JPH07210331A - Disk array device - Google Patents

Abstract

PURPOSE:To provide a disk array device capable of accurately detecting the states of respective built-in disk devices at the time of start. CONSTITUTION:When an initialization instruction from a host 21 is received, a disk array controller 13 sets parameters, then performs a test on whether or not a first disk device 14i is capable of normally performing read and write and writes management information in the management information storage area 15i of the disk device 14i when it is capable of normally performing the read and write. Also, When an (i)-th disk device 14i is not capable of normally performing the read and write, '1' is added to a counter, the contents of identification information is rewritten to (i), an error message indicating that it can not be operated even as the disk array device whose redundancy degree is zero is outputted when the counter is more than '2' and an initialization processing is ended. '1' is added to (i) when the counter is equal to or less than '1' and the test and the write of the management information are performed to the next disk device when a variable (i) is equal to or less than the number of the entire disk devices.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device, and more particularly to a disk array device for storing data in a redundant configuration.

[0002]

2. Description of the Related Art A disk array device is a storage device that is a combination of a plurality of disk devices that can operate independently, and is configured so that a host device such as a computer can handle it as one disk device.

FIG. 14 shows an example of the configuration of a conventional disk array device. The disk array device 11 comprises an interface 12, a disk array controller 13, and a plurality of disk devices 14 _{1 to} 14 ₅ . The number of built-in disk devices 14 differs depending on the disk array device. Such disk array devices include those for the purpose of improving the data access speed and those for the purpose of improving the reliability as a storage device. In the latter, the data for which a write request is made is Redundancy is added and stored in the disk array device.

The data write operation will be briefly described with reference to FIG. The redundant configuration used here is a RAID (Redundant Arrays of Inexpensive D).
It is called level 3 of isks).

The disk array controller 13 is a host
Data 20 for which a write request has been
Size, for example, divided into sector units, and the disk device 1
Four₁~ 14_FourStorage area 16₁ ~ 16_FourDispersed and written
Imprint At this time, the disk array controller 13
Is the data D₁, D₂, D₃, D_FourExclusive OR operation of
And the parity P that is the calculation result is
Storage 14_FiveRecording area 16 _FiveYou can also write to
To go.

As a result of such a write operation, a state in which the data stored in a certain disk device can be calculated from the data stored in another disk device and the parity is formed in the disk array device. . For example, the data D ₄ matches the exclusive OR operation result of the data D ₁ , D ₂ , D ₃ and the parity P. Therefore, it is possible to respond to the request without using the disk device in which the data requested to be read is actually stored. In a disk array device, if a read request is made to the disk when an access failure occurs in one of the built-in disk devices, it is necessary to use the information stored in other disk devices in this way. Responding to the read request by reconstructing the stored data.

In order to determine whether the response is due to this rebuilding, it is necessary for the disk array controller to recognize the disk device having the access failure. There are various means for this purpose, and, for example, it has been practiced to store information for specifying a disk device having a failure (failed disk device) in a memory in the disk array controller.

However, when the information for identifying the failed disk device is stored in the volatile memory, the information is lost when the system is stopped. Therefore, the user of the disk array device is required to restart the system. However, there is a problem that the information specifying the failed disk device must be reset. For this reason, various devices have been proposed that store information specifying the failed disk device in addition to the volatile memory.

For example, Japanese Unexamined Patent Publication No. 3-229331 discloses a device using a non-volatile memory for state storage of a dual disk device.

FIG. 16 shows an outline of the duplicated disk device. This system includes two disk devices 22, a control unit 23 that integrally controls them, and a nonvolatile storage unit 2.
4 and an operating system (OS) 25. In this duplicated disk device, when an access failure occurs in one disk device, the disk device identification information corresponding to the other disk device is stored in the nonvolatile storage unit 2.
Written to 4. And when restarting the device,
From the disk device identification information written in the non-volatile storage unit 24, a normally functioning disk device is specified.

Further, Japanese Patent Publication No. 2-32652 discloses that
A dual disk device is disclosed which determines the status of the disk device from the information recorded on the disk medium.

Referring to FIG. 17, this dual disk device is used.
The configuration and operation of will be described. Japanese Patent Publication No. 2-32652
The duplicated disk device disclosed in the publication is two disk devices.
22 ₁, 22₂And channel device 26₁, 26₂And central office
It is composed of the processing device 27, and the data of each disk device 22 is
The history counter 29 and the history are recorded on the disk medium.
A counter verification unit 30 is provided, and in the central processing unit 27
Whether each disk device functions normally in the main memory 28
A control table 31 for storing information indicating whether or not
It has been burned.

The central processing unit 27 writes the same value in each history counter unit 29 when both the disk devices 22 are functioning normally. And
When an access failure occurs in _one disk device, for example, the disk device 22 ₁ , the other disk device 2 1
Increase 2 ₂ values of the history counter 29 _2. When the device is restarted, the history counter unit 2
The values of 9 ₁ and 29 ₂ are read and compared, and when the values are different, it is determined that the disk device in which the larger value is written is a device that functions normally.

The history counter verification unit 30 is a storage area adjacent to the history counter unit 29, and in this duplicated disk device, the history is verified by checking whether or not the data written in this portion has been changed. Whether the contents of the counter section are correct or not is determined.

[0015]

According to the two techniques for the duplicated disk device described above, the information for specifying the failed disk device is not lost even if the system is stopped. However, these techniques are for a duplicated disk device, and cannot be directly applied to a disk array device having three or more disk devices.

Further, these techniques do not deal with the case where the disk device is replaced when the system is stopped, so that the following problems may occur. For example, in a duplicated disk device that stores the state of the disk device in a non-volatile memory, if the disk device is replaced when the system is stopped, this is not reflected in the information in the non-volatile memory. For this reason, when the normally operating disk device is replaced by mistake, the control unit determines that the disk device in which information is not stored is a normal device and starts operating.

In the duplicated disk device having the history counter on the disk medium, when the normal disk device is replaced with a new disk device, the information stored in the history counter verification unit is correct. Therefore, it is determined that the device does not function normally, and the operation of the failed disk device is started assuming that it functions normally.

Therefore, an object of the present invention is to provide a disk array device capable of detecting a disk device having an access failure at startup.

Another object of the present invention is to provide a disk array device which can accurately determine the state of each disk device even if the disk device is replaced when the system is stopped. .

[0020]

The invention according to claim 1 is
A plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and determination of the status of each disk device in a predetermined storage area of each of the plurality of disk devices Initializing means for writing numerical information having the same contents as history information used for recording, detecting means for detecting occurrence of access failure in the disk device, and disk device except for the disk device in which the access failure is detected by the detecting means. And a rewriting means for rewriting the history information stored in a predetermined storage area of the disk device designated by the designating means in the order set by the setting means so that the value increases by a predetermined amount. A reading means for reading history information stored in a predetermined storage area of a plurality of disk devices at the time of startup, and a reading means for reading the history information. Based on the history information read by the stage, the setting means has a searching means for searching a disk device having a larger history information of the disk devices later ordered, and the disk device searched by this searching means has an access failure. And a determination unit that determines that the device is a disk device.

That is, according to the first aspect of the invention, when the disk array device is initialized, the same history information is written in a predetermined storage area of each disk device, and during normal operation, an access failure occurs in the disk device. When it happens,
The history information of the disk devices excluding the faulty disk device is rewritten so that the contents increase by a predetermined amount, for example, "1". This rewriting is performed in the order set by the setting means.

The reading means reads the history information in each disk device at the time of startup, and the searching means is stored in two consecutive disk devices in the order set by the setting means. The history information is compared, and a disk device in which the ordered history information of the disk device is larger is searched later. The judging means judges that the disk device searched by the searching means is a disk device having an access failure. As described above, since the history information is rewritten in a predetermined order, even if the history information rewriting by the rewriting means is interrupted, whether or not there is a disk device having an accurate access failure at the time of start-up. It will be possible to judge whether or not.

According to a second aspect of the present invention, a plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and a predetermined number for each of the plurality of disk devices. Initializing means for writing the slot number to which the disk device is connected and the history information, which is numerical information, to the storage area as management information used for determining the state of each disk device, and an access failure occurred in the disk device. Detecting means for detecting, specifying means for specifying one by one the disk devices excluding the disk device in which the access failure is detected by the detecting means, in the order set by the setting means, and the disk device specified by this specifying means Rewriting means for rewriting the history information stored in the predetermined storage area so that the value increases by a predetermined amount, and a plurality of Read-out means for reading out the management information stored in each predetermined storage area of the disk device, and whether the slot number in the management information read out by the read-out means matches the one written by the initialization means. Determination means for determining whether or not there is a change means for changing the history information read from the disk device, which is determined by the determination means to be inconsistent, to history information having a smaller value than the history information used by the initialization means for writing Corresponding to a searching means for searching a disk device having a larger history information of the disk device later ordered by the setting means, based on the history information corresponding to each disk device, and the disk device searched by this searching means If the history information to be changed has been changed by the changing means, it is determined that the disk device has been replaced. And, in other cases, the disk apparatus comprises a determining means for determining that the disk device having an access failure

That is, according to the second aspect of the invention, at the time of initialization of the disk array device, management information including slot information and history information is written in a predetermined storage area of each disk device, and during normal operation, When an access failure occurs in the disk device, the history information of the disk devices other than the failed disk device is rewritten so that the contents increase by a predetermined amount, for example, "1". This rewriting is performed according to the order set by the setting means.

The reading means reads the management information in each disk device at the time of start-up, and the judging means judges whether or not the slot information in the management information is written by the initializing means. To do. The changing means changes the history information corresponding to the disk device for which the slot information is judged to be not normal by the judging means to a value smaller than the history information used for writing by the initializing means. Note that this means does not rewrite the history information in the disk device, but only changes the read history information. Then, the searching means compares the history information corresponding to the two disk devices sequentially ordered by the setting means, and searches for a disk device in which the history information of the disk devices ordered later becomes larger. . If the history information corresponding to the disk device searched by the searching means is changed by the changing means, the judging means judges that the disk device is a replaced disk device, and the history information Is not changed by the changing means, it is determined that the disk device is a disk device having an access failure. In this way, since the history information is rewritten in the predetermined order and the determination is performed using the slot information, when the rewriting of the history information by the rewriting means is interrupted or the disk device is replaced when the system is stopped. Also, it is possible to accurately determine whether or not there is a disk device having an access failure and a correctly replaced disk device.

According to a third aspect of the present invention, a plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and a predetermined number for each of the plurality of disk devices. An initialization unit that writes first history information as history information used to determine the state of each disk device in the storage area, a detection unit that detects that an access failure has occurred in the disk device, and an access failure by this detection unit. Of the disk device specified by the specifying means and the content of the predetermined storage area of the disk device specified by this specifying means are different from the first history information. Rewriting means for rewriting to history information and a reading means for reading out information stored in a predetermined storage area of each disk device at startup. And a searching means for searching for a disk device in which information matching the second history information is stored based on the information read by the reading means, and a setting means for the disk device searched by the searching means. And a discriminating means for discriminating that the previously ordered disk device is a disk device having an access failure.

That is, according to the third aspect of the invention, when the disk array device is initialized, the first history information having the same contents is written in a predetermined storage area of each disk device, and the disk device is operated during normal operation. When an access failure occurs in, the history information of the disk devices ordered after the failed disk device is rewritten to the second history information different from the first history information.

The reading means reads the history information in the respective disk devices at the time of startup, and the searching means searches the disk device in which the second history information is stored. Then, the judging means judges that the disk device searched by the searching means and previously ordered by the setting means is a disk device having an access failure. As described above, since the history information is rewritten in a predetermined order, even if the history information rewriting by the rewriting means is interrupted, whether or not there is a disk device having an accurate access failure at the time of start-up. It will be possible to judge whether or not.

According to a fourth aspect of the present invention, a plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and a predetermined number for each of the plurality of disk devices. An initialization unit that writes first history information as history information used to determine the state of each disk device in the storage area, a detection unit that detects that an access failure has occurred in the disk device, and an access failure by this detection unit. Of the disk device specified by the specifying means and the content of the predetermined storage area of the disk device specified by this specifying means are different from the first history information. Rewriting means for rewriting to history information and a reading means for reading out information stored in a predetermined storage area of each disk device at startup. And first specifying means for specifying the disk device ordered in advance by the setting means of the disk device in which the information matching the second history information is stored, based on the information read by the reading means. Second identifying means for identifying a disk device storing information that does not match any of the first and second history information based on the information read by the reading means, and the first and second When the disk device specified by the specifying device is the same device, the first judging device for judging that the disk device is a properly replaced disk device and the specifying device by the second specifying device. When the disk device is specified by the first specifying means,
And a second judging means for judging that the disk device specified by the first specifying device is a disk device having an access failure.

That is, in the invention described in claim 4, the first history information having the same content is written in a predetermined storage area of each disk device at the time of initialization of the disk array device, and the disk device is operated during normal operation. When an access failure occurs in, the history information of the disk devices ordered after the failed disk device is rewritten to the second history information different from the first history information.

The reading means reads the history information in each disk device at the time of start-up, and the first specifying means searches the disk device in which the second history information is stored and If such a disk device can be searched for, an ordered disk device is specified before the searched disk device. The second specifying means is
A disk device having history information different from the first and second history information is searched, and when such a disk device can be searched, the searched disk device is specified (identification information is stored). .

Then, the first judging means judges that the disk device is a disk device which has been properly exchanged when the disk devices specified by the first and second specifying means are the same device. However, the second determining means is specified by the first specifying means when the second specifying means has not been specified and the first specifying means is specifying the disk device. It is determined that the disk device has the access failure. In this way, since only two types of history information are used and history information is rewritten in a predetermined order, even if a disk device is replaced when the system is stopped, a disk device having an access failure and a correct replacement will be performed. It is possible to determine whether or not there is another disk device.

[0033]

EXAMPLES The present invention will be described in detail below with reference to examples.

First embodiment

FIG. 1 shows an outline of a disk array device according to the first embodiment of the present invention. The disk array device 11 of the first embodiment comprises an interface 12, a disk array controller 13 and five disk devices 14 ₁ to 14 ₅ . The disk device 14 is a disk device that can operate independently, and a predetermined storage area of each disk device is provided with a management information storage area 15 for storing management information used for determining the state of the disk array device. ing. Disk array controller 1
A control unit ₃ integrally controls these five disk devices 14 ₁ to 14 ₅ in order to respond to a data access request from a host 21 such as a computer. The disk array controller 13 includes a control processor, a program memory that stores a program that defines the operation of the control processor, a memory that stores parameters used during the operation, and an interface with each disk device. Has been done.

The operation of this disk array device includes an initialization process which is performed prior to the use of the disk array device.
It is roughly divided into management information update processing that is performed when a failure occurs in a disk device during normal operation, and status determination processing that is performed when the device is started up in order to determine and recognize the status of each disk device. . First, the initialization process will be described.

FIG. 2 shows a disk array at the time of initialization processing.
The operation flow of the ray controller is shown. Where the processing pair
Each disk device, which is supposed to be an elephant, has a so-called format.
It is a completed device, and the disk array controller 13
From the host 21 via interface 12
When receiving the periodization instruction, the counter C_FAILAnd identification information N
_FAILTo "0" respectively, and to the variable i
"1" is set (step S101). Counter C
_FAILIs used to store the number of abnormal disk units.
Identification information N_FAILIs the counter C_FAILIs counted in
Identification information of one disk device
Is a parameter used to store the.

After setting the parameters, the disk array controller 13 tests whether or not the i-th disk device 14 _i can read and write normally (step S
102), if the data can be read and written normally (step S103; Y), the management information is written in the management information storage area 15 _i of the disk device 14 _i (step S10).
4) and proceeds to step S107. Note that step S10
In 4, the management information including the device identification information for identifying the disk array device, the slot information that is information according to the connected slot number, and the history information is written. The only information whose content changes accordingly is the slot information.

If the i-th disk device is a device that cannot read and write normally (step S103;
In N), “1” is added to the counter C _FAIL, and the content of the identification information N _{FA IL} is rewritten to i (step S10).
5). Then, only when the counter C _FAIL is equal to or less than “1” (step S106; Y), the process proceeds to step S107. When the counter C _FAIL is "2" or more (step S106; N), an error message indicating that the disk array device with zero redundancy cannot be operated is output (step S109), and initialization is performed. The process ends.

In step S107, "1" is added to i, and then it is judged whether or not the variable i is less than or equal to the total number N _{ALL of} disk devices (step S108). When the variable i is less than or equal to the total number N _{ALL of} disk devices (Y)
Returns to step S102 in order to write the test and management information to the next disk device.

Whether the counter C _FAIL is "0" when the variable i exceeds the total number N _ALL of disks (step S108; N), that is, when the test and the writing of the management information to all the disk devices are completed. Whether or not it is determined (step S110), and if it is not "0" (N), the disk device with N _FAIL number is excluded from the access target (step S111), and the initialization process is ended. The setting of the disk device excluded from the access target is performed by writing information indicating that the N _FAIL disk device cannot be used in a predetermined storage area of the memory in the disk array controller 13.

As described above, the disk array device of the first embodiment is constructed so that the use can be started even if there is a failure in one disk device, but normally all disk devices are Its use is started in a state where it can be read and written normally.

FIG. 3 schematically shows the contents of the management information storage area of each disk device when all the disk devices are normal (C _{FA IL} = 0) in the initialization processing. In the figure, "UID" written in the storage area 17 of the management information storage area 15 is device identification information, and the numerical values shown in the storage areas 18 and 19 are respectively
It is assumed that these are slot information and history information. As described above, the management information different in only the slot information is written in the management information storage area 15 of each disk device by the initialization processing shown in FIG.

This management information is information used to determine the status of each disk device. In the disk array device of the first embodiment, if a disk device fails during normal operation, the failed disk The contents of the management information storage area of the disk device other than the device are rewritten. This management information update processing is executed by the procedure described below.

FIG. 4 shows the operation flow of the disk array controller during the management information updating process. When the disk array controller receives information indicating that a failure has occurred from the disk device, the disk array controller sets the identification information of the disk device in the identification information N _FAIL (step S201). When the identification information N _FAIL is equal to the total number of disks N _ALL (step S202; Y), "1" is set to the variable i (step S203), and N _FAIL is not equal to the total number of disk devices N _ALL. In step S202; N, i is set to "N _FAIL +1" (step S204)
To do.

After that, the management information is rewritten so that the history information in the management information of the i-th disk device increases by "1" (step S205). In this step,
The management information is read, the history information in the read management information is updated, and the updated management information is written in the management information storage area.

Next, the disk array controller adds "1" to the variable i (step S206), and i is N.
Only when _ALL is exceeded (step S207; Y), i
Is changed to "1" (step S208). Then, the variable i and the identification information N _FAIL are compared, and if i does not match N _FAIL (step S209; N), step S2 is performed to update the management information of the next disk device.
Return to 05. This series of processing is performed by the variable i and the identification information N.
_{When the FAILs} match (step S209; Y), that is, when the management information of the disk device other than the disk device designated by the identification information N _FAIL is updated, the process ends.

[0048] The processing of steps S202 through S204 is a process for determining a disk apparatus of updating the first management information, by these processes, N _ALL
If the No. 1 disk device (“5” in the disk array device of the embodiment) is the failed disk device, the No. 1 disk device is the update target of the management information; otherwise, the failed disk device. The disk device connected to the slot having a slot number that is larger by "1" than the slot number to which the device is connected is the first update target. The processing of steps S206 to S208 is also processing for determining the disk device for which the management information is updated next, according to the same order rule.

FIG. 5 schematically shows that the contents of the management information storage area of each disk device are updated by this management information updating process. When the content of the management information of each disk device is in the state as shown in (a), for example, a disk device with a star mark (disk device No. 2)
When a failure occurs, the history information in the management information of the disk device No. 3 is first rewritten as shown in (b), and then the management information is updated in the order of No. 4, No. 5, No. 1. It is updated, and finally, the state shown in (c) is formed.

This management information is read out when the apparatus is started up and is used for determining whether there is a disk apparatus that should not be accessed or a disk apparatus that may start the recovery process. Since the operation procedure of this state determination process is complicated, before giving a detailed description, the relationship between the content of the management information read from each disk device and the operation content that should be selected by the disk array controller depending on the content. Will be explained.

6 and 7 schematically show the contents of the management information which may be read from each disk device. FIG. 6C shows the contents of each piece of management information when a failure has occurred in the second disk device and the management information update processing has been completed. In this case, the disk array controller should exclude the second disk device from the access target and operate the disk array device as a storage device with zero redundancy. The management information storage area of the second disk device may or may not be readable.

FIG. 6D shows the contents of the management information when the faulty disk device is correctly replaced as indicated by the solid arrow. The management information storage area of the replaced disk device is blank to indicate that the management information is not written. The state of the disk array device in this case is a state in which the recovery process is performed on the second disk device to make all the disk devices available and then to operate. If the exchanged disk device has just been formatted, a predetermined bit pattern (for example, a bit pattern of "E5" in hexadecimal notation) is written in the storage area corresponding to the management information storage area. ing.

FIG. 6 (e) shows an example of the case of erroneous disk device replacement as indicated by the broken line arrow in the state of (c). In this case, the disk array controller should determine the status of the disk array device as inoperable.

Further, when the management information updating process is interrupted, the contents of the management information are different from those in FIG. 6 even though the state of the disk array device is the same.

FIG. 7 shows the contents of the management information of each disk device when the update of the management information is interrupted. here,
It is assumed that the processing is interrupted when the management information of one disk device is updated. 7 (b),
The states of the disk devices in (f) and (g) are the same as the states of the disk devices in FIGS. 6 (c), (d), and (e), respectively. That is, (b) is a state in which the second disk device should be excluded from access targets and operated as a disk array device with zero redundancy,
(D) shows a state in which the disk array device should be operated as the disk array device using the original redundant configuration after the recovery process is performed for the second disk device, and (e) shows the operation of the disk array device. This is a state that should not be allowed, and a state in which the user should be notified that an incorrect exchange has been made.

In the disk array device of the first embodiment,
Such state determination is realized by the following procedure.

FIG. 8 shows the flow of the first operation performed by the disk array controller at startup. When the power is turned on, the disk array controller first initializes the parameters used to determine the status of each disk device (step S301). The parameters initialized in this step are C _FAIL and H _j (j = 1 to 1).
N _ALL ), N _INIT , i, N _UP, and N _DOWN , and C _FAIL , H _j , N _UP, and N _DOWN are set to “0” and N _INIT and i, respectively.
Is initialized to "1". It should be noted that H _j is a parameter mainly used for storing the history information in the management information, and N _INIT is a parameter for storing the slot number of the disk device which is apparently abnormal. . Further, N _UP and N _DOWN are parameters used in the operation described later with reference to FIG.

After the initialization of these parameters, the disk array controller attempts to read the management information of the i-th disk device (step S302), and if it cannot be read normally (step S303; N), Set "-1" to H _i (step S309),
The counter C _FAIL is incremented by “1” and N _INIT is set to i (step S310), and then step S310.
Proceed to 307.

When the management information of the i-th disk device can be read normally (step S303; Y).
Is determined by comparing the device identification information and the slot information in the read management information with the device identification information and the slot information written at the initialization (step S304). Make a decision. Then, if the device identification information and the slot information match the device identification information and the slot information written at the time of initialization, it is determined that the disk device is normally connected (step S305;
Y), the history information in the management information is stored as H _i (step S306), and the process proceeds to step S307.

In the judgment of step S305,
The disk device branched to the “Y” side is not always a normally functioning disk device. That is, even if the disk device is a failed disk device, the connection is correct,
If the contents of the management information storage area can be read, the determination in step S305 is "Y".

If either the device identification information or the slot number is different from the contents written during the initialization processing, it is determined that the disk device is not normally connected (step S305; N). ), "1" is added to the counter C _FAIL , i is set to N _INIT (step S310), and the process proceeds to step S307.

In step S307, "1" is added to i, and if i is equal to or less than the total number N _{ALL of} disk devices (step S308; N), the process returns to step S302.
Processing is performed for the next disk device. When i exceeds the number N _{ALL of} all disk devices (step S308;
Y), that is, after the reading of the management information storage areas of all the disk devices is attempted, this process ends.

When this processing ends, if there is a disk device from which the management information cannot be read, the variable H corresponding to the device becomes "-1", and either the device identification information or the slot information is initialized. If there is a disk device different from the information written at the time of writing, the value of the variable H corresponding to the device remains "0". The value of the variable H corresponding to the exchanged disk device is also
Since the management information of the predetermined pattern is not written in the storage area corresponding to the management information storage area, step S3
In 05, it was judged that it was not connected properly,
It remains “0”.

The counter C _FAIL has a disk device whose H value is "-1" or "0", that is,
The number of disk devices that are certainly not in a normal state is stored, and the variable N _INIT stores the slot number of one of them. If the counter C _FAIL is “0”, the variable N _INIT is rewritten (step S31).
0) is not executed, the value of N _INIT is
The value set in 01 remains "1".

FIG. 9 shows a flow of operations executed by the disk array controller subsequent to the flow of FIG. The disk array controller first determines whether or not the value of the counter C _FAIL obtained in the processing of FIG. 8 is larger than "1" (step S401). This counter C
_As described above, the number of disk devices that are surely not in a normal state is set in _FAIL , so when this value is "2" or more, the disk array device is stored with zero redundancy. It is decided that the device cannot be operated as well. Therefore, the counter C _FAIL
However, if it is larger than "1" (step S401; Y), an error message indicating that the apparatus cannot be started up is output (step S402) and the process ends.

Of the states illustrated in FIGS. 6 and 7, it is determined that the apparatus cannot be started up at the stage when the determination in step S401 is made.
In the states (e) and (g), the management information of the faulty disk device marked with * cannot be read.

Counter C_FAILWhen the value of is less than "1"
(Step S401; N), the content of the variable i is set to N
_INITTo the variable j (step S403)
Value is "1" or more, N_ALL"I + 1" as follows
Or "i + 1-N_ALLSet "(step S40
4) Do. Then, the contents are set by the processing of FIG.
H _jAnd H_iComparison of the size relationship (step S405)
No, h_jIs H_iGreater than N_UPIs "0"
(Step S406; Y), N_UPSet i to
Step S407) and then H_jIs H_iGreater than N_UPBut
If it is not "0" (step S406; N), N_UP
Without changing, the process proceeds to step S410.

If H _j is smaller than H _i and N _DOWN is "0" (step S408; Y), N
I is set to _DOWN (step S409), and H _j is H _i.
If it is larger and N _DOWN is not “0” (step S4)
08; N), the process proceeds to step S410 without changing N _DOWN . Further, in the comparison in step S405, if H _j and H _i match, the process directly proceeds to step S410.

The branching performed in step S406 or S408 is to store the value of the variable i when H _j and H _i first differ from each other in N _UP or N _DOWN . For example, when the management information of the failed disk device can be read in the state of FIG. _6E , N _INIT is set to “3” by the process of FIG. 8, so that in step S405 of FIG. A comparison of H ₄ and H ₃ is made, after which H ₅ and H ₄ , H ₁ and H ₅ , H ₂ and H ₁ ,
The comparison of H ₃ and H ₂ will be performed in this order.

After these five comparisons, step S406
H is branched to_FourAnd H₃When the comparison of
However, when branching to the step S408 side
Is H ₂And H₁When the comparison of₃And H₂
Exists twice when the comparison is made. Such a place
In step S408, N_DOWNIs “0”
Since it has been determined whether or not it is the second branch,
N_DOWNWill not be rewritten, and in the end,
The information at the time of the first branch is N_DOWNTo be remembered in
Become.

In step S410, it is determined whether or not all comparisons have been completed, and if j does not match N _INIT (step S410; N), there are combinations to be compared. Therefore, the content of i is rewritten to j (step S411), and the process returns to step S404.
When the variable j and N _INIT match (step S4)
10; Y), this process ends, and the final determination process described below is executed.

FIG. 10 shows a flow of operations performed by the disk array controller subsequent to the flow of FIG. The disk array controller first determines whether N _UP is "0" (step S501). If N _UP is "0" (Y), all disk devices can operate normally. Then, the startup operation is ended. From this determination, it can be easily understood that all the disk devices can be regarded as normal, because when N _UP is “0”, the history information of all the disk devices is the same. See.

When N _UP is not "0" (step S501; N), it is determined whether the counter C _{FA IL} is "0" (step S502) and it is "0". In the case (Y), the N _UP disk device is excluded from the access targets (step S509), post-processing (step S510) is performed, and then the startup processing ends. The state is determined by the determination in step S502 when the management information storage area of the failed disk device can be read in FIGS. 7B and 6C.

The post-processing performed in step S510 is a processing for completing the management information update processing when the management information update processing is interrupted, and by this processing, for example, FIG. The management information of each disk device in the state of () is corrected to the state of FIG. The actual operation is to restart the management information update process shown in FIG. 4 from the interrupted step (N _DOWN is the information indicating the disk device updated immediately before the interrupt).

When the counter C _FAIL is not "0" (step S502; N), H which is N _UP number H
Determination whether _N is "-1" (step S503)
If it is “−1” (Y), that is, if the disk device cannot read the management information, the N _UP number disk device is specified as the failed disk device (step S509). ), The post-processing (step S510) is performed, and then the startup processing is ended. The state is determined by the determination in step S503 when the reading of the management information storage area of the failed disk device is impossible in the states of (b) and (c).

When H _N is not "-1" (step S503; N), the value of variable j is "1" or more,
"N _DOWN +1" or "N" so that N _ALL or less
_DOWN + 1-N _ALL "is set (step S504). Then, N _UP and j are compared (step S505), and when N _UP and j do not match (step S50).
5; N), the disk device numbered N _UP is specified as the failed disk device. The state is determined by the determination in step S505 when the management information storage area of the failed disk device can be read in the states (f) and (e).

When N _UP and j match (Y)
For this, a test is performed as to whether or not the N _DOWN disk device can normally read and write (step S506). If the disk device is normal (step S
In 507; Y), it is determined that the N _UP disk device is a disk device that has been correctly replaced, and recovery processing is performed (step S508), and the startup processing ends. Of course, in this step S508, the contents of the management information are also set. If the N _DOWN disk device cannot normally read / write (step S507; N),
The N _UP disk device is specified as the failed disk device (step S509), post-processing (step S510) is performed, and then the start-up process is terminated.

It is in the states of (d) and (g) that the determination up to step S507 is performed. These states differ only in the contents of the history information, but have the same relative positional relationship, and cannot be discriminated from the information obtained by the processing up to step S505. The processing performed for this determination is the processing of steps S506 and S507, and an incorrect exchange was made (g).
In the state of N, the N _DOWN disk device corresponds to the faulty disk device, so normal reading and writing cannot be performed, whereas in the state of (d), the N _DOWN disk device is properly replaced. Uses the fact that it can read and write normally to distinguish between them.

Although this disk array device uses management information consisting of device identification information, slot information, and history information, management information consisting of slot information and history information, or management consisting of history information only. Information may be used. When using the management information including the slot information and the history information, the device may be configured so that only the slot information in the management information is compared in step S304 of FIG.

When only history information is used, the range of numerical values used as history information is, for example, 1 to 1.
It is limited to 00, and in step S304 of FIG. 8, it is evaluated whether or not the read history information is within the range, and if it is out of the range, step S305.
In the above, it may be configured to branch to the “N” side.
When replacing a failed disk unit, a disk unit in a formatted state, that is, in a state in which no data is written, is used, so if the contents of the storage area corresponding to the management information storage area are read out as numerical values ,
It does not fall within the above range. Therefore, even by such a judgment, the replaced disk device can be identified, and by comparing with the management information (history information) stored in another disk device, the replacement is correctly performed. It is possible to determine whether or not it is an item. Also, before use for replacement, specific information is written in the management information storage area of the disk device, and the presence or absence of the specific information is detected at startup, and the disk device is replaced. You may make it detect whether or not.

Second embodiment

The disk array device of the first embodiment determines the status of each disk device using only the magnitude relation of the history information in the management information, but the disk array device of the second embodiment does The state of the disk array device is determined by using the two types of history information that have meaning.

The configuration of the disk array device of the second embodiment is the same as that shown in FIG. 1, so the description of the configuration will be omitted. Also, the operation at the time of initialization is almost the same as the operation of the disk device of the first embodiment described with reference to FIG. However, there are only two types of information used as history information in the management information in the disk array device of the second embodiment, one is information written at initialization, and the other is a failure in the disk device. This is the information written at that time. These pieces of history information are used for determination as bit strings, not as numerical values.

FIG. 11 shows the flow of management information update processing in the disk array system of the second embodiment. Upon receiving the information indicating that a failure has occurred from the disk device, the disk array controller sets the identification information of the disk device in the identification information N _FAIL (step S601). When N _FAIL is equal to the total number of disks N _ALL (step S602; Y), "1" is set to the variable i (step S603), and when N _FAIL is not equal to the total number of disk devices N _ALL (step S603). In S602; N), i is set to "N _FAIL +1" (step S604).
To do.

Then, the history information in the management information of the i-th disk device is rewritten to the second history information (step S
605), the process ends. As described above, in the disk array device of the second embodiment, only the contents of the management information of the device next to the failed disk device are rewritten.

In the disk array device of the second embodiment,
At start-up, first, the same processing as that shown in FIG. 8 is performed to read the management information of the i-th disk device,
Further, when the device identification information and the slot information in the management information match the information written at the time of initialization, the content of the history information is stored in H _i . If the management information of the i-th disk device cannot be read, H _i
Is set to "-1" and the device identification information and slot information in the management information do not match the information written at the initialization, "0" is stored in H _i . Counter C
The number of disk devices whose H is "-1" or "0" is stored in _FAIL . Then, the disk array device of the second embodiment uses the information thus obtained to judge the state of each disk device.

FIG. 12 shows the disk array of the second embodiment.
The flow of the determination operation performed by the controller is shown. Processing of FIG.
The disk array controller that has finished
Ta C _FAILIs determined to be greater than “1” (step
S701), if it is larger than “1” (Y),
Error message (step S711) and stand
The lifting operation ends.

When the counter C _FAIL is smaller than "1" (step S701; N), _X where H _X = H _FAIL.
Is searched (step S702). Where H _FAIL is
The second history information written during the management information update processing is shown. If such H _X exists (step S703; Y), the information i for identifying the disk device ordered before the disk device is calculated (step S704). Also, H _X = H
If there is no H _X that becomes _FAIL (step S
703; N), i is set to "0" (step S
705). Next, the disk array controller
A search for j for H _j = 0 is performed (step S706).
In this step, if _j for which H _j = 0 does not exist, “0” is set to j.

When the variable j is "0" (step S7)
07; Y), that is, if there is no disk device whose device identification information or slot information in the management information is different from the information written at initialization, i
It is determined whether or not is "0" (step S708). The fact that the variable i is “0” (Y) indicates that there is no disk device in which the second management information has been written, so the disk array controller determines that all disk devices are normal. Then, the start-up process ends.

If the variable i is not "0" (step S708; N), the i-th disk device is excluded from the access targets (step S709), so that the disk array device is a storage device with zero redundancy. To operate as.

When the variable j is not "0" (step S7)
07; N), the jth disk device is a device that cannot read the management information or the disk device has been replaced, so j and i are compared (step S71).
By performing 0), it is determined whether or not the j-th disk device is the replaced disk device. If the contents of j and i do not match (N), there is a disk device that cannot be used in addition to the jth device which is considered to be the faulty disk device due to the presence of the second management information. Since it exists, the fact that the disk array device cannot be operated is output (step S711), and the startup process is terminated.

If the contents of j and i match (step S710; Y), it means that the correct exchange has been performed, and therefore the recovery process is performed for the i (= j) th disk device ( In step S712), the disk array device is returned to a state in which all the disk devices can be used, and the startup process ends. In step S712, the Xth disk device (that is, the second history information is written in order to store on the disk medium that all the disk devices are normal after the completion of the recovery process. The operation of returning the history information in the management information of the disk device) to the first history information is also performed.

Here, the fact that the state of the disk array device can be accurately determined by this determination operation will be described with reference to the drawings.

FIG. 13 shows the relationship between the state of the disk array device and the contents of the history information, which should be judged by the disk array controller of the second embodiment, and is obtained when the above processing is applied to each disk array device. The values of i and j are shown. In this figure, only the history information in the management information is shown, where “good” indicates the first history information and “fail” indicates the second history information. Further, it is assumed that the history information marked with a star is the history information of the disk device in which the failure has occurred, and the management information storage area can be read.

As shown in FIG. 13A, when the history information of all the disk devices is the first history information "good", the counter C _FAIL becomes "0" and the second history information is obtained. Since the disk device where the information was written cannot be searched,
i becomes "0", and since there is no disk device whose device identification information or slot information is different from the value written at the time of initialization, j also becomes "0". Therefore, in steps S707 and S708, both branches to the "Y" side, and it is determined that all the disk devices are normal.

Further, as shown in (B), when the history information of the disk device No. 3 is rewritten to the second history information "fail" according to the flow shown in FIG. If a disk device failure has occurred in, since the H _X = H _FAIL H _X is present, in step S704, "2" is set to i. Further, since there is no disk device in which the device identification information or the slot information is different from the value written at the time of initialization, j becomes "0". After all, in this case, step S
At 708, it is branched to the “N” side, and the second disk device is excluded from the access target.

Further, when the disk device is replaced from the state of (B) by removing one disk device and installing a new disk device,
As in (C), when the correct exchange is made,
It is conceivable that the wrong replacement is performed as in (D) and (E). In the cases of (C) and (E), i is set to "2" as in the case of (B), but since the replaced disk device exists, j is set to "0". Not
In (C), j is set to "2", and in (E), it is set to "4". Therefore, both are determined in step S710, (C) is determined that the disk device is correctly exchanged, recovery processing is started, and (E) is erroneous exchange. It is judged and an error message is output.

The state (D) is characteristic in that the disk device in which the second history information has been written is mistakenly replaced, but in this case, i is "0".
Then, since j becomes "3", it is branched to the "N" side in step S710, it is judged that an incorrect exchange is made, and an error message is output.
If the contents of the management information storage area marked with a star cannot be read, (D) and (E) are determined to be inoperable in step S701.

Also in the disk array device of the second embodiment, the management information can be composed of slot information and history information, or only history information. For replacement such as removing the disk device and installing a new disk device, the state can always be accurately determined. However, the history information alone cannot identify the case where the disk device that should be connected to the slot No. 5 is connected to the slot No. 4, for example. In a disk array device that needs to be removed, it is desirable to include at least slot information in the management information.

[0100]

As described above in detail, according to the first aspect of the present invention, the failed disk device is specified based on the information written in the disk device at the time of start-up. Information is not lost when the system is stopped.

According to the second aspect of the invention, based on the information written in the disk device, it is determined whether or not there is a disk device that has been replaced with the identification of the failed disk device, and if there is, Since it can be judged whether the exchange is correct or not, it is possible to prevent data loss due to human error.

According to the third aspect of the present invention, since the failed disk device is specified based on the information written in the disk device at the time of start-up, the information regarding the failed disk device can be obtained even if the system is stopped. Never lost.

According to the fourth aspect of the present invention, by writing the two types of management information in the disk device, it is determined whether or not there is a disk device that has been exchanged with the identification of the faulty disk device. Can determine whether the exchange is correct or not, so that data loss due to human error can be prevented.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an outline of a disk array device according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a flow of an initialization operation of the disk array system of the first embodiment.

FIG. 3 is an explanatory diagram showing an outline of management information stored in each disk device at initialization in the disk array device of the first embodiment.

FIG. 4 is a flowchart showing a flow of management information update processing performed when a failure occurs in a disk device in the disk array device of the first embodiment.

FIG. 5 is an explanatory diagram showing how the management information stored in each disk device is updated during the management information updating process in the disk array device of the first embodiment.

FIG. 6 is an explanatory diagram showing how the management information changes in the disk array device of the first embodiment when the disk device is replaced when the system is stopped.

FIG. 7 is an explanatory diagram showing how the management information changes in the disk array device of the first embodiment after the management information update process is interrupted and the disk device is replaced when the system is stopped. is there.

FIG. 8 is a flowchart showing an operation flow at the time of startup of the disk array system of the first embodiment.

9 is a diagram showing the disk array device of the first embodiment as shown in FIG.
6 is a flowchart showing a flow of operations performed subsequently to the flow of FIG.

FIG. 10 is a flow chart showing a flow of operations performed subsequently to the flow of FIG. 9 by the disk array device of the first embodiment.

FIG. 11 is a flowchart showing a flow of management information update processing performed by the disk array device of the second embodiment when a failure occurs in the disk device.

FIG. 12 is a flowchart showing the flow of operations performed at startup by the disk array device of the second embodiment.

FIG. 13 is an explanatory diagram illustrating a determination operation performed by the disk array device according to the second embodiment at startup.

FIG. 14 is a configuration diagram showing an outline of a conventional disk array device.

FIG. 15 is an explanatory diagram showing an outline of a data write operation in the disk array device.

FIG. 16 is a configuration diagram showing an outline of a conventional dual disk device using a nonvolatile memory.

FIG. 17 is a configuration diagram showing an outline of a conventional dual disk device using a history counter.

[Explanation of symbols]

11 ... Disk array device, 12 ... Interface, 1
3 ... Disk array controller, 14, 22 ... Disk device, 15 ... Management information recording area, 16 ... Recording area, 1
7 ... Device identification information storage area, 18 ... Slot information storage area, 19 ... History information storage area, 20 ... Write request data, 21 ... Host computer, 23 ... Control section, 24 ...
Non-volatile storage unit, 25 ... Operating system, 2
6 ... Channel device, 27 ... Central processing unit, 28 ... Main memory, 29 ... History counter unit, 30 ... History counter verification unit, 31 ... Control table

Claims (4)

[Claims] 1. A plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and a predetermined storage area for each of the plurality of disk devices. An initialization unit that writes numerical information of the same content as history information used to determine the state of the disk device, a detection unit that detects that an access failure has occurred in the disk device, and an access failure is detected by this detection unit. Designating means for designating the disk devices other than the disk device one by one in the order set by the setting means, and the history information stored in the predetermined storage area of the disk device designated by the designating means as its value. And a rewriting unit that rewrites the data so that it increases by a predetermined amount, and the data is stored in the predetermined storage area of the plurality of disk devices at startup. Read means for reading the history information, and search means for searching for a disk device having a larger history information of the disk device later ordered by the setting means, based on the history information read by the reading means, A disc array device comprising: a discriminating unit that discriminates the disc unit searched by the unit as a disc unit having an access failure. 2. A plurality of disk devices for storing data, setting means for setting the order in which all of these disk devices are circulated, and a predetermined storage area for each of the plurality of disk devices. Initializing means for writing the slot number to which the disk device is connected and history information which is numerical information as management information used for judging the state of the disk device, and detecting means for detecting that an access failure has occurred in the disk device. Specifying means for specifying the disk devices one by one in the order set by the setting means, excluding the disk device in which the access failure is detected by the detecting means, and the predetermined operation for the disk device specified by the specifying means. Rewriting means for rewriting the history information stored in the storage area of the plurality of units so that the value increases by a predetermined amount; Read-out means for reading out the management information stored in the predetermined storage area of each of the disk devices, and the slot number in the management information read out by the read-out means coincides with the one written by the initialization means. And the history information read from the disk device determined by the determination means to be different from the history information used for writing by the initialization means. Changing means, searching means for searching for a disk device having a larger history information of the disk device ordered by the setting means after the history information corresponding to each disk device, If the history information corresponding to the disk device searched by is changed by the changing means, the disk is changed. A disk array device comprising: a judgment unit that judges that the device is a replaced disk device, and otherwise judges that the disk device is a disk device having an access failure. 3. A plurality of disk devices for storing data, setting means for setting the order in which all these disk devices are circulated, and a predetermined storage area for each of the plurality of disk devices. An initialization unit that writes first history information as history information used to determine the state of the disk device, a detection unit that detects that an access failure has occurred in the disk device, and an access failure is detected by this detection unit. Specifying means for specifying a disk device to be ordered later by the setting means of the disk device; and a second history different from the first history information in the contents of the predetermined storage area of the disk device specified by the specifying means. Rewriting means for rewriting to information and reading information stored in the predetermined storage area of each disk device at startup The reading means and the second reading means based on the information read by the reading means.
Search means for searching the disk device in which the information matching the history information of the disk device is stored, and the disk device previously ordered by the setting means of the disk device searched by the searching means is a disk device having an access failure. A disk array device comprising: a determination unit that determines that 4. A plurality of disk devices for storing data, setting means for setting the order in which all these disk devices are circulated, and a predetermined storage area for each of the plurality of disk devices. An initialization unit that writes first history information as history information used to determine the state of the disk device, a detection unit that detects that an access failure has occurred in the disk device, and an access failure is detected by this detection unit. Specifying means for specifying a disk device to be ordered later by the setting means of the disk device; and a second history different from the first history information in the contents of the predetermined storage area of the disk device specified by the specifying means. Rewriting means for rewriting to information and reading information stored in the predetermined storage area of each disk device at startup The reading means and the second reading means based on the information read by the reading means.
Based on the information read by the reading means, the first specifying means for specifying the disk device previously ordered by the setting means of the disk device storing the information matching the history information of 1
The second specifying means for specifying the disk device that stores the information that does not match the first and second history information and the disk device specified by the first and second specifying means are the same device. Sometimes, the first discriminating unit discriminates that the disc device is a properly exchanged disc device, and the discriminating unit is not discriminated by the second discriminating unit. A disk array device, comprising: a second judging means for judging that the disk device specified by the first specifying means is a disk device having an access failure when the specifying is performed.