JPH02195420A - Error recovery processing system for common magnetic disk device - Google Patents

Abstract

PURPOSE:To evade the system down of the whole system even when a fault is generated in a specific host and to improve the reliability of the system by detecting a defective path and a defective disk, releasing the reserve state of the defective host and continuing processing. CONSTITUTION:When specific disks 14, 15 are in the reserve state by the system down of a system 8, all starts from systems 1 to 7 are device busy. When a device-free interruption is invalid at the time of common operation of the systems, a defective path is retrieved by a path information control table 18. When a using flag 21 is not ON, whether a reserve flag 22 is ON or not is discriminated, and when the flag 22 is ON, the defective path 11 and the defective disks 14, 15 are detected. When the systems are not in the common operation, the device busy state can be released by issuing an UR command.

Description

[Detailed Description of the Invention] [Summary] Regarding an error recovery processing method in a shared magnetic disk device, the present invention relates to an error recovery processing method for a shared magnetic disk device, which can improve system reliability by avoiding system down of the entire system even if a failure occurs in a specific host. The purpose of this technology is to provide an error recovery processing method for magnetic disk devices, and it is a magnetic disk control device that controls multiple disks and has a connection path that connects to multiple hosts and a director unit that controls data transfer. , a path information management table having a connection host identification section, a device address, an in-use flag, and a reserve flag is provided between the connection path and the director section, and a failed path and a failed disk are identified by searching the path information management table. It was configured to detect the failed host, remove it from the reserved state, and continue processing.

[Industrial Application Field] The present invention relates to an error recovery processing method in a shared magnetic disk device.

In large-scale information processing systems, a plurality of magnetic disk devices are shared and operated by a large number of hosts.

In this case, if the reserved state of the magnetic disk unit continues forever due to a failure of a specific host, when accessed by a normal host, the magnetic disk unit will respond with a device bus response and become permanently unusable. becomes.

Therefore, in such a case, it is necessary to release the reserved state of the faulty host to make the corresponding magnetic disk device usable.

[Prior Art] As a conventional magnetic disk device, there is one shown in FIG. 5, for example.

In FIG. 5, 1 is a connection path connected to a plurality of hosts (systems), 2 and 3 are mutually independent director units, and 4 is a disk controlled by the director units 2 and 3.

The disk 4 is shared by a large number of hosts, and when it is reserved by a specific host, all activations from other hosts result in the device being busy.

[Problems to be Solved by the Invention] However, in such conventional magnetic disk drives, path management is performed on a per-director unit basis, and there is no function for transmitting information between director units or between paths. During shared operation, if a failure occurs on a specific host, the disk (device) becomes permanently reserved.
Other hosts are rarely able to use the disk. That is, as shown in FIG. 6, there is a problem that an infinite loop occurs and the entire system eventually goes down.

The present invention has been made in view of these conventional problems, and is a shared system that can avoid system failure of the entire system and improve system reliability even if a failure occurs in a specific host. The purpose of this paper is to provide an error recovery processing method for magnetic disk devices.

[Means to solve the problem] FIG. 1 is a diagram explaining the principle of the present invention.

In FIG. 1, 11 is a connection path connected to a plurality of hosts, 12.13 is a director unit that controls data transfer, and 1B is the connection path 11 and the director unit 12.13.
13, is a path information management table having connected host identification, device address, in-use flag, and reserve flag.

[Operation] In the present invention, a path information management table is provided between the connection path and the director unit, and when a failure occurs in a specific host during shared operation, the path information management table is searched and the in-use flag is set. When it is off and the reserve flag is on, a faulty path or faulty disk is detected, so issuing the LJR command clears the device busy state and allows booting from the host on Sen. System processing can continue.

As a result, it is possible to prevent a failure of a specific host from spreading to other systems, and the reliability of the system can be improved.

[Example] Embodiments of the present invention will be described below based on the drawings.

FIGS. 2 to 4 are diagrams showing one embodiment of the present invention.

In FIG. 2, reference numeral 11 indicates connection paths 12 and 13 that are connected to multiple hosts (systems 1 to 8), and 14 and 15 indicate director units that control data transfer. The disk is controlled via

Here, as shown in FIG. 3, a path information management table 1 is provided between the connection path 11 and the director section 12.
8 is provided.

The path information management table 18 includes a connected host identification section 19, a device address 20, an in-use flag 21, and a reserve flag 22.

Therefore, it is possible to monitor and control the path usage status within the director section 12.13 and between the director sections 12.13. Note that the connection path 11, the path information management table 18, and the director section 12.1
3 constitutes a magnetic disk control 311 device as a whole.

Next, the operation will be explained based on the flowchart shown in FIG.

First, assume that a system down occurs in the system 8 in step S1. Disk 14.1 in step S2
When disk 5 is not in the reserved state, there is no effect due to the system down of system 8 (see step S3), but if a specific disk 14 or 15 is in the reserved state due to system down due to system 8, the system down from systems 1 to 7 is All startups result in a device busy state (see step S4).This device busy state is not permanently released unless a system reset is issued from the faulty system 8.

Next, in step S5, it is determined whether the system is in shared operation. If it is not a shared operation, select U in step S6.
R (Unconditional Re5erve)
By issuing a command, the device busy state can be released, and in step S7, it becomes possible to start up from systems 1 to 7, and system processing can be continued.

On the other hand, when the system is in shared operation, it is determined in step S8 whether or not device-free interrupts are possible, and if possible, the process advances to step S7 to continue system processing. If device-free interrupts are not possible, time monitoring is performed for 3 minutes, for example, in step S9. If a device-free interrupt cannot be performed even after time monitoring, a search for a faulty path is performed using the path information management table 18 in step S10.

Next, determine whether the in-use flag 21 is on,
When the in-use flag 21 is on, the process advances to step 312, and the process returns to step S10, assuming that the system is in operation.

If the in-use flag 21 is not on, step S13
It is determined whether the reserve flag 22 is on or not, and if it is not on, the process proceeds to step 314, and returns to step S10 as the system is not in operation. Reserve flag 22
is on, the faulty path 11 and faulty disk 14°15 are detected in step S15.

In other words, the in-use flag 21 is turned on only when the host is in use and there is access within a certain period of time (for example, 10 minutes), and it is turned off when there is no access due to a system down, etc., so the previous j host As described above, when the in-use flag 21 is off and the reserve flag 22 is on, it is determined that the faulty path 11 and faulty disks 14 and 15 are present due to system failure.

When the faulty path 11 and faulty disk 14.15 are detected, the system is in a state equivalent to not being shared, so the process advances to step S6, issues the LIR command, releases the device busy state, and In step S7, it becomes possible to start up from systems 1 to 7, so system processing continues.

Note that such search and recovery processing for the failed path 11 is executed by the system that first detected the error among the normal systems 1 to 7.

[Effects of the Invention] As described above, according to the present invention, the faulty path and faulty disk are detected as soon as a fault occurs on a specific host during shared system operation, so that the fault is not caused by other faults. It is possible to prevent the problem from spreading to the system, and system processing can be continued. The conclusion,
System reliability can be improved.

[Brief explanation of the drawing]

FIG. 1 is a diagram explaining the principle of the present invention, FIG. 2 is a block diagram showing an embodiment of the present invention, FIG. 3 is a diagram showing a path information management table, and FIG. 4 is a flowchart for explaining the operation. FIG. 5 is a block diagram showing a conventional example, and FIG. 6 is a conventional flow chart. In the figure, 11...Connection path, 12.13...Director section, 14.15...Disk, 16.17...Adapter, 18...Path information management table, 1...Connection Host identification part, 20... Device address, 21... In use flag, 22... Reserved flag. Heather sales! The Moon's Rhinoceros Face Sun Moon Decoy Figure 1 Subtitles: Figure 5 Figure 4 Water 70-F Figure 6

Claims (1)

[Claims] In a magnetic disk control device that controls a plurality of disks (14) and has a connection path (11) connected to a plurality of hosts and a director section (12), (13) that controls data transfer, the connection A path information management table (18) having a connection host identification section, a device address, an in-use flag, and a reserve flag is provided between the path (11) and the director sections (12) and (13) to manage the path information. An error recovery method for a shared magnetic disk device, characterized in that a failed path and a failed disk are detected by searching a table (18), the reserved state of the failed host is released, and processing is continued.