JPS6010328A - Input and output error demarcating and processing system - Google Patents

Abstract

PURPOSE:To save man-hours and the time for diagnosis by allowing the system to access other device and then a maintenance area of a faulty device when an input/output error occurs in one device accessed directly to automate the demarcation of the error. CONSTITUTION:If an input/output error occurs in any of magnetic disc devices 3-1-3-3 connected under the control of a channel device 2, e.g., the 3-1, the other magnetic disc devices 3-2 and 3-3 are accessed with a command of the identical content, the result is informed to a channel system error identification section 8, and whether or not all the devices are in error is judged and if in error, it is recognized that the error is an error of a channel system. When the accessing is normal, a maintenance cylinder 4-1 of the magnetic disc device 3-1 is accessed, the result is informed to a magnetic disc/medium error identification section 10, and when the same error occurs, it is judged as an error of a device system and if not, it is judged as an error of medium system.

Description

Detailed Description of the Invention (8) Technical Field of the Invention The present invention provides an input/output fault isolation processing method, particularly when an input/output fault occurs in a direct access device such as a magnetic disk device (2. I/O fault isolation, which can be detected visually, is related to the processing method.

■ Technology background and issues Improving the reliability of data processing systems (= systems) is an important issue, but at the same time, when a failure occurs, it is important to quickly identify the cause of the failure and the location of the failure. It is desirable to be clear so that the problem can be addressed as early as possible.

Direct access devices such as magnetic disk devices (DASD-)
is the most commonly used external storage device. For example, if an input/output command to a magnetic disk device terminates abnormally, what are the causes of the failure?

Channel device, drive device, drive control device, head.

There are nine possible causes, including abnormalities in the medium and the power supply system of the system. Depending on the cause of the failure, the countermeasures for the failure will differ.For example, if the cause is caused solely by the medium, such as a scratch on the medium, you may need to reinstall a new medium, allocate a spare track, back up the information, etc. You just have to restore it based on (= restore it. Again.

If the fault is in the head, drive unit, etc., either disconnect the faulty device from the system and reinstall the medium in another normal empty device and use it, or if there are two faulty devices C, repair it immediately. (Conventionally, in order to identify the above-mentioned fault location), maintenance personnel use the sense information outputted by the system to perform trial-and-error procedures, for example, using fault tracing procedures determined by experience. In the past, efforts were being made to investigate the cause of the failure. Therefore, there are problems in that it takes time to identify the faulty part, and that erroneous judgments may occur due to operational errors or the like.

(Q. Purpose and Structure of the Invention The present invention aims to solve the above-mentioned problems, automates the isolation of faults, makes it possible to save labor and time for diagnosis, and improves the reliability of the results in fault isolation. The purpose is to improve. Therefore.

By using multiple devices to isolate faults and performing special access, we can compare and analyze their error information and sense information, thereby narrowing down the focus and pointing out the location of the fault.

In other words, what is the input/output fault isolation processing method of the present invention?

Input/output fault isolation for identifying the failure location of an input/output fault for a direct access device in a data processing system is performed in a processing method by providing a maintenance area for test access in advance (2. a first access means for accessing another direct access device under the same channel device as the direct access device when an input/output failure occurs;
a channel system fault isolation means for determining a channel system fault based on normality or abnormality of access by the access means; a second access means for accessing the maintenance area of the direct access device in which the fault has occurred; The present invention is characterized in that a device/medium fault isolation means is provided for determining whether the fault is a media-related fault or a device-related fault based on whether the access is normal or abnormal. Embodiments will be described below with reference to the drawings.

0 Embodiment of the Invention FIG. 1 shows the configuration of an embodiment of the invention, and FIG. 2 shows a process explanatory diagram of the embodiment of the invention.

In the figure, ] is the main system device, 2 is the channel device, 3-1 to 3-3 are the magnetic disk devices, 4-1 to 4-3 are the maintenance cylinders, 5 is the fault detection section, and 6 is the fault identification control section. , 7 is another device access section, 8 is a channel system fault identification section, 9 is a maintenance area access section, 10
1 is a device/medium failure identification unit, 11 is a failure data editing unit, and 1 is a failure data editing unit.
2 represents a display device.

The main system device 1 is composed of a central processing unit, a main memory device, etc., and is a device that sequentially fetches and executes instructions prepared on the main memory device. The channel device 2 is a device that controls information transfer between the nine-core system device 1 and the magnetic disk devices 3-1 to 3-3.

Typically, a plurality of devices of the same type are connected to the channel device 2. Magnetic disk device 3-1
3-3 are so-called direct access devices that can directly access any storage area, and the address of the storage area is 1, for example, a cylinder number.

It is given by head number, record number, etc.

Many storage media, that is, volumes, are removable. Compared to those in which the media is fixed and cannot be attached or detached (1), it is possible to quickly distinguish between device failures such as drive parts and media failures due to scratches (2). is large.If it is a simple media failure, use can be continued without waiting for the device to be replaced or repaired.Magnetic disk devices 3-1 to 3-3
Maintenance cylinders 4-1 to 4-3 are prepared in advance for volumes to be mounted. The maintenance cylinders 4-1 to 4-3 are storage areas provided for writing or reading for maintenance and diagnosis.
Therefore, this area is designated not to be used in normal business data processing. Therefore, any test or
Even if patterns are written, data necessary for business operations will not be destroyed.

The failure detection unit 5 is a part of the input/output management unit of the operating system, and detects the occurrence of a failure based on the condition code of the input/output activation instruction or the information notified from the device side by the input/output interrupt. The cause of the failure can be determined to some extent by the error status information reported to the central processing unit as a so-called sense byte.However, the sense byte is not always effective. No. In the case of the present invention, as described below (2), input/output for diagnosis is performed on the fR to narrow down the range in which a fault may exist.

When the failure occurrence detection unit 5 detects an input/output error, it activates the failure identification control unit 6.

The fault identification control section 6 is activated by the fault detection section 5 and controls the activation of each of the following processing sections for diagnosis. First, the other device access section 7 is activated. Other device access section 7.

For example, when a failure occurs in the magnetic disk device 3-1, other magnetic disk devices 3-2 to 3-3 connected under the same channel device 2 are accessed.

The channel command used at this time has almost the same content as the command that encountered the failure. however,
That command.

In the case of a (WRITB) type command that involves data writing, the maintenance cylinder 4-2.4-3 is selected as the access area in order to prevent normal data from being destroyed by writing. These access results are notified to the channel system fault identification section 8.

The channel system fault identification unit 8 is connected to the other device access unit 7.
Based on the access results of If this also occurs in devices 4-2 and 4-3, it is recognized that the common cause is a channel system failure. The recognition results are reported to the failure data editing section 11. Channel failures cannot be diagnosed in detail by external access by a channel program, so it is necessary to start a microprogram on the channel side. In this regard, detailed information can be collected by performing conventional channel diagnosis.

If the fault is not in the channel system, the maintenance area access unit 9 is called. Maintenance area access section 9
This is to check whether the failure that has occurred in the primary area of the magnetic disk device 3-1 also occurs in the maintenance cylinder 4-1 by executing a command under the same conditions. The access result is notified to the device/medium failure identification section 10.
teeth.

If the same failure occurs in the maintenance cylinder 4-1, it is determined that it is a device-related failure; otherwise, it is determined that it is a media-related failure. The determination result and status information based on the access are reported to the failure data editing unit 11. In the case of a device system failure, it is possible to further isolate the head failure by accessing by switching heads. In case of failure other than the head (
= analyzes the failure location and obtains detailed information by comparing the status of other devices. In case of media failure.

Access to the vicinity of the failed address, retry of recalibration plate/seek, etc. Detailed fault isolation is possible by replacing write commands with read commands and accessing the faulty address.

The fault data editing unit 11 edits the reported fault severity information and fault status information and displays them on the display device 12 of the system console.

The operator (2) reports the results, codes and summarizes the diagnostic results, and logs them as maintenance information.

Next, referring to FIG. 2, an overview of the processing of the embodiment shown in FIG. 1 will be explained. When a failure occurs, first the process 20 shown in FIG.
With the same command.

Access other devices. If multiple devices of the same type are connected under the same channel device, access each one. Process 21 (2) Determine whether all accesses have become abnormal. If all accesses have become abnormal, process 22 recognizes that it is a channel system failure. and can be accessed from a different access path.

You can also use that access information.

Process 20 (If the access by 2 is normal, the channel device is considered to be normal. In that case, process 2 is performed to distinguish between a device system failure and a media system failure.
3 to access the maintenance area of the faulty device. A command with the same content as the command that encountered the problem, or
Check the occurrence of failures by performing write tests with unique patterns that increase the probability of errors. In process 24, it is determined whether everything has ended normally.

If it is normal, processing 25 recognizes that there is a media system failure. Also, if a similar abnormality occurs.

Through process 26, it is recognized that the failure is in the device system. These recognition results will be edited and output as a message. Therefore, upon receiving the above-mentioned message report from the operator, the maintenance personnel can immediately prepare and bring the parts necessary for repair and perform replacement/repair. That is, it automates a series of processes from the occurrence of a fault, starting nine diagnostic processes, segregating the details of the fault, and outputting a detailed message about the fault.

This makes it possible to simplify maintenance procedures, shorten system repair time, and speed up factory testing.

(G) Detailed Description of the Invention As described above, according to the present invention, failures can be isolated down to the smallest detail without any intervention, and the time required for maintenance can be significantly reduced. Furthermore, since classification is performed based on actual access results, the accuracy of identification is also good. In particular, it is of great significance in creating patterns for maintenance and testing systems, which have varied until now.

[Brief explanation of drawings]

FIG. 1 shows the configuration of an embodiment of the present invention, and FIG. 2 shows a processing explanatory diagram of an embodiment of the present invention. In the figure, 1 is the main system device, 2 is the channel device, 3-1 to 3
-3 is a magnetic disk device, 4-1 to 4-3 are maintenance cylinders, and 5 is a failure detection unit. 6 is a fault identification control unit, 7 is another device access unit, 8
9 is a maintenance area access unit; and 10 is a device/medium failure identification unit. Jl represents a fault data editing section, and 12 represents a display device. Patent applicant Fujitsu Limited

Claims (1)

[Claims] Input/output failure isolation for identifying the location of an input/output failure for a direct access device in a data processing system is a processing method (=) in which a maintenance area for test access is provided in advance on the medium of the direct access device, and
When an input/output failure occurs in a direct access device, a first access means accesses another direct access device under the same channel device as the direct access device, and a channel system failure is detected depending on whether the access by the first access means is normal or abnormal. Channel system fault isolation is determined by means,
a second access means for accessing the maintenance area of the faulty direct access device; and a device/medium for determining whether it is a media system failure or a device system failure based on normality or abnormality of access by the second access means. An input/output fault isolation processing method characterized by having a fault isolation means.