple paths, such as a storage area network (SAN), are used to ensure connectivity is maintained in any path failure.

For example, the central processing unit of the high availability system 130 uses the checkpoint information stored in the checkpoint area 124a of the second node 120 when a failure of the first node 110 occurs. 110 may be recovered.

Accordingly, the checkpoint control apparatus according to another embodiment of the present invention is applied to a high availability system in which two physically different devices are logically configured as one device to prepare for a failure.

As described above, the present invention has been described with reference to the embodiments shown in the drawings, which are merely exemplary, and a person of ordinary skill in the art without departing from the spirit and scope of the present invention. It will be apparent that various modifications, changes, and equivalent other embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

As described above, the present invention allocates a RAM disk as a checkpoint area to store checkpoint information. Accordingly, the present invention has the following effects.

First, the present invention manages the checkpoint area like a file system, thereby improving manageability of the checkpoint information.

In addition, since the present invention uses a memory file system, a processing speed faster than that of a disk file system can be obtained.

In addition, the present invention may access all processes to the RAM disk, and may limit processes accessible to the checkpoint area through security settings.

In addition, the present invention provides convenience for tracking the checkpoint information required because all processes having access rights retrieve the checkpoint information by the file input / output method.

Claims (17)

A checkpoint controller for storing, transferring, changing, and deleting checkpoint information to recover a system failure; And at least two systems including a checkpoint storage unit configured to store the checkpoint information, wherein the systems are set as a shared area of an accessible file system format. The method of claim 1, wherein the checkpoint storage unit, Checkpoint control device, characterized in that the RAM disk. The method of claim 1 or 2, wherein the checkpoint control unit, And a portion of the checkpoint storage unit is allocated to a checkpoint area. The method of claim 1 or 2, wherein the checkpoint information, Checkpoint control device, characterized in that generated in accordance with the AIS Spec (Service Availability Forum). The method of claim 1 or 2, wherein the checkpoint control unit, And a checkpoint control apparatus for controlling access and setting of a readpoint and a read / write operation of the checkpoint area according to a file input / output method. The method of claim 1 or 2, wherein the checkpoint control unit, And extracting a data offset, a data size, a data length, and data from the checkpoint information and storing the data offset, data size, data length, and data in the checkpoint area using an input / output system call function. The method of claim 1 or 2, wherein the checkpoint control unit, And a checkpoint controller for selectively controlling access to the checkpoint storage unit according to a preset security setting. The method of claim 7, wherein the security setting, Checkpoint control device, characterized in that configured for each node, processor or user. The method of claim 1, And a file system provided with the checkpoint storage unit to separately backup and store in case a failure recovery through another system is impossible. The method of claim 9, wherein the file system, And a cluster file system in which the systems share stored information. Creating a ram disk for storing checkpoint information in at least two systems; Allocating a checkpoint area to a predetermined area of the created RAM disk; Storing the checkpoint information in the allocated checkpoint area; And recovering a failing system using the stored checkpoint information. The method of claim 11, wherein the ram disk, A checkpoint control method, characterized in that the file system format is set as a shared area. The method of claim 11, wherein the storing step, And a security setting step of setting an access restriction of the stored checkpoint information. The method of claim 13, wherein the security setting step, Checkpoint control method characterized in that the access rights are set for each node, processor, or user. The method of claim 11, wherein the storing step, And backing up the checkpoint information to a file system set as a shared area so as to recover the system in the event of an abnormality of the RAM disk. The method of claim 11, wherein the checkpoint information, Checkpoint control method characterized in that stored by the file I / O method (file I / O method). The method of claim 11, wherein the storing step, And extracting and storing a data offset, a data size, a data length, and data from the checkpoint information.

Images