JPH03144721A - Magnetic disk control device - Google Patents

Abstract

PURPOSE:To make it unnecessary to secure the required time of disk copying for system down by executing disk copying operation guaranteeing consistency during the operation of a system. CONSTITUTION:When a write request from a user program in the other host system 14 to an area to be monitored arrives after reading out corresponding data from a master disk 12 to the area to be monitored in a slave disk 13, the command processing of the master disk 12 is executed, but an error is returned to the command processing of the slave disk 13, and after ending a copying write command, the command processing of the slave disk 13 is executed by retrial from the other host system 14. Consequently, exclusive control between a copy program and a user program can be attained, disk copying operation guaranteeing consistency is executed during the operation of the system without stopping the system and the need for securing the required time of disk copying for system down can be eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a magnetic disk control device used in a data processing system having a disk subsystem with dual disks.

(Prior Art) Conventionally, in a data processing system that has a disk subsystem with duplicated disks in order to improve reliability, a failure occurs in one of the duplicated disks, resulting in failure of one of the disks. When this happens, degenerate operation is performed. After disk replacement, in order to return to normal operation with duplicated disks, it is necessary to copy all data from the normal disk to the replaced disk. FIG. 6 shows the state transition of the disk subsystem at this time. In conventional systems, the sum of the drive exchange time (TI) and the disk copy time (T2) (TI + T2) is the system down time, and if this time is long, , the impact on users was large.

Therefore, in order to reduce the time that the system goes down, it is possible to perform disk copying while the system is running, but in this case, multiple O8s share one disk subsystem as shown in Figure 5. When considering a loosely coupled system, in the past, when attempting to execute a disk copy while the system was running, it was not possible to perform exclusive control between the copy program and the user program at the OS level. That is, conventionally, in a loosely coupled system in which multiple O5s share one disk subsystem as shown in FIG. 5, a magnetic disk control system with a control function that allows exclusive control of copy programs and user programs The device was not present.

Here, the reason why disk copying cannot be performed in the conventional system while the system is in operation will be explained using FIG. 4.

In FIG. 4, it is assumed that the copy program 41 under the control of 0840 is currently trying to copy block A of the mask disk 43 in the disk subsystem 42 to the slave disk 44. First, the copy program 41 reads out program A in the mask disk 43 via the magnetic disk controller 45a (t(aol)). Subsequently, when an input/output request comes from the user program 45 of another system when trying to write to the slave disk 44, the O
8, it is not known which one will be executed first.

A problem arises when a request from the user program 45 is executed first and that request writes into block A. Now, the user program 45 writes to block B inside block A (
a02), and then writing of the copy program 41 to the slave disk 44 is executed (a03). Then, on the mask disk 43, block B exists inside program A, but on the slave disk 44, block B is written and block A is overwritten, resulting in a data mismatch between the mask disk 43 and the slave disk 44. occurs.

(Problems to be Solved by the Invention) As described above, in the past, there was a problem in that the system had to be brought down when copying a disk after replacing a failed disk. In addition, in a loosely coupled system where multiple O5s share one disk subsystem, there is a problem that disk copying cannot be performed while the system is running because the magnetic disk control device does not have an exclusive control function between the copy program and the user program. was there.

The present invention has been made in view of the above circumstances, and it is possible to restore data by copying the disk while the system is running after replacing a failed disk in a data processing system having a disk subsystem with duplicated disks. It is an object of the present invention to provide a magnetic disk control device that eliminates the copying time (see T2 in FIG. 6) that accounts for system down time.

[Structure of the Invention (Means and Effects for Solving the Problems)] The present invention provides a data processing system having a disk subsystem with dual disks, in which a slave disk is used during disk copying due to a failure of one disk. Under the same mode setting, only write commands for copying are accepted as blocks to be monitored in a certain range (copy block length) from logical address zero (0), and other write commands are accepted. and a means for executing the same process on the next certain range as the next block to be monitored when the copy processing of the block to be monitored is completed. This enables exclusive control between the copy program and the user program, making it possible to perform disk copy with guaranteed consistency while the system is running without stopping the system, eliminating the time required for disk copying during system downtime. can do.

(Example) An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram for explaining the disk copy processing operation in one embodiment of the present invention, and FIG. 2 I shows the hardware configuration of one magnetic disk control device in the same embodiment. It is a block diagram.

In the figure, 10 is a disk subsystem, tta, tt
b is a magnetic disk control device, 12 is a mask disk which becomes the read side during disk copying, 13 is a slave disk which becomes the write side during disk copy, 1414,...
. is a host system that accesses the disk subsystem IO. Reference numerals 111 to 11B respectively constitute l\-ware components of the magnetic disk control device 11a (llb) provided in the disk subsystem lO, 111 is a CPU that controls the entire magnetic disk control device, and 112 is a CPU that controls the entire magnetic disk control device. (13) and the host system 14 side; 113 is a ROM in which a control program for controlling processing of the CPU 111 is stored; ), a data bus 115 is connected to the disk 12 under the control of the CPU 111.
(13) A disk control unit that controls access to the host system 14.14. ...and RAM 112
host system, including the transfer of read/write data to and from 14.14. ... and the magnetic disk control device 11a
This is a host control unit that controls the exchange of information with (llb).

FIG. 3 is a flowchart showing disk copy processing in the embodiment of the present invention.

The operation of an embodiment of the present invention will now be described with reference to FIGS. 1 to 3.

In order to perform disk copying that guarantees data consistency while the system is running, the copy program under the control of the O8 of the host system 14 runs on the disk subsystem lO8.
magnetic disk control device 1 on the slave disk I3 side in
Issue a copy mode setting command to 1b, and
A copy mode flag is set in the status register under the control of the U 111, and the magnetic disk device 11b is notified that data will be copied from now on (first
Figure al, Figure 3, step S10. 5ll). Having received this notification, the magnetic disk device flb on the slave disk 13 side first starts from logical address zero (0) to a certain range (
The copy block length (for example, 84 bytes) is set as the first area to be monitored, and an error is returned in response to a write command of the user program within this area (step S12 in FIG. 3). Next, the copy program is executed by the magnetic disk controller 1 on the mask disk 12 side.
A read command is issued to 1a, and data corresponding to the monitoring target area of the slave disk 13 is read from the mask disk 12 (step a2 in FIG. 1, step S in FIG. 3).
13). The read data is then written to the monitored area of the slave disk 13, and execution of the user program write command to the monitored area on the slave disk 13 is prohibited while this copy write command is in progress (the third Figure steps S14-S1B, S20-5
24).

That is, from the mask disk 12 to the slave disk 1
After reading the data corresponding to the monitoring target area on 3, when a write request to the monitoring target area arrives from the user program of the other system host system 14 (step 520 in FIG. 3), command processing to the mask disk 12 is performed. is executed (Figure 1 aJ Figure 3 steps 521-324)
, an error is returned for command processing to the slave disk 13, and after the copy write command is completed, the command processing to the slave disk 13 is executed by a connection attempt from the other host system 14 ( Figure 1 a4.a3-, Figure 3 Step S23.S
20-524).

When the write command for copying a fixed block length (64 bytes) is successfully completed, the monitoring target area is updated and the next fixed block length (64 bytes) on the mask disk 12 is completed normally.
The data of 4 bytes) is read as the data of the new monitoring target area, and all the data on the mask disk 12 is copied, thereby ending the copy write command (step S17.31g in FIG. 3).

When all data copying from the mask disk 12 to the slave disk 13 is completed, the copy mode is canceled by the copy mode setting command, the copy mode flag is reset, and the normal operation mode is restored (third
Figure step S19. S20).

In this way, while executing the copy write command,
Execution of the user program write command to the monitored program on the slave disk 13 is prohibited, and the user program write command is executed after the copy write command is completed. This allows disk copying with guaranteed consistency to be performed while the system is running.

[Effects of the Invention] As described in detail above, according to the present invention, in a magnetic disk control device of a data processing system having a disk subsystem with duplicated disks, a disk that is replaced due to the occurrence of a failure, etc. A means for setting a copy mode when performing a disk copy and restoring data while the system is running, a means for specifying a block to be monitored on a slave disk that is the writing side under the same setting mode, and A means for prohibiting data writing to the monitored block except for copyright command processing, and updating the monitored block of the slave disk upon completion of data writing to the monitored block by the copyright command processing. By adopting a configuration having a means for The time required for disk copying during system downtime can be eliminated.

[Brief explanation of the drawing]

FIG. 1 is a diagram for explaining the disk copy processing operation in one embodiment of the present invention, and FIG. 2 is a block diagram showing the hardware configuration of one magnetic disk control device in the same embodiment. FIG. 3 is a flowchart showing the disk copy processing in the above embodiment, FIG. 4 is a diagram for explaining the conventional disk copy processing operation, and FIG. 5 shows the loosely coupled system configuration targeted by the present invention. Block diagram, 6th
The figure is a state transition diagram of the disk subsystem. 10... Disk subsystem, Lla, Ilb
...Magnetic disk control device, 12...Mask disk, Part 3...Slave disk, 14...Host system, Port 1...CPU, 112...RAM5IL
:(...ROM%114...Data bus, 115...
... Disk control unit, 116... Host control unit.

Claims (1)

[Claims] In a data processing system that has a disk subsystem with duplicated disks, set the copy mode when performing disk copy and restoring data on a disk that has been replaced due to a failure etc. while the system is running. a means to monitor the slave disk that is the writing side under the same setting mode, a means to specify the target block,
and a means for prohibiting data writing to the monitored block except for copyright command processing, and a means for prohibiting data writing to the monitored block of the slave disk upon completion of data writing to the monitored block by the copyright command processing. What is claimed is: 1. A magnetic disk control device characterized in that the magnetic disk controller has a means for updating a disk, and executes disk copying with consistency guaranteed during system operation.