JPH0420487B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to a multiplexed data processing device;
In particular, it relates to update management of duplex files in a duplex system.

Prior Art and Problems In data processing equipment, the host CPU (central processing unit) and
Duplicating everything, including DASD (direct access storage devices), is becoming popular. In a redundant system, processed data is recorded twice on two DASDs, so even if one DASD goes down, the other DASD
This allows data processing to continue.
The record or history information (update management information) of write processing performed on the DASD is stored in the control adapter of the DASD, and since it is a duplicate file, the update management information of each DASD will have the same content if normal. It is. If one side goes down, the update management information after that time will be blank, so both
By comparing the update management information of the DASD, it is possible to know to what extent the DASD was normal and where to start updating the DASD that has gone down, and if the update management information is saved, startup after the failure is repaired can be carried out advantageously.

In a system where the update management information for such duplexed files is stored in an external storage device (in the above example, DASD), the external storage device performs update management as long as it is operating, and updates are not performed even though they do not belong to the duplexed files. The drawback is that management is done even when it is not necessary.

Purpose of the invention The present invention aims to improve the above points.
The reliability of the update management information is improved by instructing the host to enable/disable the update management function of the duplicated file so that the update management information is not taken when unnecessary, and the information is not recorded when it is unnecessary. This is intended to reduce the load on the control device of the external storage device.

Structure of the Invention The present invention provides update management information for multiplexed files in an external storage device constituting the files,
When recovering a multiplexed file, the data in the normal file is written to the failed file based on the update management information of the normal file, thereby recovering the failed file. In the data processing device, a flag is provided in the update management information to enable or disable the function of collecting file update management information, and the flag is enabled or disabled by a command from the central processing unit of the data processing device. This feature is set to an invalid state, which will be explained next with reference to the drawings.

Embodiment of the Invention FIG. 1 shows an outline of a duplex data processing device to which the present invention is applied, and 10 and 12 are host
A CPU, 14 and 16 are control devices for a DASD magnetic disk device in this example, 18 and 20 are disk control adapters, and 22 and 24 are magnetic disk devices.
Although a large number of disks 22 and 24 are provided in a large-scale data processing apparatus, only two of each are shown here. As shown in the diagram, this system has duplicate hosts, disks, and other components to increase reliability. The area within the chain line frame 26 constitutes an external storage device for the internal storage device (main storage) of the host. The disk control adapters 18 and 20 have an update management area 18.
a, 20a are provided, and update management information is stored in these areas.

FIG. 2 is a detailed explanatory diagram of the disk control adapter 18, in which the same parts as in FIG. 1 are designated by the same numbers. 27 is the upper interface circuit, 28
is an address pointer, 29 is a read circuit, 30 is a write circuit, and 31 is a lower interface circuit. The command from the control device 14 is decoded by the upper interface circuit 27, and if it is a write command, a predetermined address in the update management area 18a is accessed by the address pointer 28, and the data is written by the write circuit 30. In the case of a read-related command, data is read by the read circuit 30.

FIG. 3 shows information stored in the update management areas 18a and 20a. AA, AB... The first symbols A and B of BD indicate device names, and the following symbols A to D indicate channels. What is a device? Disk 22, 24
In this example, there are two of each, so they are called A and B.
It is shown in Channels refer to the ports of the disk control adapters 18 and 20 to the control devices 14 and 16, and in this example there are two ports each, but in FIG. 3 there are four ports.
These are shown as A, B, C, and D assuming that there are several cases. In the update management area, as shown in the figure, write processing history is kept for each device and channel, and the history contents include data valid flag a, update start date and time b, update start block address c, update completion block address d, and data check. Consists of flag e. Of these, b to d form the main part of the update management information, a indicates whether or not these data are valid, and e indicates a parity bit for error checking of these data. The data check flag e may have a data correction function such as ECC.

Writing and reading of such update management information is performed by a host command such as a Set Buffer Parameter command, and this operation will be explained using the flowchart shown in FIG.
When the Set Buffer Parameter command is issued, if the write parameter described later is specified in this command, if update management is to be started, the update management area write command issued subsequently will The update management data shown is set, and the valid flag a is set on. When update management is no longer necessary, the validity flag indicated by 3 is reset to OFF using a termination parameter described later. Further, when temporarily freezing the update management, a freezing flag indicated by 4 is set using a freezing parameter described later (the freezing flag itself is not shown).
On the other hand, when a read parameter, which will be described later, is specified in the set buffer parameter command, the update management data indicated by 2 is read and sent by the subsequently issued update management area read command. The following parameters are prepared as parameters.

The parameter is a light parameter that notifies the control device that duplex control is in progress. Then, as mentioned above, the update management area write command issued following the set buffer parameter command writes data a, b, c, d, and e explained in FIG. The updated block address is updated for each command. The processing flow for write commands issued thereafter follows the route shown in 5 in the figure. This processing flow writes the block number to the update management area when the valid flag a is on and update management is not frozen.

The parameter is a read parameter, and as mentioned above, the 4 unit data of the update management area is read by the update management area read command issued subsequently.
It allows you to read AA~AD, BA~BD.

The parameter is a frozen parameter, which indicates to the control device that it is not necessary to update the updated block address due to circumstances such as the data not needing to be repaired in the event of a failure even though it is operating independently or under duplex control. It is something to inform.

The parameter is an end parameter that causes data a to e to be written to any block by the update management area write command issued subsequently, but in this case, update data is not taken even if a write command is issued thereafter. It is something to do.

The data valid flag a indicates whether data b to d are valid or invalid.If you do not want to collect update management information anymore, use a parameter to set the flag a.
Set to “0” (invalid). Of course, when starting to obtain update management information, the flag a is set to "1" by a parameter. One disk system 1
8, 22 or 20, 24 are down, read and compare the contents of the update management areas 18a, 20a of the adapters 18, 20 as shown in FIG. If d is different, device A (that in each system)
It is sufficient to perform write processing from channel B to the down device.

Data processing equipment that collects update management information automatically collects this information using hardware.
In the present invention, the information is also transmitted to the disk control adapter 18,
20 hardware samples. That is, information on the update management area 18a is collected by the address pointer 28, read circuit 29, and write circuit 30 shown in FIG. As mentioned above, if this function cannot be enabled or disabled quickly, it will collect information even when it should not be collected, which will eventually make the update management information itself unreliable. In this regard, if it is enabled and disabled easily and quickly using host commands as in the present invention,
If one of the DASDs goes down, or if the duplication of the DASD is canceled and the DASD operates independently, collection of update management information can be immediately prohibited, which is extremely effective.

The present invention can be applied not only to magnetic disk devices but also to data processing devices that are configured as multiplexed files and that collect history information to obtain similar effects.

Effects of the Invention As explained above, according to the present invention, the update management function can be freely enabled/disabled using the OS (management program), thereby providing versatility and improving reliability. Furthermore, when the update management function is disabled, the update management function does not operate, which has the effect of reducing the load on the control device of the external storage device.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a block diagram showing details of the disk control adapter, Fig. 3 is an explanatory diagram showing the contents of its update management area, and Fig. 4 is an update management area. 2 is a flowchart illustrating the processing. In the drawing, 26 is an external storage device, 10 and 12 are central processing units, and 18a and 20a are areas for storing update management information.

Claims (1)

[Claims] 1 The update management information of a multiplexed file is stored in the external storage device that constitutes the file, and when the multiplexed file is restored, the data of the normal file is updated based on the update management information of the normal file. In a multiplex data processing device configured to recover a file in which a failure has occurred by writing to the file in which a failure has occurred, enable or disable a function for collecting file update management information in the update management information. A data processing apparatus, characterized in that a flag is provided, and the flag is set to be enabled or disabled by a command from a central processing unit of the data processing apparatus.