JPH06103453B2 - Redundant volume management information recovery processing method - Google Patents

Description

DETAILED DESCRIPTION [Overview] A duplicated volume management information table that stores configuration control information of a duplicated volume is provided for each physical volume, and a journal area for saving the contents of this duplicated volume management information table is provided. When updating each duplicated volume, the pre-update journal is acquired and the state transition is managed by enabling / disabling the journal to shorten the recovery time after a system down during management information update, and after system startup. It is possible to maintain the redundancy of the management information in 2x.

[Industrial application field]

The present invention relates to a method of updating and recovering a management information table having configuration control information of a redundant volume in a data processing system having a redundant volume management function for managing one logical volume in association with, for example, two physical volumes. In particular, the present invention relates to a dual volume management information recovery processing method that can improve reliability by shortening recovery time and maintaining redundancy.

FIG. 6 is a diagram for explaining the configuration of a general duplex volume.

In managing redundant volumes, one virtual logical volume 21 is replaced by two physical volumes 22X, 22 inside the system.
In association with Y, the WRITE access to the logical volume 21 by the data processing unit 31 writes the same contents to the two physical volumes 22X and 22Y. As a result, even if one physical volume fails, normal data can be read from the other.

The configuration management information 33 is required to manage the configuration 32 of such a redundant volume.
The 33 needs to be updated when the system is switched from duplex to single or from single to duplex due to system convenience, or when switching to another physical volume due to a physical volume failure.

It is necessary to be able to recover the management information quickly and correctly even if the system goes down or an I / O failure occurs during the updating of the management information.

[Conventional technology]

Generally, as a countermeasure against a failure of a data set or the like, a method in which the same information is multiplexed and held and a journal method in which update information is collected on a magnetic tape are used.

Conventionally, redundancy control has been performed on the management information of the duplicated volume, and a journal is recorded on the magnetic tape for the update information.

[Problems to be solved by the invention]

In the conventional method, it is necessary to recover the management information of the duplicated volume from the journal, which is the update information on the magnetic tape, when it cannot be restored in the event of a failure. However, when recovering from a magnetic tape journal, there is a problem that operator intervention is inevitable and recovery time is long.

In addition, while updating the redundant management information, the system
When a down occurs, in the recovery after the down, the system is restarted with the redundancy reduced from two to one, so that there is a problem that the reliability thereafter decreases.

[Means for solving problems]

 FIG. 1 shows a basic configuration example of the present invention.

In FIG. 1, reference numeral 10 is a processing device including a CPU and a memory, 11 is a redundant volume operation command processing section for processing an operator's operation command for changing the configuration of a redundant volume, and 12 is for setting / restoring a redundant volume. Redundant volume management utility, 13 is a dataset access processing unit that accesses and processes data on a redundant volume, 14 is a redundant volume configuration management unit that manages redundant volume configuration management information, and 15 is a redundant volume management information The pre-update journal acquisition unit that acquires the pre-update journal, 16 is the table update unit that updates the duplicated volume management information table, 17 is the journal invalidation unit that invalidates the journal after updating the table, and 18 is the journal valid / invalid A recovery processing unit that recovers the management information table, 19 is a management information table on the memory, 19X and 19Y are management information tables on physical volumes, 20 is an input / output control unit, 21 is a logical volume which is a storage medium to be virtually accessed, and 22X and 22Y are logical volumes. Physical volume associated with 21, 2
3X and 23Y represent journal areas in which the management information tables 19X and 19Y are respectively saved, and 24X and 24Y represent data sets in which data having the same contents for users is stored.

The journal areas 23X and 23Y are provided not on the magnetic tape but on the same medium as the medium storing the management information tables 19X and 19Y.

When the logical volume 21 configuration is changed or the attribute is changed by the command input by the redundant volume operation command processing unit 11, the redundant volume configuration management unit 14 first updates the configuration control information of the management information tables 19X and 19Y. Call the pre-update journal acquisition unit 15.

The pre-update journal acquisition unit 15 sequentially reads the management information tables 19X and 19Y on the medium into the management information table 19 on the memory, and the contents obtained by adding the journal valid information to the contents thereof are journal areas 23X and 23Y, respectively. Export to.
After that, the table updating unit 16 rewrites the configuration control information of the management information table 19 on the memory and writes the contents to the management information tables 19X and 19Y on the physical volume.

After updating the contents of the management information tables 19X and 19Y, the journal invalidation unit 17 is called, and the journal invalidation unit 17 writes the contents in which the journal valid information is released to the journal areas 23X and 23Y.

If a system down occurs during updating of the management information tables 19X, 19Y, etc., the recovery processing unit 18 checks the journal valid information in the journal area 23X at the time of system restart initialization, and if valid information is set, , The contents of the journal area 23X are restored to the management information table 19X, and recovery is performed. Also, for the other journal area 23Y, check the validity / invalidity, and if the journal is valid,
Similarly, the management information table 19Y is recovered by the information in the journal area 23Y.

[Action]

In the case of the present invention, the journal areas 23X and 23Y are respectively provided in advance corresponding to the duplicated management information tables 19X and 19Y, and the contents before the update are updated during the update of the management information tables 19X and 19Y. It is held in the journal areas 23X and 23Y together with the journal valid information.

Therefore, even if the system goes down during the update of the management information, the recovery can be performed quickly without the intervention of the operator.

If the redundancy is duplicated, by storing the identification numbers of the devices on which the other volume is mounted in the management information tables 19X and 19Y, the configuration control information will be inconsistent in the event of an error. If a problem occurs, the correct logical volume configuration can be reassembled.

〔Example〕

FIG. 2 is a configuration example of a management information table according to the present invention,
FIG. 4 is an example of management information update processing according to an embodiment of the present invention, FIG. 4 is an explanatory view of the relationship between update status and recovery, and FIG. 5 is an explanatory view of configuration and assembly by partner identification number according to an embodiment of the present invention. Indicates.

The management information table 19 for storing the configuration control information of the duplicated volume has the contents shown in FIG. 2, for example.

VCP1 and VCP2 are so-called validity check patterns. Information including time information at the time of updating is set in VCP1 and VCP2 so that VCP1 and VCP2 have the same contents. At the time of recovery, the contents of VCP1 and VCP2 are collated, and if they do not match, it can be recognized that a failure occurred during writing of the management information table, so the contents of the management information table are incorrect. Reject as. The same applies to journals.

TBLID is a table identifier, and an ID for identifying a management information table is stored here.

MSTSB is the status of the own volume, and OSTSB is
This is the status of the other party's volume. When this is hexadecimal '01', it indicates that the own volume or the partner volume is included in the duplex volume. Hereinafter, this state is called ON DUAL. When this is '02', it indicates that the volume is separated from the redundant volume configuration. This state is called OFF DUAL. '03' NEUTRA
L indicates the state in the middle of transition from single-duplex to dual-duplex. ANY of '04' is a value set when status information has no meaning.

FLAG is a journal flag indicating whether the journal contents are valid or invalid in the journal area. This flag has meaning only in the journal area. In the case of the present embodiment, consideration is given to the sharing of the duplicated volume between systems, and the state in which the journal is valid and the other system is being notified can also be managed by the flag.

The identification number of the device on which the remote volume is mounted is set in OTHERPDN. If there is no other volume, the identification number of the own volume is set.

Next, the management information update processing by the duplicated volume configuration management unit 14 shown in FIG. 1 will be described in accordance with the processings shown in FIG.

Prior to updating the management information table, first read the management information table 19X of the primary volume (X), and then enter the F shown in FIG.
Set LAG to hexadecimal'FF 'and write to journal area 23X.

Similar to the processing, in the contents of the management information table 19Y,
Set the journal valid information (FLAG = X'FF '),
Write to journal area 23Y.

Next, the management information table 19X for the primary volume (X)
The configuration control information of is updated to the new contents.

Management information table for secondary volume (Y)
Update 19Y configuration control information.

X'00 'is set in the journal flag in the journal area 23X to invalidate the journal of the main volume (X).

Also for the journal area 23Y of the secondary volume (Y), X'00 'is set in the flag and the journal is similarly invalidated.

When a redundant volume is shared by multiple processing units, a process such as notifying the other system of the redundant volume configuration change and waiting for a response is performed before disabling the journal. By the same status management, it is possible to recover from a failure during notification to another system.

In response to the above processing or processing, the update status of the management information table transits to statuses S0 to S12 shown in FIG. The journal of the main volume is valid from status S2 to status S8. In the case of a secondary volume, the journals of that management information table are valid from status S4 to status S10.

In state S1, state S9 and state S3, state S11, since the journal area is being updated, it is uncertain whether it is valid or invalid. VCP shown in Fig. 2 in the journal area
If 1 and VCP2 do not match, the journal is rejected as invalid. If they match, the validity / invalidity is determined according to the contents of the journal flag.

It is assumed that a command is issued that allocates a logical volume to two physical volumes 22X and 22Y and, for example, separates the physical volume 22Y while operating in a redundant configuration. In this case, it is necessary to update the management information tables 19X and 19Y to update the configuration control information from the state (a) shown in FIG. 4 to the state (b).

Although the update is performed by the processing shown in FIG. 3, there is a possibility that the system will be down during the update due to a failure of the main body system or the like. When the system goes down during the management information update, in the recovery, when the update status is from S0 to S8, the configuration control information in the management information table, especially by the pre-update journal in the main volume,
It can be restored to the state of (a) shown in FIG. Further, in the case of the state S9 or later, since the new configuration control information of the management information table in the main volume is valid, it is possible to recover to the state (b) shown in FIG.

Since each management information table 19X, 19Y stores the identification number of the device on which the partner volume is mounted, it is prevented from being reassembled with an incorrect logical volume configuration in the event of an abnormality.

FIG. 5 is a diagram for explaining the construction and assembly based on the identification number of the partner device. As shown in FIG. 5 (a), the partner identification number in the management information table points to the identification number of the own volume, and when there is no partner volume, the configuration and assembly are performed in the single-sided state of the X side. As shown in Fig. 5 (b), the status of the own volume and the status of the partner volume in the Y-side management information table are both
It is ONDUAL, and it is assumed that the other party's identification number on the X side points to its own device. In this state, it can be considered that the configuration control information was updated only in the X plane due to an I / O abnormality in the Y plane. Therefore, in this case, construction and assembly are performed with the single side of the X side.

In the case of FIG. 5C, the components are assembled and assembled in the duplicated state. For example, in the case of an inconsistent state in the physical volume switching as shown in FIG. 5D, the Y surface, which is a pair of the X surface, is set as the physical volume of # 0, and the configuration and assembly are performed in a duplicated state. In the case of the state of FIG. 5 (e), it is considered that the configuration control information could only be updated on the Y side due to an I / O error on the X side. Therefore, in this case no automatic assembly is performed. Also in the case of FIG. 5 (f), the structural assembly is not performed. The construction and assembly can be similarly considered in various other cases.

〔The invention's effect〕

As described above, according to the present invention, when the system goes down while updating the duplicated volume management information, the recovery can be performed at high speed without operator intervention. Further, when the redundancy is multiplexed, the redundancy can be used as it is by the recovery after the down.

[Brief description of drawings]

FIG. 1 is a basic configuration example of the present invention, FIG. 2 is a configuration example of a management information table according to the present invention, FIG. 3 is an example of management information update processing according to an embodiment of the present invention, and FIG. And FIG. 5 is a diagram for explaining the relationship between the recovery and recovery, FIG. 5 is a diagram for explaining the configuration and assembly by the partner identification number according to one embodiment of the present invention, and FIG. 6 is a diagram for explaining the configuration of a general duplicated volume. In the figure, 10 is a processing device, 11 is a redundant volume operation command processing unit, 12 is a redundant volume management utility, 13 is a data set access processing unit, 14 is a redundant volume configuration management unit, 15 is a pre-update journal acquisition unit, and 16 is a Table update unit, 17 journal invalidation unit, 18 recovery processing unit, 19,
19X and 19Y are management information tables, 20 is an input / output control unit, 21 is a logical volume, 22X and 22Y are physical volumes, 23X and 23Y are journal areas, and 24X and 24Y are data sets.

Claims (1)

[Claims] 1. In a data processing system having a redundant volume management function for managing a logical volume in association with a physical volume, a redundant volume management information table (19X) in which configuration control information of the redundant volume is stored for each physical volume. , 19Y), the journal areas (23X, 23Y) whose contents are respectively saved on the same medium in correspondence with the above redundant volume management information table, and the above redundant volume management information when updating each redundant volume management information. A pre-update journal acquisition unit (15) that transfers the contents of the table to which journal valid information is added to the journal area, and a table update unit (15) that updates each of the above-mentioned duplicated volume management information tables after acquisition of the pre-update journal ( 16) and the above redundant volume management information After the table is updated, the pre-update journal invalidation means (17) for canceling the valid display of the journal in the journal area and invalidating the journal, and the failure in the journal area for the failure during the update of each redundant volume management information It is provided with a recovery processing means (18) for checking the validity / invalidity of the journal for each physical volume, restoring each redundant volume management information according to the state, and reconstructing the redundant volume. Redundant volume management information recovery processing method.