JPS6225318A - Automatic bus blocking system - Google Patents

Abstract

PURPOSE:To shorten the execution processing time of input/output instructions by blocking a trouble bus automatically by a disk controller on the basis of specific condition and returning unusable state information for input/output instructions from a host device. CONSTITUTION:If an error detecting circuit detects an error when data are transferred between a low-priority device and respective devices in a magnetic disk device with input/output instructions from the host device, an MPU decides whether bus trouble occurs or not. When the bus trouble is decided, counted values in corresponding device areas #0-#7 of a bus trouble occurrence frequency calculation table corresponding to the bus in the data transfer processing are summed up. Then, if there is a table area which exceeds a prescribed frequency, the magnetic disk device blocks the bus automatically and when a next input/output instruction is generated through the bus, unusable state information is sent back. The host device on receiving the information is allowed to sends and input/output instruction to the device from another bus.

Description

[Detailed description of the invention] [Already required] In the path automatic blockage method, when a path failure occurs in the disk subsystem, the disk control device automatically closes the failed path based on predetermined conditions, and the host device The execution time of the input/output command can be reduced by returning unusable state information in response to the input/output command from the host device, and having it executed on the target device immediately by accessing from another path using the software MJI-Ly from the host device. Shortening is possible.

[Industrial application field]

The present invention relates to a path automatic closing method that automatically closes a corresponding path when a path failure occurs due to a failure of a disk control device or a disk device. ``As the use of information processing systems has become more widespread and the methods of use have become more sophisticated and complex, the demand for higher processing speeds has become stronger. In particular, as a countermeasure for solving these problems, it is more important to realize processing efficiency in a disk subsystem whose processing speed is relatively slow.

[Prior art and problems to be solved by the invention] The prior art will be explained using a magnetic disk subsystem as an example.

FIG. 4 shows a system configuration diagram illustrating a conventional example. still,
In this example, it is assumed that eight units of devices are connected inside the magnetic disk drive 4.

In a magnetic disk subsystem as shown in FIG. 4, a magnetic disk control unit 2. When a path failure occurs due to a failure in the magnetic disk device 4 or the like, the magnetic disk control device 2 reports an error in response to an input/output command from a host device (not shown).

As a result, the host device (not shown) attempts to recover from the error through software using O5 (software that controls the execution of computer programs) stored in its internal storage. Note that as a general software attempt, the same input/output command is issued several times to the path where the failure has occurred in an attempt to recover from the error.

However, if retries due to the same failure fail a specified number of times, then the same input/output command is issued and executed from another path that reaches the target device 4(0) and 4(7).

Furthermore, since new input/output commands are generally issued with priority given to certain fixed paths, there are many cases where new input/output commands are issued again to the failed path.

Each time, this input/output command is processed as an error, and a similar software attempt is made as described above.

In other words, for each input/output command from a host device (not shown), "Issue input/output command" - "Error processing" - "Retry execution for failed path" - "Failure for specified number of times" - "Retry from other path" Since "try execution" is repeated, there is a problem in that the execution time of one input/output instruction becomes long.

[Means for solving problems]

FIG. 1 shows a block diagram illustrating the invention in detail.

This block diagram shows a part of the internal configuration of the magnetic disk control device 2 explained in FIG. 4, and is provided corresponding to the channel interfaces a to c with the host device.
Channel interface control circuits 21a to 21c that control the channel interfaces a to c, and a path blocking control unit that blocks the corresponding path when a path failure exceeds a predetermined number in the path failure frequency calculation tables 23a to 23c. 22a to 22c, and path failure occurrence count calculation tables 23a to 23c that count the number of path failures that have occurred in each device for each path; A control interface control circuit 25 for controlling interfaces d and e between a control circuit 24 and a magnetic disk device comprising a plurality of devices and an adapter unit that monitors and controls execution processing of the plurality of devices; Control device 2
A control storage 26 stores control programs etc. for controlling various operations of the magnetic disk controller 2 according to predetermined procedures, and the control programs stored in the control storage 26 control various operations of the magnetic disk controller 2 according to predetermined procedures. A device (function) consisting of a microprocessor (hereinafter referred to as MPU) 27, an error detection circuit 28 that detects errors from the data transfer control circuit 24, and an error detection circuit 29 that detects errors from the control interface control circuit 25. When an error is detected by the error detection circuit when data is being transferred between the host device and each device in the magnetic disk drive based on an input/output command from the host device, the MPU determines whether or not there is a path failure. .

If it is determined that a path failure has occurred, the count values of the corresponding device areas #0 to #7 of the path failure occurrence count calculation table corresponding to the corresponding path on which data transfer processing is being performed are added.

If there is a table area where the count value of each device in the path failure occurrence count calculation table exceeds the predefined number of times,
The magnetic disk controller immediately automatically blocks the relevant path, and when the next I/O command is issued through the relevant path, unusable state information (for example, condition code -3) is sent.
)return it.

Upon receiving this, the host device can immediately issue an input/output command to the corresponding device from another path, and the processing time for executing the input/output command can be shortened in a simple manner.

〔Example〕

The gist of the present invention will be specifically explained below with reference to embodiments shown in FIGS. 1 to 3.

FIG. 2 shows an embodiment of a path failure occurrence count calculation table according to the present invention, and FIG. 3 shows an operation diagram of the embodiment of the present invention. The same reference numerals indicate the same objects throughout the drawings.

Path failure occurrence count calculation tables 23a to 2 shown in FIG.
3c corresponds to the channel interface 3-%-C, and the calculation tables 23a to 23c each have device table areas #0 to #7 corresponding to the devices 4(0) to 4(7).

Next, the operation of this embodiment will be explained based on the operation shown in FIG. It is assumed that eight units of devices are connected in this embodiment.

The main unit U27 in the magnetic disk control device 2 is a host device (CP).
U) After confirming that the input/output command issued from la or 1b has been received, which higher-level device (CPU) la,
It is issued from lb, and it is determined to which devices 4(0) to 4(7) it has been issued. (Step 100)
As a result of the determination, the path failure frequency calculation tables 23a to 23c corresponding to the current devices 4(0) to 4(7) of the channel interface aNC corresponding to the relevant host device (CPU) la, lb are read. (If the values of the tables 23a to 23c read in step 10 do not exceed the preset specified values, the corresponding device 4 (
0) to 4(7) to execute input/output commands.

(Steps 102 and 104) If an error is detected by the error detection circuits 28, 29, etc. during this execution process, the error is detected by the magnetic disk controller 2. The MPU 27 determines whether the path failure is due to a failure of the magnetic disk device 4 or the like. (Steps 105.106
) In the case of a path failure, the relevant higher-level device (CPU) la,
After adding +1 to the path failure occurrence count calculation tables 23a to 23c corresponding to the corresponding devices 4(0) to 4(7) of the channel interfaces a-c corresponding to lb, the error processing completion report is sent to the corresponding upper device ( CPU) l
a, lb. (Steps 107 and 108) On the other hand, in the judgment of step 102, the read table 23
If the value of a to 23c exceeds a preset value, the relevant path is automatically blocked at that point, so the relevant host device (CP
U) Report unavailability status information (eg, condesion code -3) to la, lb.

(Step 103) Also, if it is determined in step 105 that no error has occurred, the magnetic disk control device 2 reports normal completion to the relevant host device (CPU) Ia, lb after the input/output command execution processing is completed. Ru. (Step 109) As described above, the magnetic disk controller 2. If a path failure occurs due to a failure in the magnetic disk device 4, etc., the magnetic disk controller 2
automatically closes the faulty path and responds with unusable status information to the next I/O command.

The host device (CPUHa) that received this unusable state information
, lb (according to the built-in O8), immediately after outputting a message to that effect, the target device 4 (0) ~
Issue the same input/output command from another path to reach 4(7). As a result, software trials are omitted, and the processing time for executing input/output instructions can be significantly shortened.

〔Effect of the invention〕

According to the present invention as described above, the execution processing time of input/output instructions can be shortened in a simple manner.

[Brief explanation of the drawing]

Fig. 1 is a block diagram explaining the present invention in detail, Fig. 2 is an embodiment of a table for calculating the number of path failure occurrences in the present invention, Fig. 3 is an operation diagram of the embodiment of the present invention, and Fig. 4 is a conventional example. A system configuration diagram illustrating an example is shown, respectively. In the figure, la and lb are host devices (CPUs), 2 is a magnetic disk control device, 3a and 3b are adapter units, 4 is a magnetic disk device,
4(0) to 4(7) are devices, 21a to 21c are channel interface control circuits, 2
2a to 22c are path blockage control units; 23a to 23c are path failure frequency calculation tables; 24
25 is a data transfer control circuit, 25 is a control interface control circuit, 26 is a control storage, 27 is an MPU, and 28 and 29 are error detection circuits. , I complete g month l 2, ko)' l b 58 ° sunn miya keisei ro shio says J [boots') is the actual left third f row turn 2 concave 721, ita 5 g month σ fuzu P Threatening dfu hrp Figure X
ff11 row-xiLgq system groove wide fertilizer ≦ 4 items

Claims (1)

[Scope of Claims] A disk control device (2) having a plurality of channel interfaces (ac) with a host device; and a disk control device (2) that is controlled by the disk control device (2) and has a plurality of In a disk subsystem consisting of a disk device having a plurality of devices controlled via control interfaces (d, e), the number of path failures occurring in each device in the disk control device (2) is calculated. Path failure counting means (23a to 23c) that counts each device on the corresponding path; and path blocking means that automatically blocks the corresponding path when the count value of the path failure counting means (23a to 23c) exceeds a predetermined value. (22a
~22c), and when a path failure occurs due to a failure of the disk control device (2) and the device during execution of an input/output command from the host device, the path failure counting means (23a~22c) is provided.
In step 3c), the number of path failures that have occurred for each device on the relevant path is counted, and if the counted value exceeds a predetermined number of times, the relevant path is automatically blocked, and unusable state information is sent to the higher-level device. A path automatic closure method characterized by responding to