JP2001154929A - Management method and system for substituting path system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically detect the event that intermittent faults occur for a long time and to automatically disconnect a fault path from a substituting path. SOLUTION: The information on the operation states of plural paths and the executed result of a test command is gathered for every prescribed cycle, the gathered information is statistically processed and the presence/absence of abnormality in the paths are judged. The path judged as abnormal as the result of judgment is disconnected from the substituting path.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a path switching method of an alternative path system having a plurality of paths including a spare path between a host computer and a file device and a system thereof.

[0002]

2. Description of the Related Art In a large-scale computer system, in order to process a large amount of data at a high speed, a fiber or fiber device such as a file device or a tape device using an optical interface is used.
Channel Arbitrated Loop (Fiber Channe
l Arbitrated Loop (hereinafter referred to as FC-AL). And in general, there are at least two positive and
There are two I / O paths, and the FC-AL control and data I / O processing are performed using the positive I / O path with the secondary I / O path as a spare in normal times. When a failure occurs in the input / output path, the processing is switched to a secondary input / output path for operation, and this configuration is generally called an alternative path configuration.

Conventionally, there is the following disclosure as a measure for improving reliability against intermittent failures of a small-scale file device.

The "file control method" disclosed in Japanese Patent Application Laid-Open No. 5-81109 is an obstacle caused by dust on a medium or a displacement between a track and a head when accessing a magnetic file device or the like.
As a measure to improve reliability before an intermittent failure in which access is successful as a result of retry becomes a fixed failure over time, the number of retries or failures for failures that occurred when accessing each file was performed. Accumulate the number of accesses,
When the number of times exceeds a predetermined value, the data of the corresponding file is copied to a newly allocated file.

The "file manager" disclosed in Japanese Patent Laid-Open No. 7-169135 evaluates the reliability of a magneto-optical disk, which is considered to be less reliable than a magnetic disk, due to aging, and issues an alarm according to the state of deterioration. For the purpose of emitting,
Means for grasping the deterioration state of the magneto-optical disk and deterioration state warning means for giving a warning to the execution program when an abnormal value of deterioration exceeds a reference value are provided.

On the other hand, alternative path management techniques for improving the reliability and processing capacity of a large-scale disk subsystem are disclosed in JP-A-63-104126 and JP-A-5-100801. There is something.

Japanese Patent Application Laid-Open No. 63-104126 discloses an "error retry method for a disk subsystem" for the purpose of controlling the number of unnecessary retries for a failed path. An error counter to be stored is provided, and when the counted-up error counter value becomes equal to or more than a predetermined value, the path to be accessed is switched to an alternative path and a retry operation is repeated to improve the efficiency of error processing.

Japanese Patent Laid-Open Publication No. Hei 5-100801 discloses a "high reliability system for a disk subsystem", which is used for a drive of an off-line processing system having two systems, an active system and a standby system. In order to avoid the high cost of online processing systems that write data and to improve the fault tolerance against drive access failures, correctable data transfer system failures that occur when accessing the current drive are The number of times is counted for each drive, the contents of the current drive in which the number of occurrences exceeds a predetermined value are copied to the spare drive, and the spare drive is used as a substitute for the current drive.

[0009]

In a conventional alternative path management method, an FC-input / output (test I / O) method is used.
Although the operating state of the AL was confirmed, this method was only able to make a binary decision between normal and abnormal. Therefore, even if the detection of a fixed failure and the disconnection of the path in the extension can be realized,
An event in which an intermittent failure occurs for a long time cannot be automatically detected, and the operator needs to periodically monitor the operation state of the system.

The switching operation of the alternative path management method is generally executed in the file system of the host computer, but if a fixed failure occurs in the used positive I / O path,
Switching to the secondary I / O path is automatically performed. However, if an intermittent failure occurs in both the primary and secondary I / O paths, the following problem occurs.

A) When an intermittent failure is detected in the positive I / O path, the path is switched to the secondary I / O path.

B) Subsequently, when an intermittent failure is detected in the secondary I / O path, the path is switched to the positive I / O path.

C) The operations a) and b) are repeated for several minutes.

D) The intermittent failure of the primary and secondary I / O paths is recovered.

E) a) to d) occur intermittently.

When an intermittent failure occurs in such a pattern, the user's I / O request succeeds as a result of the retry operation, but the processing time is delayed and the business program does not end within a predetermined time. There was a problem.

An object of the present invention is to solve the above-mentioned problems, to automatically detect an event in which an intermittent failure occurs for a long time, and to automatically disconnect a failed path from an alternative path. A path management method and system are provided.

[0018]

An alternative path management method according to the present invention collects information on the operation states of a plurality of paths and the execution results of test commands at predetermined intervals, and statistically processes the collected information. To determine whether there is an error in the path.
The path determined to be abnormal is separated from the alternative path.

Information on the operation states of a plurality of paths and the execution results of the test commands are collected by issuing an I / O test command to the file device. If a retry operation occurs due to an I / O error, it is regarded as an intermittent failure. , Retry occurrence information, or a history of I / O events that have become abnormal when data is read from or written to the file device.

Then, retry occurrence information is acquired at a predetermined retry occurrence information acquisition time interval, stored as a record, and the stored statistical information is periodically read out. If the information is continuous, it is determined whether the threshold value has been exceeded after editing the statistical information, the path failure is identified, and an alarm is notified to the operator for the path identified as the failure, and the replacement is performed. By separating from the path, a smooth operation environment of the system can be provided.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a system configuration diagram of one embodiment of the present invention.

The alternative path management method shown in FIG. 1 uses a file system 1 for processing data from a business application 6, a statistical information management device 2 for monitoring input / output operations in the file system 1, and an FC-AL device 4 And a device driver 3 for controlling the file device 5.

The file system 1 receives data requested by the business application 6 and splits / assembles the data.
I / O event management unit 11 that manages schedules and retries, and I that manages I / O issued to device drivers
/ O control unit 12, execution of testI / O, switching of alternative path,
The system includes a path management unit 13 that performs path management such as switching back, and an application program interface (API) unit 14 that connects the file system 1 to the outside.

The input / output test (test I / O) method is a procedure for continuously issuing I / O until a response is received within a certain period of time, and if a response is received from the file device 5, the path of the FC-AL device 4 Is recognized as normal, but the I / O due to the retry is abnormal, and if there is no response, the path is recognized as abnormal and the path is switched to another path.

The statistical information management device 2 has a file system 1
Information acquisition unit 21 that acquires the operation history of I / O events that occurred in the server, statistical information determination unit 22 that analyzes the acquired event information and determines intermittent failures, and controls the separation of paths used by the file system 1 Path control unit 23 that performs
It is composed of a statistical information file 24 for storing statistical information and an alarm notifying unit 25 for notifying an operator of an alarm.

The device driver 3 has a disk controller 31 for managing the file device 5 and an FC-A for managing the FC-AL device 4.
An L control unit 32 is provided.

Next, the operation of the embodiment shown in FIG. 1 will be described in detail with reference to the flowchart of FIG.

The first embodiment shown in FIG. 1 statistically monitors the status of retry operation in test I / O,
This is an example of detecting an event in which an intermittent failure occurs for a long time.

First, the state of a path by test I / O is confirmed (step S1). In the file system 1, the path management unit 13 issues a test I / O to the file device 5 using the device driver 3 in order to check the operation state of the FC-AL device 4, and uses the test I / O to If a retry operation occurs due to a / O error, it is regarded as an intermittent failure,
The retry occurrence information is stored in the path management unit 13.

Next, statistical information is obtained (step S
2). The event information acquisition unit 21 acquires retry occurrence information at one-minute intervals, which is the retry occurrence information acquisition time, from the path management unit 13 of the file system 1, and obtains the statistical information file 24.
And store it as a record.

Next, it is determined whether the threshold value has been exceeded after the statistical information has been edited (step S3). The statistical information determination unit 22 periodically reads out the statistical information stored in the statistical information file 24, and if the retry occurrence information is continuous during the intermittent failure monitoring time of 10 minutes, identifies the path as a failure. I do.

Next, an action for the failed path is executed (step S4). An alarm is raised from the statistical information determination unit 22 to the alarm notification unit 25 for a path recognized as a failure, and an alarm is notified to the operator. Further, the path is separated from the alternative path by the path control unit 23.
This is realized by the path disconnection notification from the path 23 to the path management unit 13 of the file system 1.

Next, a second embodiment of the present invention will be described.

As a second embodiment, there is a method of monitoring the occurrence of an I / O event failure. The operation of this embodiment will be described in detail with reference to the flowchart of FIG.

In this embodiment shown in FIG. 1, an event in which an intermittent failure occurs for a long time is obtained by observing the history of I / O events in the file system and statistically calculating the occurrence of an abnormality in the unit of an alternative path. This is an example of detection.

First, the status of an I / O event handled by the file system is confirmed (step T1). In the file system 1, the I / O event management unit 11 reads and writes data from and to the file device 5 via the device driver 3, and stores a history of I / O events that have become abnormal at this time. The event management unit 11 holds the information.

Next, statistical information is obtained (step T).
2). In the event information acquisition unit 21, the file system 1
I / O event history information is acquired from the I / O event management unit 11 at intervals of one minute, which is the acquisition time of I / O error event occurrence information, and stored as a record in the statistical information file 24.

Next, the statistical information is edited and it is calculated whether the threshold value has been exceeded (step T3). Statistical information judgment unit 22
Then, the statistical information stored in the statistical information file 24 is periodically read out, and the intermittent failure monitoring time is set to 10 minutes.
If the history of I / O events that have caused an I / O error is continuous, the path is identified as a failure.

Next, an action for the failed path is executed (step T4). An alarm is raised from the statistical information determination unit 22 to the alarm notification unit 25 for a path recognized as a failure, and an alarm is notified to the operator. Further, the path is separated from the alternative path by the path control unit 23.
This is realized by a path disconnection notification from 23 to the path management unit of the file system 1.

The present invention is effective not only in the case where the number of the alternative paths is two but also in the case where the number of the alternative paths is three or more, and the same effect can be expected for a tape or a printer as a medium other than the file device.

The first effect of these embodiments is that an intermittent failure of a path can be detected at an early stage and separated from a group of alternative paths. As a result, I / O delay and I / O request due to frequent occurrence of path switching and switching back at the time of I / O abnormality are not retried many times, so that the business program can be processed without delay. . The second effect is that the intermittent failure of the path can be detected and separated at an early stage, so that the I / O performance deterioration due to the data write back has been eliminated. In other words, if a mirror configuration that duplicates data in the file device is adopted, data is written from a normal file device to the failed file device in order to guarantee data consistency after recovery from the failure. In some cases, the performance of business programs is affected by the input / output load. However, according to the present invention,
Intermittent failures of the path can be detected and separated early, eliminating I / O performance degradation due to data writeback.

The third effect is that the occurrence status of the intermittent failure is collected as statistical information, so that the operator can use it for system analysis.

[0043]

According to the present invention, information on the operation states of a plurality of paths and the execution results of test commands are collected at predetermined intervals, and the collected information is statistically processed to determine whether there is an abnormality in the paths. By separating the path determined to be abnormal from the alternative path, an event in which an intermittent failure occurs for a long time can be automatically detected, and the failed path can be automatically separated from the alternative path.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a system configuration according to an embodiment of the present invention.

FIG. 2 is a flowchart of a first embodiment using test I / O.

FIG. 3 is a flowchart of a second embodiment based on monitoring of I / O events.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 File system 2 Statistical information management device 3 Device driver 4 FC-AL device 5 File device 6 Business application 11 I / O event management unit 12 I / O control unit 13 Path management unit 14 API unit 21 Event information acquisition unit 22 Statistical information Judgment unit 23 Path control unit 24 Statistical information file 25 Alarm notification unit 31 Disk control unit 32 FC-AL control unit S1-S4, T1-T4 steps

Claims (6)

[Claims] 1. A path switching management method of an alternative path method having a plurality of paths including a spare path between a host computer and a file device, wherein information on an operation state of the plurality of paths and an execution result of a test command are stored in a predetermined manner. An alternative path, which is collected for each cycle, statistically processes the collected information to determine whether the path is abnormal, and, as a result of the determination, separates the path determined to be abnormal from the alternative path. Management method. 2. An I / O test command is issued to a file device in order to confirm the operation state of the file device. If a retry operation occurs due to an input / output error, the retry operation is regarded as an intermittent failure and the retry occurrence information is determined. A path status check step based on the held I / O test command, a retry occurrence information acquisition step for acquiring retry occurrence information at a predetermined retry occurrence information acquisition time interval, and a statistical information acquisition step for storing as a record, and periodically reading the stored statistical information. If the retry occurrence information is continuous during the predetermined intermittent failure monitoring time, a step for determining whether the threshold value has been exceeded after editing the statistical information and certifying the path as a failure; Alerts the operator to a path that has been certified as 2. The alternative path management method according to claim 1, further comprising: 3. An I / O event status check step for retaining a history of I / O events that have become abnormal when data is read from or written to a file device, and predetermined I / O error event occurrence information I at acquisition time interval
A statistical information obtaining step of obtaining I / O event history information and storing it as a record; and periodically reading out the stored statistical information to obtain an I / O that has an input / output error during a predetermined intermittent fault monitoring time. If the event history is continuous, the statistical information is edited, and a failure detection step of the path for calculating whether or not the threshold has been exceeded, and an alarm is notified to the operator for the path that has been recognized as a failure, 2. The alternative path management method according to claim 1, further comprising the step of: performing an action on the failed path, separating from the alternative path. 4. A predetermined I / O error event occurrence information acquisition time interval is one minute, and a predetermined intermittent fault monitoring time is one.
4. The alternative path management method according to claim 2, wherein the time is set to 0 minutes. 5. A statistical information management device for monitoring an input / output operation in a file system for processing data from a business application, and a spare computer between a file computer and a host computer of a system including a device driver for controlling the file device. In an alternative path management system of an alternative path method having a plurality of paths including a path, receiving data requested from a business application,
Manage division / assembly, schedule, retry, etc.
The I / O issued to the device driver is managed, and the I / O is continuously issued until a response is received within a certain period of time. If there is a response from the file device, the path is recognized as normal, but the I / O due to the retry fails. If there is an error or no response, the path is recognized as abnormal, the file system that switches to another path, and the operation history of I / O events that have occurred in the file system are acquired, and the acquired event information is analyzed to determine an intermittent failure. And a statistical information management device that controls disconnection of a path used by the file system, stores statistical information, and notifies an operator of an alarm, and a device driver that manages the file device and the high-speed file device. An alternative path management system characterized by the following. 6. An I / O event management unit for receiving data requested from a business application and managing division / assembly, schedule, retry, etc., and manages I / O issued to a device driver. I / O to do
The controller and the control unit continue to issue I / O until a response is received within a certain period of time. If there is a response from the file device, the path is recognized as normal. A path management unit for recognizing an abnormality and switching to another path;
An event information acquisition unit for acquiring an operation history of an O event; a statistical information determination unit for analyzing the acquired event information to determine an intermittent failure; and a path control unit for controlling disconnection of a path used by the file system. , A statistical information file for storing statistical information, and an alarm notifying unit for notifying an operator of an alarm. The device driver includes a disk control unit for managing the file device and an FC-AL control unit for managing the high-speed file device. The alternative path management system according to claim 5, comprising: