JPH04315238A - Access route blocking system - Google Patents

Abstract

PURPOSE:To block an access route to a disk storage device connected under the control of a disk control adapter as a common part when a fault is detected. CONSTITUTION:The addresses to be selected of disk control adapters 4-6 set switching in advance are referred to by sub system controllers 1 and 2, and the route of access to the disk control adapters 4-6 is blocked by a block flip- flop 10. Then, the block flip-flop 10 is set by an instruction from the sub system controllers 1 and 2, and the set value of this block flip-flop can be referred to by the sub system controllers 1 and 2.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention provides an access path blockage system that blocks the access path when a device having a write control section fails in an external storage subsystem having a plurality of access paths between it and a central processing unit. It is related to the method.

0002

2. Description of the Related Art Conventionally, when accessing an external storage device that has multiple access paths, if a failure occurs, the OS (
In the input/output control of the operating system, the input/output channel command or input/output channel program is automatically retried several times, and if the retries are unsuccessful, the input/output from this access path is automatically retried. The output channel program is controlled not to be executed.

[0003] The access route in this case is the route between the channel device and the external storage device, and what is blocked is the access from that route to the external storage device where the failure has occurred; Access to other external storage devices is allowed even if they belong to the same path. When multiple external storage devices are connected under a single disk control adapter to control reading and writing of data, if a failure occurs in the common disk control adapter, each time one of the external storage devices is accessed, The I/O channel program is reported to have terminated with an error. The OS will sequentially block access routes to these external storage devices.

0004

The problem with the conventional access path blocking method described above lies in the nature of the failure of the disk control adapter as a common part. For example, in a disk storage device used as an external storage device, a failure of the write control circuit may erase the record area that the user attempts to access or make it unreadable. Even if the OS receives the error report and tries again using another route, the data has already been destroyed. There was a problem in that this problem would be reproduced every time the access path related to the disk control adapter in question was used.

[0005]

[Means for Solving the Problems] The access route blocking method of the present invention connects a plurality of disk control adapters in an imozuru style under a subsystem controller that decodes input/output channel commands and directs processing, and controls individual disks. In the subsystem that controls reading and writing of data for the disk storage device to which multiple disk storage devices are connected to the adapter, the subsystem controller refers to the selected address of the disk control adapter whose switch is set in advance. and a occlusion flip-flop that means occlusion of the access path in the disk control adapter, so that it can be set by an instruction from the subsystem controller, and the set value of this occlusion flip-flop can be referenced to the subsystem controller. When the subsystem controller executes an input/output channel command and identifies that a permanent failure has occurred in the path of the same disk control adapter, it sets the blocking flip-flop and executes the input/output channel command. When receiving subsequent input/output channel commands, if it is set by referring to this blocking flip-flop, an error completion report is made and the access path related to this disk control adapter is blocked. This is how it was done.

[0006] Furthermore, in the access path blocking method according to another aspect of the present invention, when the blocking flip-flop provided in the disk control adapter is set, the driver circuit of the disk device interface section is disabled. This gates the drive state to reduce the possibility that a failure in the disk control adapter will affect the disk device.

[0007]

[Operation] In the present invention, by blocking the disk control adapter in response to a fixed write failure, the access path from the subsystem controller is blocked as a result.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, 1 and 2 are subsystem controllers that decode input/output channel commands and direct processing, 3 is an access path, 4, 5, and 6 are disk control adapters connected in a similar manner under the subsystem controller,
7 is an interface section, 8 is an address switch, 9 is a status generation section, 10 is a blocking flip-flop for blocking the access path 3 to the disk control adapter 4;
1 is a read/write (R/W) control unit, 12 is a disk device interface unit, 13-1 to 13-n and 14-1.
-14-n are disk devices. Here, the interface section 7, the address switch 8, and the status generation section 9 constitute a means by which the subsystem controller can refer to the selected address of the disk control adapter whose switch is set in advance. The generation unit 9 and the occlusion flip-flop 10 are
It constitutes means that can be set in response to an instruction from the subsystem controller 1, and that the set value of the blocking flip-flop 10 can be referred to by the subsystem controller.

When the subsystem controller executes an input/output channel command and identifies that a permanent failure has occurred in the path of the same disk control adapter, it sets the blocking flip-flop 10 and executes the input/output channel command. When receiving subsequent input/output channel commands, if the blocking flip-flop 10 is set, an error completion report is made and the access path related to this disk control adapter is blocked. is configured to do so. Furthermore, when the blocking flip-flop 10 provided in the disk control adapter is set, the driver circuit of the disk device interface unit 12 is gated to be in a non-driving state, thereby reducing the influence of a failure in the disk control adapter. The structure is configured to reduce the possibility that the problem will reach the disk device side.

FIG. 2 is a flowchart for explaining the operation of FIG. 1, and is a control flowchart of the subsystem controller shown in FIG. And each step 101-1
14, respective predetermined processes are executed.

Next, the operation of the embodiment shown in FIG. 1 will be explained with reference to FIG. 2. First, when the central processing unit issues an input/output channel command, the subsystem controller 1 first places an address code for selecting a desired disk device on the access path 3. Each of the disk control adapters 4 to 6 then checks a part of this address code against its own address, which is set in advance by the address switch 8. and,
One of the disk control adapters 4 to 6 with a matching address is selected and can respond to the subsystem controller 1. This selected disk control adapter 4
accesses one of the disk devices 13-1 . . . 13-n specified by the remaining part of the address code according to an instruction from the subsystem controller 1. Then write as an input/output channel command (WRITE)
For example, if there is a failure in the read/write control unit 11 of the disk control adapter 4 when attempting to execute a system command, and the write current is improperly sent to the read/write head, this disk control adapter If a write-related command is executed via 4, the record to be accessed or an adjacent record may be written and erased by an illegal write current. At this time, the details of the failure detected on the read/write control section 11 or the disk device side are held in the latch circuit of the status generation section 9. The disk control adapter 4 is
If the status generation unit 9 holds the details of the failure, the occurrence of the failure is communicated to the subsystem controller 1 via the interface unit 7. Further, the subsystem controller 1 interrupts the execution of the input/output channel command being executed and reports the end of the error to the central processing unit.

Next, the OS of the central processing unit tries to recover from the error and once again follows the same access path 3, that is, via the subsystem controller 1 and the disk control adapter 4, executes the input/output channel program to the desired disk device. Try again or take another route, i.e.
The input/output channel program is re-executed on the desired disk device via the subsystem controller 2. However, as mentioned above, if the record format has been erased, the target record may not be accessible, or even if it can be accessed, re-execution will fail, or even if re-execution is successful, adjacent This means that complete recovery is not possible if the record being written has been erased. Therefore, if left as is, the same event will occur again when another disk device is accessed, and the problem, which can be called record destruction, will spread. Here, if it is possible to prevent access from the disk control adapter in which the problem has occurred, the recovery process of the OS allows access using a path that does not cause problems, and the spread of the failure can be suppressed.

FIG. 2 is a flowchart of the process for realizing this. When there is an input/output request from the central processing unit, the subsystem controller 1 first selects the desired disk device 13-1 including the disk control adapter 4, and then sends the request from the status generation unit 9 of the disk control adapter 4. Read the status (step 101). This status generation unit 9 is provided with a blocking flip-flop 10 that can be set from the subsystem controller 1.
This is done so that a logic output value of 0 is reflected in the status. Further, this status generating section 9 is arranged so that the set value of the address switch 8 is also reflected in one of the statuses.

When the blocking flip-flop 10 is in its initial state, its logical output value is "0". Therefore, first, the subsystem controller 1 refers to the status contents, confirms that the logical output value of this blocking flip-flop 10 is "0" (step 102), and executes the input/output command (step 103). Here, to explain the case where the input/output command is executed normally, if it is not a write command (step 104), the disk device 13-1 does not detect the write current (step 112), so the subsystem controller 1 is normally executed. A completion report is sent to the central processing unit (step 114).

If it is a write command (step 104), the subsystem controller 1 recognizes the address of the disk control adapter 4 based on the setting value of the address switch 8 read from the status generator 9, The control table provided in the subsystem controller 1 corresponding to this address value is referred to (step 105). This control table includes information on a "write failure flag," which is used to determine whether a write command has caused a write failure twice in a row. In the example described, since this is a normal case, no write failure is detected in either the disk control adapter 4 or the disk device 13-1 (step 106), and the "write failure flag" in the control table is reset. (Step 113), and a completion report is made (Step 114).

Next, a description will be given of processing when a write failure is detected. First, when a write current is detected during execution of a command other than a write command (step 112)
, the subsystem controller 1 sets the blocking flip-flop 10 provided in the disk control adapter 4 (
Step 108), reports the error to the central processing unit and ends (Step 110). If a write current is detected even though it is not a write command, there is a high possibility that the recording on the recording medium of the disk device is disturbed, so the blocking flip-flop 10 is immediately set. When the blocking flip-flop 10 is set (step 102), the subsystem controller 1 reports an error to the central processing unit and ends the process without executing an input/output command (step 111). If the failure is caused by the failure, at least the influence of the failure on other disk devices can be avoided.

Next, if a failure is detected in the execution of the write command (WRITE) (step 106
), the subsystem controller 1 determines whether a failure has already been detected in the execution of the previous write instruction.
The determination is made by referring to the write failure flag in the control table provided for (step 107). Then, if the write failure flag is in the reset state (OFF), this write failure flag is set (step 109),
Report the error to the central processing unit and end (step 11)
0). Here, if the write failure flag is set (ON), it means that a write failure has been detected twice in a row, and it is determined that the failure is a fixed and unrecoverable failure, and the blocking flip-flop 10 is set. (Step 108
), reports the error to the central processing unit, and exits.

The above-mentioned fault details can be known by the subsystem controller 1 reading the status generated by the status generator 9 of the disk control adapter 4 or the disk device 13-1, so the fault details can be detected by reading the status generated by the disk device 13-1. It is only necessary to control the write failure flag for things that may affect other users. Further, even if an instruction other than another write instruction is executed between write instructions, the write failure flag is controlled so that it can be updated only with respect to the write instruction. Therefore, even if a write instruction is executed and a write error is detected, the write error flag remains set until the next write instruction is received even if other instructions are executed. In other words, focusing only on write commands, if a write failure is detected continuously, it is determined that it is a fixed failure and the blockage flip-flop 10 is set, and if no write failure is detected, the write failure flag has already been set. Even if it is set, the write failure flag is determined to be an intermittent failure, ie, recovered by re-execution of the write command by the OS, and the write failure flag is reset.

Next, the subsystem controller 1 not only refers to the blocking flip-flop 10, but also controls the disk control adapter 4 and the disk device 13-1 by its output signal when the blocking flip-flop 10 is set.
If a gate circuit is constructed to drive the driver circuit (not shown) of the disk device interface unit 12 that drives the interface signal between the disk drive interface unit 13-n and the disk drive interface unit 13-n, in other words, to put it in an inactive state, at least the disk control It is possible to reduce the possibility that a failure inside the adapter 4 side will affect the operation of the disk device side. For example, depending on the specifications of the interface, if a disk device becomes active (selected) with that interface by driving a specific signal, access from other connection paths (subsystem controller 2) It is also possible that the device cannot respond to messages or appears to be busy.

[0020]

Effects of the Invention As explained above, the present invention makes it possible to block the access path from the subsystem controller by blocking the disk control adapter in response to a fixed write failure. This has the effect of preventing write failures related to the path from spreading to multiple disk devices. and,
Even in this case, the system does not become immediately unusable because the central processing unit OS can access it through other subsystem controllers instead of the blocked access path. Further, it is possible to access a disk device under a disk control adapter other than the disk control adapter in which the blocking flip-flop is set via the subsystem controller.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a control flowchart of the subsystem controller shown in FIG. 1;

[Explanation of symbols]

1, 2 Subsystem controller 3 Access paths 4 to 6 Disk control adapter 7 Interface section 8 Address switch 9 Status generation section 10 Blocking flip-flop 11 Read/write control section 12 Disk device interface section 13-1 to 13
-n disk device 14-1 to 14-n disk device

Claims (2)

[Claims] [Claim 1] A plurality of disk control adapters are connected in an imozuru style under a subsystem controller that decodes input/output channel commands and directs processing, and a plurality of disk storage devices are connected to each disk control adapter under this controller. In a subsystem that controls reading and writing of data with respect to a disk storage device, there is provided a means by which a subsystem controller can refer to a selected address of a disk control adapter whose switch is set in advance, and a means for blocking an access path to the disk control adapter. A blockage flip-flop means that the blockage flip-flop can be set according to an instruction from the subsystem controller, and a means for referring to the set value of the blockage flip-flop to the subsystem controller, and the subsystem controller can receive input/output channel commands. When it is identified that a fixed failure has occurred in the path of the same disk control adapter, the blocking flip-flop is set and an error completion report is made for the input/output channel command, and subsequent input/output is disabled. When an output channel command is received, the access path blocking method is characterized in that, if the blocking flip-flop is set, an error termination report is made with reference to the blocking flip-flop, and the access path related to this disk control adapter is blocked. 2. In the access path blocking method according to claim 1, when the blocking flip-flop provided in the disk control adapter is set, the driver circuit of the disk device interface unit is gated to be in a non-driving state. An access route blocking method characterized in that the possibility that a failure in a disk control adapter affects the disk device side is reduced in this way.