DE19758588B4 - Computer system with checkpoint and restart facility - includes output unit providing accounting information indicating charge payable as result of program execution and checkpoint and restart processes to accounting files. - Google Patents

Abstract

The system includes a program execution unit. An output unit outputs accounting information to an accounting file, indicating a charge payable as a result of the program execution. Accounting information indicating a charge payable as a result of the checkpoint and a restart process is sent to a separate accounting file. This occurs when a checkpoint process for the user program is completed within the program. A determination unit determines in the event of access of a file which has been preserved as a checkpoint, whether the file has been saved. If the file has not been saved a saving unit appends the file to the checkpoint.

Description

BACKGROUND OF THE INVENTION

1. Field of the invention

The The present invention relates generally to computer systems and in particular a checkpoint and restart facility or facility of a computer system.

2. Description of the related technology

One Computer system includes a CPU, a memory, an external storage device and a Terminal equipment, etc. The CPU is running a program loaded in memory.

One Operating system (OS) is stored in the external storage device and is loaded into memory when the computer system starts up will be to afterwards to control the computer system.

As the OS is a user program in the external storage device saved and it is executed after passing it through the OS according to a Instruction from the terminal or the like in the memory loaded.

The Loading the entirety of a large program or data in the store is impossible because the size of the memory is limited. Therefore, a so-called virtual memory scheme used, so that a Process is running, by putting sections of the program or data between the memory and the external storage device.

The Program becomes ordinary in execution units executed which is called a process become. A virtual space made up of a virtual storage space and a virtual register, etc., is assigned to a process. A job includes at least one process and Job information to process Taxes.

There a computer system is constructed so that it can be accessed by a variety of Users is used and a variety of programs is run concurrently, the OS offers several functions.

To the For example, the computer system provides a checkpoint and restart facility in preparation the occurrence of a system failure while a program is running.

A Prüfpunkteinrichtung serves to secure the runtime environment for each execution unit, such as B. for one active job or an active process. A restart facility is used to restore the runtime environment through the checkpoint facility was secured, and restarting the execution of the Program.

The OS sees system calls that contain a file input and output process for processes which are executed by different user programs together. The system call is like a subroutine for the process. When a system call is requested will, leads the OS has an associated process in it a process room of the program requesting the system call. The user an amount of money is charged, which is payable for the used CPU time, in which the OS and the user program are running.

at the checkpoint and restarting apparatus of the related art are the following Problems have been detected. For some checkpoint implementations The runtime environment of the process that is provided with a checkpoint is backed up by some previous system calls the user program was requested. That is, the checkpoint process is not distinguished from processes which are executed by the user program. As a result, will the user of the checkpoint process into account posed.

Of the Checkpoint process can after Discretion of the system operator so that it does not desirable is, the user the process, the one with a checkpoint is to be included consistently. This is a first problem the checkpoint and restarting device according to the related ones Technology.

One second problem with the checkpoint and restart means according to the related art is that billing information only be issued to a billing file when a job ends is.

It can not all user programs end normally. Some store an intermediate output in a file during a process to the execution process later using the intermediate output. In such cases it for the operation of the system desirable, the Users to count the process even if the program does not end normally.

When the checkpoint and restart process is executed, the user is repeatedly billed for a job when a restart occurs. For example, from a start time of a job to a checkpoint time, the user is charged multiple times when a job centers with a checkpoint in the middle for troubleshooting purposes is and starting from the checkpoint many times with different conditions a restart.

One third problem is that only one active file (a opened File or one being accessed) in the checkpoint process becomes. More specifically, the restart process may be the runtime environment not just by restoring the information backed up at the checkpoint if a file is updated following the checkpoint process that was not the file is at the checkpoint was secured.

The third problem will be explained in detail below.

It it is assumed that the Start the use of the file A at the time T0 takes place and a checkpoint process to the Time T1 is started. Then it is assumed that the contents of the file B during the Process X is consulted at the time T2 and the file B during the Process Y is updated at time T3. If the runtime environment, which was saved at the time T1, is restored, so that the process again starts, The content of file B differs on a restart of the process X due to the process Y at the time T3 of that he was T2 at the time, and he can not be restored.

According to DE 42 20 198 C2 and DE-Z: HARD AND SOFT 6, (1989) No. 3, pp. 44-47, in effect, secures a database application or the like files (database) and access logs of these files in a nonvolatile memory according to processing logic of itself, and thereby For example, recovery may be in the event of a system failure or the like. More specifically, information (files) to be backed up are explicitly backed up each time in checkpoint processing, and all access logs are backed up as well.

SUMMARY THE INVENTION

Therefore It is a general object of the present invention to provide a computer system to provide in which the above third problem is eliminated and a Runtime environment can also be restored, if one in the checkpoint process not active file following the checkpoint process is updated.

One another and more specific object of the present invention is to provide a computer system that has a more accurate accounting process than the conventional System offers.

These The object is solved by the features of claim 1.

According to the computer system the present invention, a billing process is possible, the with an actual execution of a user program, even if a system failure occurs in a computer system.

According to these Aspects of the present invention is a file (or computer resource), on the following accessed the checkpoint process is automatically backed up. That's why it's possible to have one restore previous runtime environment when a job resumes accomplished becomes.

BRIEF DESCRIPTION OF THE DRAWINGS

Other Aims and further features of the present invention are apparent the following detailed description in conjunction with the attached Drawings in which:

1 shows the construction of a computer system according to an embodiment of the present invention;

2 showing the construction of job information;

3 shows the construction of process information;

4 shows a record format of a billing file;

5 shows a record format of a checkpoint instruction information table;

6 show a flow of a job freezing process that represents a checkpointing process;

7 show a flow of a process responsive to a request for a system call with checkpointing means;

8th show a flow of a process that responds to a request for a system call that includes file access;

9 shows a flow of a process freezing process that represents a checkpointing process;

10 shows a sequence of a restart process; and

11 shows a second structure of the process information.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

1 shows the construction of a computer system according to an embodiment of the present invention.

The computer system includes a computer 1 , an external storage device 2 and a terminal unit 3 , The computer 1 includes a CPU 4 and a memory 5 , The memory 5 saves a loaded operating system (OS) 6 and jobs 7-1 and 7-2 ( 7-n ).

The loaded OS 6 contains a checkpoint process agent 8th , a recovery process agent 9 , a settlement agent 10 , a system call process agent 11 and a checkpoint instruction information table 12 ,

The checkpoint process agent 8th saves the runtime environment of a job in the external storage device 2 as a checkpoint file. The restart process agent 9 Restores the runtime environment of the job created by the checkpointing agent 8th in the external storage device 2 saved as a file. The settlement agent 10 manages, as billing information, the input / output (I / O) volume and CPU time used when a program is under the loaded OS 6 running.

Each of the jobs 7-1 and 7-2 ( 7-n ) includes job information 13 and at least one process. With reference to 1 contains the job 7-1 job information 13-1 and a process 14-1 , The job 7-2 contains job information 13-2 and a process 14-2 , More generally, a job 7-n contains job information 13-n and a process 14-n , The job information 13-n are information to control the job. The process 14-n is a minimal execution unit of a program. Every process 14-n Contains process management information for controlling the process that is in 3 are shown.

An operating system 15 , a checkpoint file 16 , a program 17 , a data file 18 and a billing file 19 are in the external storage device 2 saved. When the computer system starts, the operating system becomes 15 in the store 5 of the computer 1 loaded. The checkpoint file 16 is a file that stores the runtime environment of a job created by the checkpointing process agent 8th was secured. The checkpoint file 16 can be a variety of files instead of a single file.

The program 17 will be under the loaded OS 6 executed. On the data file 18 is through the program 17 accessed while it is running. The billing file 19 stores the billing information provided by the billing agent 10 to get managed.

With reference to 2 Now a description of the job information will be given 13-n , As in 2 shown, contain the job information 13-n Information required to control a job, such as: A job ID, a checkpoint process completion flag, a checkpoint filename, a plurality of process flags, and a plurality of names of backed-up files being accessed.

The job ID 13-n is a number to one of a variety of jobs under the control of the loaded OS 6 to specify. The checkpoint process completion flag is information that indicates whether the runtime environment of a job has already been saved as a checkpoint file. The checkpoint file name is a name of a file that stores the job runtime environment. The process tag is a number associated with the process representing the job. The name of a backed up file is a name of a backed up file that was active on the checkpoint (used by the job).

With reference to 3 Now a description of the process information. The process information includes the process tag (process ID), the job tag, the user tag, the CPU time used, the I / O volume, the user tag save area, the I / O volume save area, and other information (not shown) used for process management required are.

The process ID is a number to one of a plurality of processes under the control of the OS 6 to specify. The job ID is a number associated with the job to which the process belongs. The user ID indicates the user who performed the process. The CPU time used is a cumulative time in which the CPU was used by the process after its execution started. The I / O volume is a cumulative volume in bytes that indicates the input and output volumes after the process starts. The used CPU time and the I / O volume are calculated by the accounting tool 10 of the OS 6 updated at an appropriate time during the execution of the process.

The backup area of the used CPU time and the backup area of the I / O volume are used to calculate the used CPU time and temporarily secure the I / O volume.

The billing file 19 consists of a large number of records that use a record format as in 4 to have. Each of the plurality of records includes the user ID, CPU time used, I / O volume, and other required information (not shown). The user ID designates the target user for billing. The used CPU time is a time in which the CPU is used to execute the user program. The I / O volume is an input and output volume while the user program is executing. The amount of money charged to the user is determined on the basis of this and other required information.

The checkpoint instruction information table 12 includes a plurality of records with a record format as in 5 , Each of the plurality of records in the checkpoint instruction information table 12 contains the user identifier indicating a user who issued an instruction to set a checkpoint of the job, and the job ID of a job that is a target of freezing.

With reference to 6 A description will now be given of the checkpointing process performed by the checkpointing process agent 8th is performed. The checkpoint process includes a job freezing process and a process freezing process. The job freezing process is a process for backing up the runtime environment of the job and will be described with reference to FIG 6 explained. The process freezing process will be described with reference to FIG 9 explained.

With reference to 6 when the user (for example, a system administrator) is instructed to set a checkpoint of a job via the terminal device unit 3 and the job ID of the target job for setting a checkpoint in the checkpoint instruction information table 12 registered (step S21).

When registering the information in the checkpoint instruction information table 12 is completed, the process representing the job receives a message telling the consummation of the freeze process (securing the runtime environment of the process - see 9 ) (Step S22). When the message has been received, the name of an active file that has been notified is registered in an already saved file (step S23). A description of process freezing will be given later.

The Steps S22 and S23 are repeated until the completion of freezing all processes representing the target job have been reported (Step S24).

The file whose name was registered in the backed up file will be in the checkpoint file 16 saved (step S25), and the job information is in the checkpoint file 16 secured (step S26).

It A description of the process freezing process (Save the runtime environment of the process).

It should be noted that each process is executed by the loaded OS 6 a system call is requested, which is intended as a subroutine. In this invention, the checkpointing process is executed after a process is executed that responds to the system call request. 7 Figure 12 shows a flow of a process that follows a request for the system call and does not include input or output of files. A description of the process that includes access to the file will be provided later.

9 shows a flow of a process for freezing an execution state. The checkpoint instruction information table 12 is first consulted to determine if the job to which the process belongs is a target of checkpointing. If the job is not a destination, the process is terminated immediately (step S50).

Then, a billing record containing the user ID becomes a difference between the used CPU time and the value saved in the backup area of the used CPU time, and a difference between the I / O volume and the value stored in the backup area of the I / O volume, to the billing file 19 outputted (step S51). The used CPU time is saved in the backup area of the used CPU time, and the I / O volume is saved in the backup area of the I / O volume (step S52). The process freeze step is then executed, and the completion of the freeze process and the identity of the active file become the loaded OS 6 reported (step S53). The notification made at step S53 is the message received at step S22 (see FIG 6 ).

When the process freezing process is completed, the accounting record containing the user identifier designating the user issuing an instruction to set a checkpoint of the job is a difference between the CPU time used and the value stored in the backup area of the CPU used Time is assured, and a difference between the I / O volume and the value saved in the backup area of the I / O volume, to the billing file 19 outputted (step S54). The used CPU time is saved in the backup area of the used CPU time, and the I / O volume is saved in the backup area of the I / O volume (step S55). The process freezing process is then completed.

With reference to 8th Now, a description will be given of a process that responds to a request from the system call process and includes access to a file.

First there is a provision concerning whether a checkpoint process in the Job executed which the process belongs to requires access to a file based on the checkpoint process completion flag in the job information. If it is determined that the checkpoint process is not executed, control goes to step S47 (step S41).

If it is determined that the Checkpoint process is executed, there is a provision regarding of whether the destination file for the access is secured, based on the name of the secured File in the job information. If it is determined that the destination file for the Access has been secured, control proceeds to step S47 (step S42).

The billing information is then output to the billing file (step S43). This process is the same as the processes in steps S51 and S52 of FIG 9 , That is, a difference between the used CPU time and the value saved in the backup area of the used CPU time, and a difference between the I / O volume and the value stored in the backup area of the I / O volume. Volume is secured, will be sent to the billing file 19 output. Further, the used CPU time is saved in the backup area of the used CPU time, and the I / O volume is saved in the backup area of the I / O volume.

Subsequently, the destination file for access is saved (step S44). The name of the file saved at step S44 is registered under the name of the saved file which constitutes the job information arranged in the memory. The job information contained in the checkpoint file 16 are stored are updated on the basis of the job information arranged in the memory (step S45).

When the saving of the accessed file and the updating of the job information are completed, the billing information is output (step S46). This process is the same as the processes in steps S54 and S55 of FIG 9 , That is, the billing record containing the user identifier designating the user issuing an instruction for setting a checkpoint of the job, a difference between the used CPU time and the value saved in the backup area of the used CPU time , and a difference between the I / O volume and the value saved in the save area of the I / O volume is sent to the billing file 19 output. The used CPU time is saved in the backup area of the used CPU time, and the I / O volume is saved in the backup area of the I / O volume.

A description will be given of a restart process performed by the restart process means 9 is performed. A restart process is a restart process for loading the runtime environment of the job that is in the checkpoint file 16 stored in the external storage device as a result of the checkpointing process, equal to the memory.

When the job is restarted, the job information is taken from the checkpoint file 16 retrieved and loaded into the memory (step S61).

The saved file (the file that was active when the job was frozen is then restored on the basis of the content of the information that were loaded into the memory (step S62). Subsequently, will one or a variety of processes that represent the job, restored (step S63). When these restorations are completed, the billing information is output (step S64).

The output of the accounting information is the same as the corresponding processes in steps S54 and S55 of FIG 9 , That is, the billing record containing the user identifier designating the user issuing an instruction to restart the job, a difference between the used CPU time and the value saved in the backup area of the used CPU time, and a difference between the I / O volume and the value saved in the save area of the I / O volume is sent to the billing file 19 output. Further, the used CPU time is saved in the backup area of the used CPU time, and the I / O volume is saved in the backup area of the I / O volume.

The The above problems peculiar to the related art are accomplished the computer system according to the above described embodiment solved.

With reference to 11 A description will now be given of a variation of the embodiment of the present invention. With reference to 11 For example, the process information may include a second backup area of the CPU time used and a second backup area of the I / O volume.

According to this variation, the billing information available when the normal process is started (that is, when the checkpoint process ends) is saved in a second billing information backup area composed of the second backup area of the CPU time used and the second backup area of the billing information I / O volume is formed. The billing information that is available when the normal process ends (that is, when the checkpoint process starts) is saved in a first billing information backup area, which consists of the first backup area of the CPU time used and the first backup area of the I / O server. Volume is formed. When the checkpoint process ends, a billing record is sent to the billing file 19 outputted indicative of an amount of money corresponding to a difference between the value saved in the first accounting information backup area and the value saved in the second accounting information backup area and credited to the user who runs the program (Job , Process). A billing record indicating the amount of money corresponding to a difference between the current billing information and the value saved in the first billing information backup area and credited to the user requesting the checkpoint of the job is output.

So Also, the problems inherent in the related art are through the change the embodiment solved.

While the above description relates to a file access that occurs after the checkpoint process, the present invention is not intended solely for file access, but the same solution can for Computer resources that are not files.

at In the above description, it is assumed that the job information with connected to a job and the process information associated with a process are. However, you can the job information and the process information in a memory stored by the 05 or the external storage device is managed.

While the Invention has been described with reference to specific embodiments is the description of the specific embodiments only illustrative, and it should not be considered as limiting the scope of the invention become. Professionals will be various other modifications and changes come up without the basic idea and scope of protection of the present To leave invention.

Claims (2)

Computer system provided with a resource access means and checkpoint and restart means, characterized by comprising: a program execution means ( 4 ) for executing a program using at least one execution unit that processes a plurality of resources; a backup means for backing up the content of an active resource; determination means for determining, in the case where an execution unit accessing a resource is preserved as a checkpoint, whether the resource being accessed has been saved; and securing means for securing the resource accessed when the determining means determines that that resource has not been saved to attach to the checkpoint. Computer system according to claim 1, characterized that the Resource is a file.