JP2500735B2 - Checkpoint / restart controller - Google Patents

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION The present invention relates to a checkpoint / restart controller.

[0002]

2. Description of the Related Art A conventional checkpoint / restart control device is provided with a process identifier reservation table, one bit of this table is associated with one process identifier, a checkpoint is taken, and a restart is performed based on the information. At that time, the process was restarted using the corresponding process identifier.

Further, a process identifier reservation file for saving the information of the corresponding process identifier reservation table is prepared,
Even when the system is started / started up, the process identifier reservation table is created from the process identifier reservation file to guarantee the process identifier.

[0004]

In the above-mentioned conventional checkpoint / restart control device, there is a guarantee when the system is started / started up, but one checkpoint for one process identifier. Since only information (one restart file) can be collected, there was a problem when collecting multiple checkpoint information.

The process identifier is reserved by collecting checkpoint information, and this process identifier cannot be used during normal process generation. Therefore, depending on the operation method of the checkpoint / restart function, all process identifiers may be reserved, and in some cases, the system may not be able to create a completely new process.

[0006]

The device of the present invention comprises a CPU
Process generation means that generates a process to be allocated, checkpoint execution means that executes checkpoint processing when a checkpoint request is accepted during process execution, and process state at that time is frozen to a re-executable state. A restart file for storing the process identifier, a process identifier reservation table indicating whether or not the process identifier needs to be saved for restart processing, and a process identifier called by the checkpoint execution unit at the time of checkpointing the process identifier. Process identifier reservation means for making a reservation in the process identifier reservation table,
In a checkpoint / restart control device in a computer system having restart file deleting means for deleting the restart file when it is no longer needed, a process for performing a check for determining a process identifier to be assigned when the process is generated A generation time control means and a process identifier reservation release control means for releasing the reservation of the process identifier when the restart file is deleted are provided, and the process identifier reservation table has a checkpoint executed once. , The value of the process identifier area is incremented by 1,
When the value is 1 or more, the corresponding process identifier cannot be used in normal process generation, and even if one of the restart files is deleted, a restart file for the corresponding process identifier exists. The feature is that the reservation of the corresponding process identifier is not released, and control is performed so that restart cannot be disabled due to duplication of process identifiers at restart.

[0007]

Next, the present invention will be described in detail with reference to the drawings.

Referring to FIG. 1 showing a first embodiment of the present invention, in the present embodiment, a process generation means 1, a checkpoint execution means 2, a process identifier reservation means 3, and a restart execution means 4 are provided. , Restart file deleting means 5
And a process identifier monitoring unit 6 (consisting of a process generation control unit 7 and a process identifier reservation release control unit 8), a restart file 101, a process control table 102, and a process identifier reservation table 105. .

FIG. 4 shows the process identifier reservation table 10
It is a block diagram of FIG. This process identifier reservation table 1
In 05, a 1-byte area is assigned to one process identifier, and the area is prepared in an array corresponding to the maximum process identifier prepared as a system parameter. For example, when a checkpoint request is issued to a process whose process identifier is "30", the process identifier reservation unit 3 increments the area (1 byte configuration) allocated for this process identifier "30" by one. . At this time, the process identifier is already “30”.
On the other hand, if one restart file 101 is created, the value of the area becomes “2” by the increment of 1 this time. A process identifier having a value of "1" or more in this area is a rule not used during normal process generation.

After that, when the restart file 101 is deleted by the restart file deletion means 5, the process identifier reservation release control means 8 is called and decremented by 1 from the area of the corresponding process identifier. When the value of this area becomes “0”, the corresponding process identifier can be used during normal process generation.

FIG. 5 shows a processing flow executed by calling the process identifier reservation means 3 from the checkpoint execution means 2.

When a checkpoint request is issued to the process having the process identifier "30" (step 3 in FIG. 5).
-1), the checkpoint execution means 2 is activated, and first the state of the process with the process identifier "30" is frozen (step 3-2). Then, the restart file 101 which stores the state of this process is prepared (step 3-3). After that, prepared restart file 10
1 stores information necessary for process restart, such as a process control table 102 and a file management table (not shown). At this time, the process identifier "30" of this process is stored in the restart file 101 together with other information (step 3-4). At this time,
The process identifier “30” is stored in the newly prepared process identifier storage area.

The checkpoint execution means 2 transfers control to the process identifier reservation means 3, and the process identifier reservation means 3 increments the area corresponding to the process identifier "30" in the process identifier reservation table 105 by 1 (step 3- 5). As a result, the process identifier saving process in the checkpoint process ends (step 3-6).

Next, a normal process generation process will be described with reference to FIG.

First, the process generation means 1 is called,
The process identifier is acquired (step 4-1). Since the recently acquired process identifier is stored as the process identifier state variable, the next identifier to be assigned is determined from this variable. When the process identifier exceeds the maximum process identification number (NO in determination 4-2), the identifier to be assigned is retaken from the young process identifier again (step 4-3).

Once the identifier has been determined, the process identifier reservation table 10 is set by the process creation control means 7.
It is checked whether the area of the process identifier of 5 is 1 or more. If the value is 1 or more (NO in decision 4-4), this process identifier is reserved and cannot be assigned. Therefore, another process identifier is taken again (step 4-5). By the above processing, if the unreserved process identifier can be finally obtained (step 4-6), the process generation means 1 continues the normal processing.

Next, the process of deleting the restart file 101 will be described with reference to FIG.

For example, it is assumed that a request for taking a checkpoint or deleting the start file 101 is issued to the process having the process identifier "30" (step 5-1). The restart file deletion means 5 calls the process identifier reservation release control means 8 before deleting the file and opens the restart file 101 (step 5-2). Then, the process identifiers constituting the restart file 101 are acquired from the process identifier storage area in the restart file 105 (step 5-3).

The process identifier reservation table 10
The corresponding area of the process identifier of 5 is decremented by 1 (step 5-4). Note that a plurality of process identifiers may be reserved and stored in one restart file 101. In that case, this process is performed for each process identifier.

Thereafter, the process identifier reservation release control means 8 returns control to the restart file deletion means 5 and executes the original restart file deletion processing (step 5).
-5). By the above processing, when the area of the process identifier reservation table 105 is decremented by 1 and its value becomes "0", the process identifier corresponding to this area can be assigned during normal process generation. .

Next, a second embodiment of the present invention shown in FIG. 2 will be described. In this embodiment, process identifier allowable range control means 9 is added to the first embodiment shown in FIG.

First, the checkpoint processing is shown in FIG.
This will be described below. Now, it is assumed that a checkpoint request is issued to the process with the process identifier "30" (step 6-1). After this, the processing contents up to step 6-4 are the same as the processing contents up to step 3-4 in FIG.

Thereafter, control is passed to the process identifier allowable range control means 9, which compares the process identifier reserved maximum number with the process identifier reserved current number, and the process identifier reserved current number still has the process identifier reserved maximum number. If it does not exceed (YES in decision 6-5), the process identifier reservation means 3 increments the relevant part of the process identifier "30" in the process identifier reservation table 105 by 1 (step 6-7). This completes the process identifier saving process in the checkpoint process,
Normal checkpoint processing is executed (step 6-
8).

Further, the process identifier allowable range control means 9
If the current number of reserved process identifiers exceeds the maximum number of reserved process identifiers (NO in decision 6-5), the upper 1 bit of the corresponding process identifier storage area is set in order to remember that the process identifier is not reserved this time. It is turned on and this information is written in the restart file 101 (step 6-8). In short, in this case, a checkpoint is taken without reserving the process identifier.

Next, the process of deleting the restart file 101 will be described with reference to FIG. In this case, the restart file deletion means 5 calls the process identifier reservation release control means 8 and the process identifier allowable range control means 9.

The process (step 7-1 to step 7-3) from the issuance of the request to delete the restart file 101 to the acquisition of the relevant process identifier in the restart file 101 is the same as the process described in FIG. Steps 5-1
This is the same as step 5-3). Thereafter, the process identifier allowable range control means 9 checks whether or not the upper 1 bit of the acquired process identifier storage area is ON,
If it is not ON, it means that this process identifier is reserved (YES in decision 7-4), so the value of the corresponding area is decremented by 1 (step 7-5).

If the value of the corresponding area becomes "0" (YES in decision 7-6), 1 is obtained from the current number of reserved process identifiers.
Decrement is performed (step 7-7). Finally, restart file deletion processing is executed (step 7-
8). With the above processing, it is possible to control the allowable range for performing checkpoint collection as a system.

Next, a third embodiment of the present invention shown in FIG. 3 will be described. In this embodiment, a process identifier dynamic control means 10 is added to the second embodiment shown in FIG. 2, and its flowchart is shown in FIG.

As a result of monitoring the state of the system, it was found that it is better to increase the number of target processes for taking checkpoints from the time of initial design. Therefore, the system is requested to change the maximum number of reserved process identifiers ( Step 8-1) in FIG. Upon receiving this request, the process identifier dynamic control means 10 confirms that the number of requests is larger than the maximum number of process identifier reservations (YE of decision 8-2).
S), and the maximum process identifier 10 in the entire system
It is confirmed that the number is within 4 (YES in decision 8-3), and the maximum number of process identifier reservations is changed to the request number (step 8-4).

With the above processing, the maximum number of reserved process identifiers in the system parameters can be changed according to the system operation mode without dropping the system, and the allowable checkpoint range of the system can be expanded.
Tuning that reduces the size is possible.

[0031]

As described above, the present invention has been
Even if the checkpoint succeeds, the process identifier that was attempted to be restarted at the time of the restart request was in use and the restart did not always succeed. It guarantees success.

In executing the program of this function, a plurality of restart files can be freely collected. This enables generation management of the restart file, and since deletion of the restart file and process release are linked in the system, there is no need to pay special attention to the process identifier.

From the above, the use of checkpoint / restart can be simplified and the system can be operated efficiently.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a block diagram showing a second embodiment of the present invention.

FIG. 3 is a block diagram showing a third embodiment of the present invention.

FIG. 4 is a process identifier reservation table 1 according to the present invention.
It is a block diagram of 05.

FIG. 5 is a flowchart of checkpoint processing in the first embodiment.

FIG. 6 is a flowchart of a process generation process according to the present invention.

FIG. 7 is a flowchart of restart file deletion processing according to the first embodiment.

FIG. 8 is a flowchart of checkpoint processing in the second embodiment.

FIG. 9 is a flowchart of a restart file deletion process according to the second embodiment.

FIG. 10 is a flowchart of a process identifier maximum reservation number changing process in the third embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 process generation means 2 checkpoint execution means 3 process identifier reservation means 4 restart execution means 5 restart file deletion means 6 process identifier monitoring means 7 process generation control means 8 process identifier reservation release control means 9 process identifier allowable range control means 10 process identifier dynamic control means 101 restart file 102 process control table

Claims (3)

(57) [Claims] 1. A process generating means for generating a process to be a CPU allocation target, a checkpoint executing means for executing a checkpoint process when a checkpoint request is accepted during process execution, and a process state at that time again. A restart file for freezing and storing in an executable state, a process identifier reservation table indicating whether or not the process identifier needs to be saved for restart processing, and a checkpoint executing unit at the time of a checkpoint. A checkpoint / restart system in a computer system having a process identifier reservation unit that is called to reserve a process identifier in the process identifier reservation table and a restart file deletion unit that deletes the restart file when it is no longer needed In the device, a process creation time control unit that performs a check for determining a process identifier to be assigned when the process is created, and a process identifier reservation release control unit that releases the reservation of the process identifier when the restart file is deleted. In the process identifier reservation table, when a checkpoint is executed once, the value of the area of the process identifier is incremented by 1. When the value is 1 or more, the process identifier is a normal process generation. If the restart file for the corresponding process identifier exists even if one of the restart files is deleted, the reservation of the corresponding process identifier is not released, and the process identifier of the process identifier is not restarted at restart. Control so that restart is not disabled due to duplication A checkpoint / restart control device characterized by: 2. A process identifier allowable range control means for controlling the number of reserved process identifiers in the entire system, a maximum number of process identifier reservations as a system parameter used at that time, and a checkpoint target as a system variable. The process identifier allowable range control means called from the checkpoint executing means at the time of the checkpoint request is added with the process identifier reservation current number for storing the number of process identifiers If is not reserved, it is confirmed whether the current number of process identifier reservations has reached the maximum number of process identifier reservations, and if not, there is no problem and it is a checkpoint permissible range as a system. Characterized by processing that enables checkpoints The checkpoint / restart control device according to claim 1. 3. A process identifier dynamic control means for changing a permissible range of a process identifier according to a system condition is added, and the process identifier dynamic control means adjusts the maximum number of reserved process identifiers to a system operating form. The system operator can change without dropping the system.
With the newly prepared command, the current reserved number of process identifiers and the maximum reserved number of process identifiers are referred to, the maximum reserved number of process identifiers is changed appropriately according to the system operation mode, and then the check is performed. 3. The checkpoint / restart control device according to claim 2, wherein the point execution means checks whether the system is within a checkpoint allowable range, and if the checkpoint is possible, normal processing is performed.