JPS63136142A - Error recovery system for logical unit - Google Patents

Abstract

PURPOSE:To reduce danger that causes a system breakdown or job-aboard state, etc., by using another normal logical device to perform retrial of an instruction that is turned into an error in case such an error that can be retried and also cannot be retried is produced or such a fixed error that can be retried is produced. CONSTITUTION:When a logical unit 11 has a trouble, a monitor means 14 interrupts the processing of the unit 11 and stores the internal state of the unit 11. A diagnosis processor 2 judges an error detected out of the read-out internal state through a retrial possibility judging means 22. If an error that can be retired and also cannot be retried is judged or a fixed error that can be retried is judged, the succession information is complied and produced from the read-out internal state by a succession information compiling/producing means 23 and then set to a normal logical unit 13 by a succession information setting means 24. Then a succession processing re-execution/succession instructing means 25 gives a instruction to the unit 13 for succession of the processing. Then the unit 13 re-executes and succeeds the processing.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field The present invention relates to an error recovery method for a logical device, and particularly to a logic system for an information processing system in which a plurality of logical devices share a main storage device. Regarding error recovery methods for equipment.

Prior Art Conventionally, this type of error recovery system has been disclosed, for example, in Japanese Patent Application Laid-Open No. 57-064849 (Japanese Patent Application No. 55-141323).
The method shown in No. 1) is known.

In the conventional method described above, when an error occurs in a certain logical device (central FFI logical device) and it is determined that the instruction that caused the error can be retried due to memory rewriting conditions, etc., the logical device that caused the error is first If the error is not recovered even after a predetermined number of retries, the instruction is configured to be retried using another normal logical device.

In this error recovery method, if the cause of the error is an intermittent failure and retry is possible, and if the retry is successful on the logical device where the error occurred, system failure can be avoided, and if the cause of the error is due to a fixed failure. If it is possible to retry even if the logical device is a logical device, a system failure or the like can be avoided by successfully retrying with another normal logical device.

However, even if there is an intermittent failure, retrying is not always possible; retrying may not be possible depending on the memory rewriting timing during instruction execution. In addition, since it is thought that a logic device that has experienced an intermittent failure once has a high probability of causing an intermittent failure again, this method of first performing a retry on the logical device where the error occurred will result in an intermittent failure occurring again during the retry. An error may occur and the instruction in error may not be retriable. In such a case, since retrying is not possible, there is a problem that the system may go down or the job may be aborted.

This error recovery method is revised to 7r, L, Tachi, and when an error occurs in a logical device and it is determined that the error instruction can be retried, the error is retried in the logical device where the error occurred. Instead, a method is available in which another normal logical device is made to retry the error instruction. In this method, for safety reasons, if a failure occurs even once, the faulty device is disconnected, so retries are made at 100%, such as in the case of a fault in advance control.
Even for possible failures (failures that cannot be retried), they are removed without retrying the logical device in which the error occurred. Therefore, there is a drawback that the performance of the system is unnecessarily degraded.

Purpose of the Invention The present invention solves these problems, and its purpose is to reduce the risk of system downtime, job aborts, etc., while minimizing performance degradation to the logical device. An object of the present invention is to provide an error recovery method capable of performing error recovery.

According to the present invention, there is provided an error recovery method for a logical device in a J5 information processing system in which a plurality of logical devices share a main storage device, in which when an error occurs in the logical device, the command that caused the error is recovered. When the above error is determined to be retriable and this error is determined to be an error that cannot be retried, or if it is determined to be a retryable error but the retries have failed several times, the above error is generated. According to the present invention, there is provided an error recovery method characterized in that, without retrying the instruction in the errored logic device, other normal logic devices are made to continue retrying the instruction that caused the error.

Example Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 is a block diagram of an embodiment of the invention.

In the figure, 1 is an information processing device, and 2 is a diagnostic device.

3 is a diagnostic interface for connecting the information processing device 1 and the diagnostic device 2;

The information processing device 1 includes a plurality of logical devices 11.13, which are, for example, central processing units, and a main storage device 12 that is connected to the plurality of logical devices 11.13 and accessible from the two-way processing device.
and a monitoring means 14 connected to a logical device 11.13 and a main memory 12, one operating
controlled by the system. The monitoring means 14 has a function of detecting a failure (error) in the logical device 11.13, and when a failure is detected, saves the internal state of the failed logical device and transmits it to the diagnostic processing device 2 via the diagnostic interface 3. It also has a function to notify you of this.

Note that although Figure 1 shows the case where there are two logical devices,
The present invention is also applicable to information processing systems in which three or more logical devices work together.

Further, the diagnostic processing device 2 includes an internal state reading means 21,
It includes a retry possibility determination means 22, a handover information editing/creation means 23, a handover information setting means 24, and a first stage 25 for instructing re-execution/continuation of the handover process. this house,
The internal state reading means 21 is a means for reading out the internal state of the faulty logic device held by the monitoring means 14 when the above-mentioned notification is received from the monitoring means 14. The retry possibility determining means 22 determines the possibility of retrying an instruction that has caused an error based on the read internal state, and determines whether the error is a fixed error (retry). This is a means of determining errors in cases where several attempts have failed.

The handover information editing/creation means 23 detects a failure when the retry possibility determination means 22 determines that the error is retryable and cannot be retried, or when it is determined that the error is retryable and fixed. This is a means for editing and creating handover information for the process that was being executed on the logical device from the internal state. The handover information setting means 24 is means for setting the handover information created by the handover information editing/creation means 23 to a normal logical device other than the logical device in which the failure has been detected. Further, the re-execution/continuation instruction means 25 for instructing the re-execution/continuation of the takeover process executes the process that was being performed on the logical device in which the failure was detected on the normal logical device based on the information set by the takeover information setting means 24. This is a means of instructing that the process be taken over and re-executed.

Next, the operation of this embodiment will be explained using the flowchart of FIG. 2, taking as an example the case where a failure occurs in the logic device 11 in FIG.

When a failure occurs in the logical device 11, the monitoring means 14
Detected in When the monitoring means 14 detects this, it interrupts the processing of the logical device 11, saves its internal state, and notifies the diagnostic processing V7L unit 2 via the diagnostic interface 3 that the logical device 11 has failed.

In response to this, the diagnostic processing device 2 activates the internal state reading means 21, and uses this means 21 to read out the internal state of the logic device 11 saved by the monitoring means 14, as shown in process 51 in FIG. . Next, the diagnostic processing device 2 uses the retry possibility determining means 22 to determine whether the error detected from the read internal state is an error that can be retried and cannot be retried, or an error that can be retried and is fixed. If the condition is not satisfied, failure processing is performed for the logical device 11 in which the error has occurred. This obstacle is 91! [9! In step , if the error is one that can be retried, failure processing for the failed logical device 11, that is, retry processing is performed, and if the retry is successful, operation continues. Otherwise, the error occurs in logical unit 1.
1 system color separation is performed.

On the other hand, in process 52, retry is possible and retry is not possible, and 1q
If the error is determined to be a retryable and fixed error (an error that occurs after several failed retries), the takeover information is extracted from the internal state read by the takeover information editing/creation means 23. is edited and created (process 53), and this inheritance information is set for the normal logical device 13 by the setting means 24 (process 54). In the judgment made in this process 52, an "error that can be retried but cannot be retried" means that an error that occurs at a certain time can be retried because the content of the command is different depending on the difference in the time of occurrence of the error, but an error that occurs at another time can be retried. This means that if an error occurs, it may not be possible to try again. [Retryable and fixed error 4] means an error that can be retried and has failed several times as a result of retrying. The possibility of retrying such an error is determined based on the contents of the error flip-flop box indicating the error occurrence state, and a known technique may be used.

Next, the instructing means 25 for re-executing/continuing the takeover process instructs the normal logical device 13 to take over the process 55.
), causing the logical device 13 to re-execute and continue the inherited processing. As a result, the error processing that was being executed on the logical device in which the retryable failure was detected is re-executed not on the logical device in which the failure was detected, but on another normal logical device, and the logical '1tall Recovery from the error that occurred is performed, and subsequent processing is taken over by a normal logical device instead of the failed logical device. When a normal logical device takes over the processing of a failed logical device, the failed logical device is usually logically disconnected from the system.

The above explanation is for the case where the logical device 11 fails, but
Even if the logical device 13 fails, the processing is taken over to the normal logical device @11 after the υ control of the diagnostic processing device 2 as described above, and the processing that was being performed on the logical device 13 is continued. be able to.

DETAILED DESCRIPTION OF THE INVENTION According to the present invention, when an error that can be retried and cannot be retried, or an error that is retriable and fixed, the retry of the erroneous instruction is performed without an error. without running on the logical device where the
It is executed by another normal logical device, and if an error occurs due to an intermittent failure during retry, the error instruction cannot be retried, resulting in a system down or abort. can be done, and 100
% In the case of a retryable error, the logical device in which the error has occurred is not disconnected and continues to operate by retrying itself, which has the effect of suppressing performance degradation to the utmost.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a flowchart h showing a processing example of the diagnostic processing device in the blocks of FIG. Explanation of symbols of main parts 11.13...Logic device 12...Main storage device

Claims (1)

[Claims] An error recovery method for a logical device in an information processing system in which multiple logical devices share a main memory, in which when an error occurs in the logical device, the instruction in error can be retried and the error can be retried. When an error is determined to be a non-tryable error, or an error is determined to be a retryable error for which several retries have failed, an error is detected without retrying on the logical device that generated the error. An error recovery method characterized in that the normal logical device of the computer is made to continue retrying the instruction that caused the error.