JPS6220037A - Command retry system for information processing system - Google Patents

Abstract

PURPOSE:To perform retry as a host system expects, by issuing a command to which information indicating that it is a retry command is added if a command issued from the host system results in unsuccess. CONSTITUTION:When detecting the occurrence of the trouble that the end status is not returned from a subordinate system 20 within a prescribed time for the command issued from a command issuing part 110 of a higher system 10, the command is issued again from a command reissuing part 130. In this case, information indicating that this command should be retried is added by a retry indication adding part 120. The retry command is applied to a retry discriminating part 210 of the higher system 20 and is identified as the retry command, and the control is transferred to a retry processing part 220. The retry processing part 220 executes the command with a logic different from that of a command processing part 240 is consideration of contents of the received command and the retry command and returns the end status to the host system 10.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to failure handling in an information processing system, and in particular, when a command given from a higher-level system to a lower-level system fails due to some kind of failure, This invention relates to a command retry method that recovers from failures by issuing commands.

[Conventional technology]

Conventionally, this type of command retry method has been: ∎ A method in which the same command is simply reissued. ∎ The state before the command is issued is saved in the higher-level system, and when a failure occurs, the lower-level system is redirected from the higher-level system based on the saved information. A method in which a command is issued to return the system to the state before the command was issued, and then the same command is reissued. [Problems to be solved by the invention] Method When a command is received, if the lower system is in state A, for example, if the lower system is in state A, the command is sent from the higher system. This is effective if a failure occurs before issuing a command and executing the command, that is, before changing state BA to state B, but it is effective when the lower system executes the command and changes state A to state GH. If a failure occurs thereafter, the problem is that if the higher-level system makes a retry, the lower-level system will move to state C, meaning that there is no point in retrying.

In this respect, method ① is that if a failure occurs after the lower system changes status BA to status B, status BB can be returned to status BA by a command from the higher system, and the next retry command will return the status as expected. It is possible to change the shape LiA to the shape LiB. However, according to this method, it is necessary to save the state in the upper system, and it is also necessary to issue a command to restore the state to the original state separately from the retry command, which reduces processing efficiency. There is. Also, method ■
On the contrary, if a failure occurs before state 8A is changed to state B, the command to return the state to the state before the command was issued will be interpreted by the lower system as a meaningless command, and it will be treated as if an illegal command had been issued. There are also issues to be addressed.

The present invention solves these conventional problems.
The purpose of this is to avoid the need to save the state before the command was issued in the upper system, or to issue a command to return the state of the lower system to the state before the command was issued, and to regenerate the system as expected by the upper system. The purpose is to provide a method that can be tried.

[Means for solving problems]

If a command issued by a higher system to a lower system fails due to a failure, it is quite difficult for the higher system to correctly recognize at what stage the command failed, but the lower system executing the command It can be easily identified from the side. Therefore, it is better to leave the details of the processing to be performed at the time of retry to the lower system side.

The present invention has been made with attention to these points, and includes a command retry method for an information processing system that includes a higher-level system and a lower-level system that executes commands from the higher-level system. a command issuing processing means for issuing a command; a retry display adding means for adding information to the command indicating that a retry is to be performed when the command issued by the command issuing processing means is unsuccessful due to a failure; and command reissue means for reissuing the command to which information indicating that it is a retry has been added by the trial display adding means, and the lower system is configured to identify the retry display information in the command received from the upper system. The present invention is provided with a retry determining means for determining whether a command is a retry command or not, and a retry processing means for processing a command determined to be a retry command by the retry determining means.

[Effect]

If a command issued by the command issuing processing means of the upper system is unsuccessful, a command with information indicating that it is a retry command is issued from the command reissuing means, and the lower system processes the retry command in the command. By identifying the trial display information, it is fully determined that the command is a retry command, and the retry processing means performs retry processing in accordance with the command and the state of the lower system.

〔Example〕

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention. The host system 10 includes a command issuing section 110, a retry display adding section 120, and a command reissuing section 130. It is connected to the input/output control device 30. The lower system 20 also includes a retry determining section 210, a retry processing section 220, a state storage section 230, and a command processing section 24.
0, and is connected to the input/output control device 30 via the device ink face 50.

Commands from the host system 10 are issued by the command issuing unit 110, and are sent to the channel 40. Input/output side? ■Device 30. The information is sent to the lower system 20 via the device ink face 50 and input to the retry determination unit 210 of the lower system 20. Retry determination section 210 determines from the retry display included in the command that the command is a command other than retry, and passes control along with the received command to command processing section 240 . The command processing section 240 decodes the contents of the command and performs processing such as rewriting the contents of the state storage section 230 according to the contents of the command. When the command processing unit 240 completes a predetermined process, the completion status is displayed as the device ink face 50. Input/output side? 1
1 device 113o.

It is sent back to the host system 10 via the channel 40. When the host system 10 receives the termination status, it determines its contents and, if it has terminated normally, executes the next operation. Further, when it is detected that a failure has occurred, for example, the end status is not returned within a predetermined time, the command reissue unit 130 issues the command again. At this time, the retry display adding section 120 adds information indicating that the command requires a retry.

The retry command issued by the command reissue unit 130 is sent to channel 40. Input/output control device 30. Device interface 50
The command is added to the retry determining section 210 of the lower system 20 via the , where it is identified as a retry command and control is passed to the retry processing section 220 .

The retry processing unit 220 considers the contents of the received command and that the command is a retry command.
The command is executed using logic different from 0. That is, for example,
If the process corresponding to the received command has already been executed, only the process of including the result in the end status and returning it to the higher-level system 10 is performed, and if the process corresponding to the received command has not been executed yet, the process After executing the process, the process includes the result in the end status and sends it back to the host system 10.

The above completion status is device ink face 50. Input/output control device 30. Upper system 1 via channel 40
Returned to 0.

In the system configured as described above, various commands are issued from the upper system 10 to the lower system 20.
) command as an example. This lock block command is a command issued from the higher system 10 to the lower system 20 when a certain task exclusively uses a certain block (storage area) of the lower system 20, and when this command is issued, If the block specified by the command is not used by any task, the lower system 20 makes it available for use by the task that issued the command, and notifies the higher system 1 to that effect.
0 and notifies the higher-level system 10 with the completion status of the command that it cannot be used if another task is using it.

The lock block command as described above includes, for example, a command code, a flag, a data size, a task name, a file name, a block name, etc., as shown in FIG. Here, the difference from the conventional method is that, for example, one bit in the flag is assigned to a retry display bit that indicates whether the command in question is a retry command.

When the lower system 20 receives a command that is not a retry command as shown in FIG. a block table 232 including a block name, a task link field, and a file link field; and a file table 233 including a link field and a file name.
is created in the state storage unit 230, and if the block specified by the command is not used by another task, write the address of the block table 232 in the link field of the task table 231 and the link field of the file table 233 to link it. , and registers that the block described in the block table 232 is used by the task that issued the command. Furthermore, when the block specified by another task is being used exclusively, this fact is notified to the higher-level system 10 by means of the termination status.

On the other hand, when the lower system 20 receives a command as shown in FIG. 2, which is a retry command, the lower system 20
0 retry processing unit 220 leaves the registration status as it is when there is already a registration in the status storage unit 230 in which the same block is used by the same task, and indicates in the upper end status that the block specified by the command can be used. The system 10 is notified, and if there is no registration in the status storage unit 230 that the same block is used by the same task, the same process as the command processing unit 240 is performed.

Now, in FIG. 1, when a lock block command is issued from the command issuing unit 110 of the host system 10, this command is sent to the channel 40. It is sent to the lower system 20 via the input/output control device 30 and the device ink face 50.

The lower system 20 determines whether the block specified by the command is not used by any other task, and if it is not used, the lower system 20 stores the block in the state storage unit 230 so that the block specified by the command is used by the task that issued the command. Register that fact. As described above, this registration method is performed by creating a task table 231 and a file table 233, creating a block table 232 that describes the used blocks, and creating a link.

After registering in the status storage unit 230 as described above, the lower system 20 notifies the upper system 10 that the specified block can be used using the command completion status. At this time, if the termination notification cannot be notified to the higher-level system 10 due to a failure in the channel 40, for example, the higher-level system 1
0 means that the command was not completed normally, so the command is retried. For this retry, the retry display addition unit 120 adds a retry information bit in the flag of the command shown in FIG. 2, and the command reissue unit 130 sends the command to the lower system 20. This is done by reissuing it.

When the lower system 20 receives this command, the retry determining unit 210 checks the retry information bit in the command, and since a retry is displayed, the retry processing unit 22
Transfer control to 0. The retry processing unit 220 includes the state storage unit 2
Since there is already a description in 30 that the same block is used in the same task, the status is left as is and the host system 10 is notified by the end status that the block specified by the command can be used. Note that if you retry as in the conventional method (■) without adding a retry indication to the command, the lower system 20 will determine that it is a desodorofter because there is already a description of using the same block in the same task. However, the termination status indicating fraud will be notified to the higher-level system 10, and the higher-level system 10 will have to cancel the job, resubmit the job, analyze the cause, etc., and the system efficiency will be significantly reduced. Become.

〔Effect of the invention〕

As explained above, the present invention provides the following advantages when reissuing a directive:
Information indicating whether the command is a retry command is added and sent from the upper system to the lower system, so the lower system can identify whether the command requires a retry or not. , retry processing can be performed in the lower system that knows the best way to complete processing in the command where the failure occurred, and retry processing can be performed as expected by the upper system. becomes possible.

Furthermore, according to the present invention, there is no need to save the state before the command is issued in the upper system as in method (2), and a special command is issued to return the state of the lower system to the state before the command was issued. This is no longer necessary, and processing efficiency can be improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention, FIG. 2 is an explanatory diagram of the contents of a command, and FIG. 3 is an explanatory diagram of the contents stored in the state storage section 230. In the figure, 10 is the upper system, 20 is the lower system, 30 is the input/output control device, 40 is the channel, 50 is the device ink face, 110 is the command issuing section, 120 is the retry display adding section, and 130 is the command reissuing section. , 210 is a retry determination unit, 220 is a retry processing unit, 230 is a state storage unit, 2
40 is a command processing section. A block diagram showing the overall configuration of an embodiment of the present invention. An explanatory diagram of the contents of the command. Fig. 2

Claims (1)

[Scope of Claims] A command retry method for an information processing system including a higher-level system and a lower-level system that executes a command from the higher-level system, comprising: a command issuing processing means for issuing a command to the higher-level system; retry display adding means for adding information to the command indicating that the command is to be retried when the command issued by the issuing processing means is unsuccessful due to a failure; a command reissuing means for reissuing a command with information added thereto; the lower system identifies retry display information in the command received from the upper system to determine whether the command is a retry command or not; 1. A command retry method for an information processing system, comprising: a retry determining means for processing a command determined to be a retry command by the retry determining means; and a retry processing means for processing a command determined to be a retry command by the retry determining means.