JPS6125250A - Fault recovery method of information processor - Google Patents

Abstract

PURPOSE:To secure the correct restart even with any type of a fault by performing the normalcy check before the information needed for restart is shunted. CONSTITUTION:An information processor contains a CPU101, a memory device 102, a file device 103 and a printer 104. The normalcy of the information processor is confirmed at a prescribed check point. Then the information needed for restart is stored in the device 103 when a system is normal. The state of the device 102 is reset at the check time point for restart bease on the latest contents shunted to the device 103.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to recovery processing in the event of a failure in an information processing device. In particular, the present invention relates to a checkpoint restart method applied to long-term processing.

(Prior Art) Conventionally, as a failure recovery method for this type of information processing device,
Various methods have been considered, and the checkpoint restart method is one of the most commonly used methods. The conventional checkpoint restart method divides a series of long-running processes into several short processes, and saves enough information after each process to restart from there. If a failure occurs during a certain process, the process can be restarted from the point where the previous process finished after the failure is removed (repaired). This is what I did.

However, there are some types of failures that can be recognized as soon as they occur based on the discovery process (for example, parity check errors, duplication comparison errors, etc., and those detected by the hardware's error detection circuit). When a failure occurs, processing continues without it being immediately recognized as a failure (such as failures that are not directly detected by the hardware error detection circuit).
There are cases where a failure is discovered much later due to some kind of inconsistency (for example, data has been corrupted due to control).

[Problem that the invention seeks to solve]

In the conventional checkpoint rescoot method described above, in response to a failure during a certain process, the process is returned to the point at which the previous process was completed, but the previous process may not have been processed normally, so this In some cases, restarting the process may result in incorrect processing results.

SUMMARY OF THE INVENTION The present invention aims to solve this drawback and to provide a fault recovery method for an information processing apparatus that can correctly recover from any type of fault.

[Means for solving problems]

In the present invention, after each divided process is finished and before saving the information necessary for restarting, it is checked whether the process was performed normally, and the process returns to this confirmed state and restarts from there. characterized in that it is configured to perform.

That is, the present invention, in an information processing device consisting of a central processing unit (CPU), a memory device, a file device, etc., checks the normality of the information processing device at a predetermined checkpoint, and restarts the system if the system is normal. This is a method of saving the necessary information to a file device, and using the latest contents saved to this file device to return the state of the memory device to the checkpoint point.This method prevents outages due to unpredictable causes. The feature is that it prevents the problem from occurring and enables a correct restart.

゛ [Operation] When each divided process is completed, the present invention checks whether the process was performed normally, and if it is normal, restarts by saving the information necessary for restarting to the file device, Eliminate troubles such as stoppages due to unpredictable causes.

〔Example〕

Next, the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 shows the equipment configuration of an information processing system to which the present invention is applied, in which a memory device 102 is connected to a file device 103 and a printing device 104 via a central processing unit 101. The file storage N103 stores data and programs to be processed, and is also used as a work area for storing data being processed by the program, and also as an evacuation area. The programs and data stored in this file device 103 are loaded into the memory device 102, processed by the central processing unit CP UIOI, and the results are output to the printing device 104. The flowcharts of FIGS. 2 and 3 provide processing flows that allow the process to be restarted correctly even if a failure occurs during the process, and will be described below.

First, the processing program is divided into processes 1 to n, and after each divided process, it is checked whether the process was performed correctly or not. If it is determined that the process was performed correctly, the information necessary for restarting is saved. Make sure to include 2nd below
The explanation will be given according to the flowcharts shown in FIGS.

When program execution starts (200), it is judged in the judgment box (box) 201 in FIG. , execute from process 1 in box (205),
If it has already been executed halfway (P≠0),
Box 202) for restoring the information necessary for restarting, Branching to the restart point (Pi) (box 203), Box 205
Thereafter, when each of the aforementioned divided processes 1 (i=1 to n) is executed, a checkpoint setting macro (CHPT) is called. As can be seen from the flowchart in Figure 3, the function of the CHPT macro is to first perform a health check to determine whether the processing up to that point was correct (box 301), and if it is normal, set the failure recovery start address. Registration of A (in this embodiment, it is the same as the program start), registration of restart address (also called checkpoint) Pi, and other information α(i), β(i) required for restarting.
, γ(i) (information on the memory that should be carried over to processing i+l and later) is saved (box 303), and if an abnormality is found in the judgment box 302, trouble handling is performed such as notifying the operator of the fault. (Box 310).

Therefore, after process 2 (box 207), the restart address and other restart necessary information are saved in areas P and SF, respectively. Areas P and SF are provided in the file device 103 shown in FIG. 1, and their information remains even if the power is turned off for repairs or the like. When the last process n is executed (box 220), all processes are completed, and if they are to be performed again, they will be performed from the beginning, so a clearing operation of 0→P (box 221) is performed and all processes are completed.

Next, for example, if a failure occurs during execution of process 2 (box 207), the power is once turned off and repaired, and then the power is turned on to start up the system and return to the program start (termination 200) point. At the starting point of the program, after determining P≠0 as described above, the area (S F) returns to the state of α, β, γ immediately after execution of process 1 and branches to P, as if it were in the middle of execution of process 2. Processing continues as if there was no failure. Also, even if a failure occurs during the execution of process 2 and the process continues without realizing it,
In 08, call the CHP'T macro and check the normality (
Since the failure is found by executing box 301), information for restarting checkpoint P2 will not be erroneously saved. Therefore, even in this case, the process will be correctly restarted from checkpoint P+ after repair.

Note that the results of the normality check are constantly stored in the memory device 2102 using a hardware test diagnosis program.

〔Effect of the invention〕

As described above, the present invention is configured to perform a normality check before saving the information necessary for restarting, thereby making it possible to restart correctly regardless of the type of failure. .

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an apparatus for an embodiment method of the present invention. FIG. 2 is a flowchart showing the processing flow of an embodiment of the present invention. FIG. 3 is a flowchart showing another processing flow of the embodiment of the present invention. 101... central processing unit ('CPU), 102...
...Memory device, 103...File device, 104.
...Printing device, A...Failure recovery start address, Pi.
・・Restart address (checkpoint), α(i
), β(i), r(i)... restart required information,
P, SF...Area, CHPT...Checkpoint setting macro, 'R3...Restart.

Claims (1)

[Claims] (1) After dividing a series of program processing into multiple short processes and saving and accumulating the necessary information so that each short process can be restarted from the state where it ended, The above series of program processing is executed so as to proceed to the next short processing mentioned above. If a failure occurs while executing this program processing, after removing the failure, A failure recovery method for an information processing device that restarts the execution of a program process from a state where the short process executed before the process has finished, using the necessary information saved and accumulated, When the execution of each short process is finished, confirm that the process was executed normally, return to the above short process where this confirmation was performed, and return from the state where the execution has finished. A failure recovery method for an information processing device, characterized by restarting the information processing device.