JP2001306262A - Method for controlling information processing system and information processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To maintain the function of a uni-directional circulation interface by specifying and excluding any defective device from plural devices connected to a uni-directional circulation interface. SOLUTION: Plural magnetic disks 2501-250n and a disk controller A (2101) are connected through a fiber switch 2104 by FC-AL interfaces 2103 and 2601-260n, and the input and output of each magnetic disk can be interrupted by by-pass switches 2301-230n in this disk sub-system. When a certain magnetic disk (250x) reports an interface disconnection failure, the disk controller A (2101) judges the magnetic disk which is not interrupted by the by-pass switch among the magnetic disks positioned at the upstream of the magnetic disk as a defective disk, and interrupts the input and output of the magnetic disk by the by-pass switch so that the functions of the FA-AL interfaces 2103 and 2601-260n can be recovered.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing system and a control technique thereof, and more particularly to an information processing system for controlling a plurality of devices using a unidirectional circular interface such as an FC-AL interface. The present invention relates to technology that is effective when applied to a processing system.

[0002]

2. Description of the Related Art In recent years, an internal magnetic disk interface in a disk subsystem has been replaced by a faster SCSI-A.
We are moving to the L interface. However FC
-In the AL interface, since the connection form of the magnetic disks is in a ring shape called a loop circuit, a failure has occurred such that one magnetic disk on the loop circuit electrically or optically disconnects the loop circuit. In this case, all the connected magnetic disks become unusable.

Japanese Patent Laid-Open Publication No. Hei 11-510933 discloses a data storage system using a fiber channel as a connection interface, in which individual disk drives are doubly connected to a plurality of controllers via a plurality of fiber channels and switches. If the Fiber Channel becomes unusable due to the failure of the disk drive connected to one Fiber Channel, the switch connects the disk drive connected to the failed Fiber Channel to another healthy fiber There is disclosed a technology that enables replacement of a failed disk drive by switching connection to a channel side without stopping disk drives other than the failed disk drive.

[0004]

However, in the above-mentioned prior art, a technique for preventing a loop circuit failure beforehand by separating and foreseeing in advance a magnetic disk which may fail and cause a loop circuit disconnection failure,
Further, when a loop circuit disconnection failure actually occurs, there is no disclosure of a technique for specifying where the failed magnetic disk is located on the loop circuit.

That is, in recent disk subsystems, the number of magnetic disk devices constituting a system is steadily increasing due to a disk array technology using a small magnetic disk device. Preventive maintenance by predicting a failure and quick identification of a failed magnetic disk device are becoming more and more important in terms of improving the operation reliability and efficient operation of the disk subsystem.

SUMMARY OF THE INVENTION It is an object of the present invention to realize a quick failure recovery by quickly and accurately specifying a failed device in an information processing system in which a plurality of devices are connected using a unidirectional circulation interface. .

Another object of the present invention is to realize preventive maintenance by predicting the occurrence of a failure of each device in an information processing system in which a plurality of devices are connected using a unidirectional circulation interface. is there.

Another object of the present invention is to realize an operation continuation by localizing a failure part in an information processing system in which a plurality of devices are connected by using a unidirectional circulation interface.

Another object of the present invention is to provide a disk subsystem in which a plurality of disk drives are connected and controlled by a one-way circular interface, and when any one of the disk drives fails, a loop circuit disconnection failure occurs. It is an object of the present invention to provide a technique for recovering a loop circuit failure by specifying and eliminating a position on a loop circuit of a drive.

Another object of the present invention is to specify a disk drive which can fail in the near future and cause a loop circuit disconnection failure in a disk subsystem in which a plurality of disk drives are connected and controlled by a unidirectional circular interface. It is to provide a technique for eliminating before a failure occurs.

[0011]

SUMMARY OF THE INVENTION According to the present invention, at least one first device and a plurality of second devices for transmitting and receiving information between the first device are provided. In a control method of an information processing system configured to be interconnected by a circulating loop-shaped first interface, each of the second
Switch means for selectively connecting and disconnecting the second device to the first interface is provided between the first device and the first interface, and any one of the second devices is provided by the switch means. By selectively connecting to the first interface and diagnosing the failure, the first device identifies the second device having a failure.

When an error is detected at the first interface in any second device, the second device located upstream of the second device in the first interface is switched to the first device by the switch means. Disconnect from the interface.

In each of the second devices, an error occurrence state relating to the first interface is stored, and when the error occurrence state stored in any of the second devices deviates from a predetermined allowable range, the second error is detected. In the one interface, a second device located upstream of the second device is separated from the first interface by a switch.

And a control logic for that purpose.

[0015] Alternatively, in addition to the above-described configuration, there is provided a switch means and a second interface connected to at least one of the first and second devices, and a second interface provided by the switch means via the second interface. The connection and disconnection of the device or the collection of information on the above-mentioned error occurrence state is performed.

More specifically, as an example, a disk subsystem which is an example of the information processing system of the present invention includes:
A cache memory for temporarily storing input / output data exchanged with the host computer, a channel controller having an input / output port for the host computer, and controlling an input / output interface with the host computer, , A disk controller that controls a plurality of storage devices via a unidirectional circulation interface such as a fiber channel, and a data input / output block between the disk controller and the plurality of storage devices. A fiber switch, a shared memory storing an interface between the channel controller and the disk controller, a channel controller and a cache memory, a channel controller and a shared memory, or a disk controller and a cache memory. And a cache switch for controlling the input and output data be between or between the disk controller and the shared memory with.

The storage device has a function of accumulating and storing faults occurring at the interface with the disk controller, and there has been a request to disclose the integrated value at the interface with the disk controller. Sometimes has a function to disclose, and also has a function to report a failure on the unidirectional circular interface or another dedicated interface at the time of loop circuit disconnection failure, and to report a failure on the dedicated interface when the integrated value exceeds a certain number. Alternatively, when the integrated value exceeds a certain number, data input / output of a storage device located upstream of its own storage device may be cut off by the fiber switch.

The disk controller has a function of interrupting data input / output of a storage device located upstream from the storage device by the fiber switch when a failure is reported from the storage device. And a function of interrupting the input / output of data of the storage device or the storage device located upstream from the storage device by the fiber switch depending on the content of the report.

The processing for obtaining the integrated value of the interface fault may be executed in synchronization with an instruction from the host computer, or may be periodically and asynchronously executed.

[0020]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a conceptual diagram of the overall configuration of a disk sub shim, which is an example of an information processing system that implements a method for controlling an information processing system according to an embodiment of the present invention.

In the disk subsystem of this embodiment, p channel controllers A (1101 to 11)
0p) and channel controller B (1201 to 12)
0p) is connected to one or more host computers 1001.

The m pairs of disk controllers A (1131)
To 113 m) and the disk controller B (1231 to
123m) are fiber switches A (1141-1), respectively.
14m) and fiber switch B (1241-124)
m) through one or more storage devices 1301
~ 130m. The input / output interface between the disk controller and the storage device is FC-A.
L interfaces 1151 to 115m and FC-AL interfaces 1251 to 125m are used.

Input / output data exchanged with the host computer 1001 is transmitted to the cache memories 1111 and 1211.
12 is temporarily stored. Further, redundancy is provided by using two cache memories.

The cache memories 1111 and 1211 and the channel controllers A (1101 to 110p) are connected via cache switches A (1121 and 1122), and the cache memories 1111 and 1211 and the channel controllers B (1201 to 120p) are connected to the cache switches B (1221, 1222). The cache memories 1111 and 1211 and the disk controller A (1131 to 113m) are connected via the cache switch A (1121 and 1122), and are connected to the disk controller B (1231 to 123m) via the cache switch B (1221 and 1222). Connected.

Channel controller A (1101 to 11)
0p) and the channel controller B (1201 to 120)
p) and disk controller A (1131-113m)
Communication between the disk controller B (1231 to 123m) is performed via the shared memories 1112 and 1212. The shared memories 1112 and 1212 are connected to the channel controllers A (1101 to 110p) via the cache switches A (1121 and 1122), and are connected to the channel controllers B (1201 to 120p) via the cache switches B (1221 and 1222). Is done. The shared memories 1112 and 1212 are connected to the disk controllers A (1131 to 113m) via the cache switches A (1121 and 1122), and are connected to the disk controllers B (1231 to 123m) via the cache switches B (1221 and 1222). It is connected.

When the channel controller A (1101) receives a data read request from the host computer 1001, the channel controller A (1101) writes "read data" to the shared memory 1112 or 1212 via the cache switch A (1121 or 1122). Or a plurality of storage devices 1301 to 13 in which
0m, and write the data to the cache ”instruction.

Next, the shared memory 1 checks whether there is an instruction to input / output data to / from a storage device connected to the shared memory 1.
The disk controller A (1131 to 113m) or the disk controller B (1231 to 123m) that cyclically monitors the data 112 and 1212 reads "data from one or a plurality of storage devices 1301 to 130m where read data exists" in the shared memory 1112 or 1212. Read and write to cache "instruction is written, the storage device 1301-1
FC-AL interface 11 connected to 30m
It is confirmed whether 51-115m is not registered as a failure port in the shared memories 1112 and 1212, and if it is not registered as a failure port, the storage device 1301-1 is checked.
Fiber switch A (1141 to 114
m) or fiber switch B (1241-124)
m), read the data via cache switch A
(1121, 1122) or cache switch B (1
221 and 1222) via the cache memory 111.
1, 1211 and then cache switch A (1
121 or 1122) or cache switch B
(1221 or 1222) via the shared memory 111
Write a “read completed” report to 2 or 1212.

Next, the channel controller A (1101), which cyclically monitors the shared memories 1112, 1212 via the cache switch A (1121, 1122), determines whether or not the reading has been completed. When the report is recognized, the request data in the cache memory 1111 or 1211 is transferred to the host computer 1001 via the cache switch A (1121 or 1122).

When the channel controller B (1201) receives a data write request from the host computer 1001, the channel controller B (1201) sends the host computer 1001 to the cache memories 1111 and 1211 via the cache switch B (1221 and 1222). After writing the data of the shared memory 1 via the cache switch B (1221 or 1222).
112 or 1212, "Data in cache is 1
Or, write an instruction to “write to a plurality of storage devices 1301 to 130m”.

Next, the shared memory 1 is checked for an instruction to input / output data to / from a storage device connected thereto.
The disk controller A (1131 to 113m) or the disk controller B (1231 to 123m) that cyclically monitors the 112 and 1212 writes "data in the cache to one or more storage devices 1301 to 130m in the shared memory 1112 or 1212. , The data is read from the cache memory 1111 or 1211 via the cache switch A (1121 or 1122) or the cache switch B (1221 or 1222).
It is checked whether the FC-AL interface connected to the specified storage device is registered in the shared memory 1112 or 1212 as a failed port. If the FC-AL interface is not registered as a failed port, the storage devices 1301 to 1313
Data is written to 0m via the fiber switch A (1141 to 114m) or the fiber switch B (1241 to 124m).

FIG. 2 is a conceptual diagram illustrating in more detail a system configuration under a pair of disk controllers in the disk subsystem illustrated in FIG.

The disk controller A (2101) is 1
FC-AL interface 210 for unidirectional transmission of pairs
3 is connected to the FC-AL interface connector 2105 of the fiber switch 2104, and n bypass switches 230 are provided in the fiber switch 2104.
1 to 230n, each of which has a magnetic disk 2 via FC-AL interfaces 2601 to 260n.
501 to 250n.

Each of the bypass switches 2301 to 2301
230n is connected to the switching bus 2106, and the disk controller A (2101) is connected to the switching bus 2106 via the switching bus interface 2102. By controlling the switching bus 2106, each of the bypass switches 2301 to 230
0n by changing the bypass state of the magnetic disk 25.
Signal flows to the FC-AL interfaces 2601 to 2602n connected to 01 to 250n can be individually cut off. The bypass switch 230 of FIG.
1, 230n are in a non-bypass state, the bypass switch 23
02 indicates a bypass state.

Similarly, the disk controller B (220
1) The FC of the fiber switch 2204 via the FC-AL interface 2203 that performs a pair of unidirectional transmissions
Connected to the AL interface connector 2205, and the fiber switch 2204 includes n bypass switches 2401 to 240n, each of which is FC-
They are connected to magnetic disks 2501 to 250n via AL interfaces 2701 to 270n. Each of the bypass switches 2401 to 240n is connected to a switching bus 2206, and the disk controller B (2201) is connected to the switching bus interface 22.
02 is connected to the switching bus 2206 via the H.02, and by controlling the switching bus 2206, the bypass state of each of the bypass switches 2401 to 240n is changed to connect the FC-AL interfaces 2701 to 270n to the magnetic disks 2501 to 250n. Can be individually blocked.

FIG. 4 shows the disk subsystem of the present embodiment exemplified in FIGS. 1 and 2 or a modified example of FIG. 1 and FIG. 3 described below, immediately after starting up or before operation after replacing a failed part. It is a flowchart which shows an example of a diagnosis.

First, in step 4001, the disk controller A (2101) controls the switching bus 2106, closes all the bypass switches 2301 to 230n, and enters the bypass state (all the magnetic disks 2501 to 25
0n is disconnected). Next, step 4002
Selects the first bypass switch 2301, sets the bypass switch 2301 to the non-bypass state (the corresponding magnetic disk 2501 is connected) in the subsequent step 4003, and initializes the loop circuit with the FC-AL interface in the step 4004. LIP (Loop
Issue InitializationPrimitive-sequence).

Next, in step 4005, it is determined whether or not the initialization of the loop circuit has been successful and the link up state has been established. If Link Up is not performed, a loop circuit is not established and data cannot be transmitted / received to / from the magnetic disk 2501. Therefore, in step 4101, the FC-AL interface 2 connected to the bypass switch 2301
601 is registered in the internal memory 2107 as a failure port, the bypass switch 2301 is closed in step 4010 to be in the bypass state, and then, in step 4011, the diagnosis of the next bypass switch 2302 is started. If step 4
If the link-up state is reached in 005, the next step 400
6 to arbitrate the transfer with the magnetic disk 2501
N_PORT L described in 23.4 of SI FC-PH-2 Rev7.4
Issue OGIN and Process LOGIN described in 21.10. The following two LOs issued in step 4007
It is checked whether the GIN has been completed normally and is ready to be transferred to the magnetic disk 2501. If the processing is not completed normally, data cannot be transferred to and from the magnetic disk 2501.
FC-AL interface 2601 connected to 301
Is registered in the internal memory 2107 as a failure port, the bypass switch 2301 is closed in step 4010 to be in a bypass state, and in step 4011 the next bypass switch 2302 is diagnosed. If the process ends normally in step 4007, the process proceeds to step 4008 where ANSI SCSI-3
In described in 5.1.1 of Primary Commands x3.301
A quiry command is issued, and in the subsequent step 4009, it is checked whether or not the processing has been completed normally. If the processing is not completed normally, the FC-AL interface 2601 connected to the bypass switch 2301 is registered as a failure port in the internal memory 2107 in Step 4101, and Step 40 is performed.
In step 10, the bypass switch 2301 is closed to enter a bypass state.
Move on to diagnosis 2. If it is determined in step 4009 that the process has been completed normally, the flow advances to step 4011 to diagnose the next bypass switch 2302.

In this way, the n-th bypass switch 2
When the diagnosis is completed up to 30n (step 4012), in step 4013, only the bypass switch that is not registered as a failure port in the internal memory 2107 is opened to set a non-bypass state (the corresponding magnetic disk is connected), and
In step 4014, the cache switches 1121, 11
The faulty port registered in the internal memory 2107 is re-registered in the shared memories 1112 and 1212 via 22.

The same diagnostic test is performed on the disk controller B (2201).

As described above, according to the disk subsystem of the present embodiment, the FC-AL
Of a plurality of devices (magnetic disks 2501 to 250n) connected to the unidirectional circular interface such as the interfaces 2103 and 2203, the device that has caused the interface disconnection failure is quickly and accurately specified,
By eliminating it, the function of the unidirectional circular interface can be maintained.

FIG. 5 shows a loop circuit disconnection fault in the FC-AL interface which occurs during operation in the disk subsystem of the present embodiment having the configuration of FIGS. 1 and 2 or FIG. 1 and a modification of FIG. 9 is a flowchart illustrating an example of a process for eliminating the exclusion.

When the device connected to the FC-AL interface detects that the loop circuit has a disconnection failure, Loop Initialization-Loop Failure; Valid AL_
Issue PA Primitive Sequence (hereinafter abbreviated as LIPf) (ANSI FC-AL Rev5.7 7.8.4).

FIG. 9 shows a model diagram when a loop circuit disconnection fault occurs. Disk controller 900
0 and devices 1 to 5 (9001 to 9005) are FC-A
They are connected by an L interface. Defined in 16.3.1 of ANSI FC-PH Rev4.3 when the loop circuit is established and there is no data exchange between devices
A signal called IDLE is circulating on the FC-AL interface, but the device 2 (9002) and the device 3 (90
Since the device 3 (9002) is disconnected due to the failure of the device 2 (9002) during the period 03), the device 3 (9003) cannot receive all the signals including the IDLE signal transmitted from the device 2 (9002). If the device 3 (9003) does not receive a signal (including the IDLE signal) from the device 2 (9002) for a certain period of time, it will send a LIPf (shown as LIPf8, 1d) including an AL_PA indicating an address on its own loop. 4 (9004) located downstream of
Send to The device 4 (9004) that has received the LIPf from the device 3 (9003) transmits the received LIPf to the device 5 (9005) located downstream of itself. In this manner, the disk controller 9000 can receive the LIPf transmitted by the device 3 (9003), and can specify the transmitting device by the AL_PA included in the LIPf. Device 2 (9
002) is eliminated (bypassed), and the device 1 (9001) and the device 3 (9003) are connected using the bypass line (9010) (in this embodiment, the bypass switch is set to the bypass state). If so, loop circuit faults can be eliminated.

The magnetic disk x (250x: 1 ≦ x) shown in FIG.
.Ltoreq.n) cannot receive a signal from the upstream for a certain period of time, and determines that a loop circuit failure has occurred and issues a LIPf. In step 5001, the disk controller A (2101) detects the LIPf issued by the magnetic disk 250x, and
A device located upstream of the magnetic disk 250x that issued f is obtained. In the following step 5002, if the disk controller A (2101) is located upstream of the magnetic disk 250x instead of the magnetic disk, it is determined that the FC-AL interface 2103 has failed, and in step 5101 the cache switch 1121,
FC to shared memory 1112 and 1212 via 1122
-Register the failure of the AL interface 2103. If it is determined in step 5002 that the device located one upstream from the magnetic disk 250x is the magnetic disk (250x-1), the disk controller A (210)
1) In step 5003, a magnetic disk (250x-
Check the bypass state of the bypass switch (230x-1) connected to 1).
F connected to this bypass switch (230x-1)
It is determined that the failure of the C-AL interface (260x-1) or the magnetic disk (250x-1) causes a loop circuit disconnection failure, and
By closing the bypass switch (230x-1) in 005 and setting the bypass state, the magnetic disk (250x-
1) FC-AL interface (260x
-1) is removed from the loop circuit, and the shared memory 1112 is
1212 FC-AL interface (260x-1)
Is registered (step 5006). If the bypass switch (230x-1) is in the bypass state in step 5003, the device x- is located further upstream.
2 is checked (step 5004, step 5002).
Similar processing is performed when the disk controller B (2201) detects LIPf.

As described above, according to the present embodiment, when a disconnection failure occurs in a part of the FC-AL interfaces 2103 and 2203 during operation, the device (magnetic disk) that causes the interface disconnection failure is specified. , The information transfer function of the FC-AL interface can be maintained.

FIG. 6 shows the disk subsystem according to the present embodiment shown in FIGS. 1 and 2 or a modification of FIG. 1 and FIG. 3 described later, which is described in ANSI FC-PH Rev4.3 and 21.4.11. 6 is a flowchart showing an example of the operation of a preventive maintenance function using an RLS (Read Link Error Status Block) function.

The disk controller A (2101) starts diagnosis from the magnetic disk 250y (y = 1) in step 6001, and in step 6002,
It is checked whether the state of the bypass switch 230y to which y is connected is in the non-bypass state and the FC-AL interface 260y is not registered as a failed port in the shared memories 1112 and 1212. At 6003, an RLS command is issued to the magnetic disk 250y. Next, step 6004
It is checked whether there is a response to the command issued in step 2. If there is no response, in step 6103 the FC-AL interface 2 connected to the magnetic disk 250y
It is determined that the failure is 60x (x = y), the bypass switch 230x is closed in step 6010 to put it in the bypass state, and then the cache switch 112
After registering the FC-AL interface 260x as a failure port in the shared memories 1112 and 1212 via the devices 1 and 1122, the process proceeds to step 6101 to diagnose the next magnetic disk drive. If there is a response in step 6004, in step 6005, the link error shown in ANSI FC-PH Rev4.3 Figure83 as illustrated in FIG.
r Check the contents of Status Block Format. Link Error
Loss of Syn among the failures indicated by Status Block Format
chronization Count or Invalid Transmission Word
If the number exceeds the allowable number of times unique to each vendor, it is determined that a device located upstream in the future may cause an FC-AL interface disconnection failure, and the device located upstream from the magnetic disk 250y in step 6006. Examine device attributes. Step 60
Magnetic disk 250 which exceeded the allowable number of times in 07
If it is determined that the device located one upstream from the y is the disk controller A (2101),
-Determine that the failure is in the AL interface 2103, register the failure in the FC-AL interface 2103 in the shared memories 1112 and 1212 via the cache switches 1121 and 1122 in step 6104, and end the processing. If it is determined in step 6007 that the device located upstream is not the disk controller A (2101),
In the following step 6008, it is confirmed whether or not the bypass switch 230x connected to the magnetic disk 250x (x = y-1) located upstream of the magnetic disk 250y is in the bypass state. If the device is in the bypass state, the flow shifts to the determination of a further upstream device (step 6009). If it is determined in step 6008 that the state is the non-bypass state, the bypass switch 230x is closed to set the bypass state (step 6010), and the FC-AL interface 2
The shared memory 1112, 12
12 and proceed to the next magnetic disk diagnosis process (step 6101). If the last magnetic disk 260
When the diagnosis of n is completed, the processing ends (step 610).
2).

If it is determined in step 6002 that the magnetic disk 25
0y is not determined to be valid, or the Error St reported by the magnetic disk 250y in step 6005 is
Loss of Synch among failures indicated by atusBlock Format
ronization Count or Invalid Transmission Word
If both are less than the vendor-specific allowable number of times, the process proceeds to diagnosis of the next magnetic disk 250y + 1 (step 6101). If the diagnosis has been completed up to the magnetic disk 250n (step 6102), the process ends.

As described above, in the present embodiment, the disk controller detects the FC-A detected by each magnetic disk.
The failure occurrence information of the L interfaces 2103 and 2203 is collected and the FC-AL interface 210 is operated during operation.
3 and 2203, a device (magnetic disk) that may cause an interface disconnection failure is identified and eliminated in advance, thereby enabling preventive maintenance.
It is possible to realize the maintenance of the function and the improvement of the reliability. Also, by setting the vendor-specific allowable number arbitrarily, accurate preventive maintenance can be performed according to various reliability levels required for individual disk subsystems.

FIG. 3 is a conceptual diagram showing a modified example of the configuration under a pair of disk controllers in the disk subsystem illustrated in FIG.

The disk controller A (3101) is 1
FC-AL interface 310 for unidirectional transmission of pairs
3 is connected to the FC-AL interface connector 3105 of the fiber switch 3104, and the fiber switch 3104 has n bypass switches 330
1 to 330 n, each of which is provided with a magnetic disk 3 via FC-AL interfaces 3601 to 360 n.
501 to 350n. Each of the bypass switches 3301 to 330n is connected to the switching bus 31.
06 and the disk controller A (31
01) is connected to the switching bus 3106 via the switching bus interface 3102, and by controlling the switching bus 3106, the bypass state of each of the bypass switches 3301 to 330n can be changed.

The magnetic disks 3501 to 350n are also connected to the switching bus 3106 by switching bus interfaces 3801 to 380n, and the magnetic disks 3501 to 350n are also connected to the switching bus 310.
6 by controlling the bypass switches 3301 through 330
n can be changed, and the switching bus 3
Via the disk controller A (3101)
And can communicate. The bypass switch 3 of FIG.
Numerals 301 and 303n indicate a non-bypass state, and a bypass switch 3302 indicates a bypass state.

Similarly, the disk controller B (320)
1) The FC of the fiber switch 3204 via the FC-AL interface 3203 which performs a pair of unidirectional transmissions
-The fiber switch 3204 is connected to the AL interface connector 3205, and includes n bypass switches 3401 to 340n, each of which is an FC-
They are connected to magnetic disks 3501 to 350n via AL interfaces 3701 to 370n. Each of the bypass switches 3401 to 340n is connected to the switching bus 3206, and the disk controller B (3201) is connected to the switching bus interface 32.
The switching state of each of the bypass switches (3401 to 340 n) can be changed by controlling the switching bus 3206 by connecting to the switching bus 3206 via the switching bus 3.

The magnetic disks 3501 to 350n are also connected to the switching bus 3206 by switching bus interfaces 3901 to 390n, and the magnetic disks 3501 to 350n are also connected to the switching bus 320.
6 by controlling the bypass switches 3401-340
n can be changed, and the switching bus 3
Via the disk controller A (3101)
And can communicate.

FIG. 7 shows an example of a process for eliminating a loop circuit disconnection fault in the FC-AL interface which occurred during operation in the disk subsystem of the present embodiment using the configuration of the modification of FIGS. It is a flowchart.

The magnetic disks 3501 to 350n are FC-
When the AL interface disconnection failure is detected, the aforementioned LIP
f and the switching bus 310 via the switching bus interfaces 3801 to 380n.
6 and the disk controller A (3101)
Notify the occurrence of a loop failure.

When the disk controller A (3101) detects the FC-AL interface disconnection failure notification of the magnetic disk x (x: 1 ≦ x ≦ n) via the switching bus 3106 in Step 7001, the disk controller A (3101) that has sent the failure notification. The attribute of the device located upstream of (350x) is checked, and the magnetic disk (350x) is determined in step 7002.
x) If the disk controller A (3101) is not located upstream of the magnetic disk but is located upstream of x), it is determined in step 7101 that a failure has occurred in the FC-AL interface 3103, and the shared memory 1112 is passed through the cache switches 1121 and 1122. , 1212, the failure of the FC-AL interface 3103 is registered. If step 7
If the device located one upstream from the magnetic disk (350x) in 002 is the magnetic disk x-1 (350x-1), the disk controller A (3101) determines in step 7003 the magnetic disk x-1 (350x-1). )
The bypass state of the bypass switch (330x-1) connected to is checked. If the bypass state is not the bypass state, the bypass switch (330x-1) is determined in step 7005.
FC-AL interface (360x-
1) Or, it is determined that a failure has occurred in the magnetic disk (350x-1), the bypass switch (330x-1) is closed to put it into a bypass state, and in step 7006, the shared memories 1112, 1212 are connected via the cache switches 1121, 1122.
12, a failure of the FC-AL interface (360x-1) is registered. If the bypass switch (330x-1) is in the bypass state in Step 7003, the device x-2 further upstream is similarly checked (Step 70).
04, step 7002). Similar processing is performed when the disk controller B (3201) detects a loop circuit failure report.

As described above, according to the present embodiment, if a disconnection failure occurs in a part of the FC-AL interfaces 2103 and 2203 during operation, a failure report is made to notify the device ( By identifying and excluding the magnetic disks, the information transfer function of the FC-AL interface can be maintained, and the individual magnetic disks can be connected via the FC-AL interfaces 2103 and 2203 and another switching bus 3106 or 3206. From the FC-AL interface 2103, 220
The operation reliability is improved as compared with the case where the failure report is made using the 3 itself.

FIG. 8 is a flowchart showing an example of the operation of the preventive maintenance function of the magnetic disk in the disk subsystem using the configuration of the modification of FIG. 1 and FIG.

Magnetic disk y (350y) (y: 1 ≦ y)
.Ltoreq.n) has the above-mentioned RLS function. When a failure having the content indicated by the Link Error Status Format described above is detected in the FC-AL interface (360y) in step 8001, each failure is determined according to its type. The integrated value is updated, and it is determined in step 8002 whether the integrated value of the failure exceeds the vendor-specific allowable value. If the integrated value does not exceed the allowable value, the process is terminated. Device x higher than yourself in 8003
(X = y-1) is obtained. If the device x is a disk controller (step 8004), the magnetic disk y (350y) accesses the switching bus 3106 via the switching bus interface (380y), and the failure of the disk controller A (3101) is unique to the vendor. Notifies that the value is equal to or greater than the integrated value. If the device x is not a disk controller in step 8004, the device accesses the switching bus 3106 to check whether the bypass switch (360x) is open (step 8005). If it is open, the bypass switch (360
x) is closed, and a failure is reported to the disk controller A (3101) in the subsequent step 8008. If the bypass switch is closed in step 8005 and the device is already in the bypass state, a device x (x = x-1) on the further upstream side is obtained in step 8006, and the process returns to the device determination process (step 8004).

The FC-AL interface (370x)
The same processing is performed when a failure occurs in.

As described above, in the present embodiment, the FC-AL interface 210 detected on each magnetic disk
3 and 2203, the FC-AL interfaces 2103 and 223 are operated while each magnetic disk is operating.
Preventive maintenance by predicting the disconnection failure of No. 03, identifying in advance other devices (magnetic disks) that may cause an interface disconnection failure, and operating and removing the bypass switch via the switching bus interface Becomes possible, and the FC-AL interface 210
3, 2203 can be maintained and reliability can be improved. Also, by setting the vendor-specific allowable number arbitrarily, accurate preventive maintenance can be performed according to various reliability levels required for individual disk subsystems.

In the example shown in FIG. 8, when a failure detected by an individual device (magnetic disk) exceeds a vendor-specific permissible value, a bypass switch of another device on the upstream side is closed to enter a bypass state. However, you may just report this to the disk controller.

The invention described in the claims of the present application is expressed in a different way as follows.

<1> A channel having a cache memory for temporarily storing input / output data exchanged with the host computer, an input / output port for the host computer, and controlling an input / output interface with the host computer. A controller, a plurality of storage devices for storing input / output data, a disk controller for controlling the plurality of storage devices, and a switch between the disk controller and the plurality of storage devices for interrupting data input / output A circuit, a shared memory for storing an interface between the channel controller and the disk controller,
A cache switch for controlling input / output data between the channel controller and the cache memory, between the channel controller and the shared memory, between the disk controller and the cache memory, or between the disk controller and the shared memory; The disk controller and the storage device are connected by a unidirectional circulation interface, and the storage device and the disk controller are set in a one-to-one state by using the switch circuit to diagnose each storage device. sub-system.

<2> A disk subsystem characterized in that the functions described in item <1> are performed during a pre-operation diagnosis, such as at system startup.

<3> The configuration of the disk subsystem according to item <1>, which has a function of notifying the disk controller via the unidirectional circular interface when the storage device detects a unidirectional circular interface disconnection failure. When the disk controller receives the unidirectional circular interface disconnection failure report, the disk controller uses the switch circuit to shut off the input / output port of the storage device located upstream of the storage device and whose input / output port is not interrupted. A disk system comprising:

<4> The disk subsystem according to item <3>, further including a dedicated bidirectional communication interface between the disk controller and the storage device, wherein the storage device is a unidirectional communication interface. When detecting a circular interface disconnection failure, the disk controller has a function of reporting to the disk controller via the bidirectional communication interface, and the disk controller is located upstream of the storage device when receiving the unidirectional circular interface disconnection failure report, and A disk system having a function of interrupting an input / output port of a storage device whose input / output port is not interrupted by using the switch circuit.

<5> The configuration of the disk subsystem according to item <1>, wherein the storage device has a function of accumulating a failure such as an out-of-synchronization of the unidirectional circular interface, incomplete data reception, and a unidirectional circular interface. The disk controller has a function of requesting the storage device to report the integrated value via a one-way interface, and when the integrated value exceeds a certain number. A disk subsystem having a function of using the switch circuit to block an input / output port of a storage device that is located upstream of the storage device and whose input / output port is not blocked.

<6> A disk subsystem characterized in that the functions described in item <5> are performed in synchronization with a command from a host computer.

<7> A disk subsystem characterized in that the functions described in the item <5> are performed asynchronously without synchronizing with instructions of a host computer.

<8> The configuration of the disk subsystem according to item <4>, wherein the storage device has a function of integrating faults such as a one-way circular interface out-of-sync or incomplete data reception and a fixed number of integrated values. Has a function of reporting to the disk controller via the two-way communication interface when the number exceeds the predetermined number. A disk subsystem which has a function of shutting down the input / output port of the storage device located upstream of the storage device and whose input / output port is not interrupted by using the switch circuit when the report is made.

<9> The disk subsystem configuration according to item <4>, wherein the storage device, the disk controller, and the switching circuit are connected by a bidirectional communication interface, and the storage device is synchronized with a unidirectional cyclic interface. It has a function of integrating faults such as shifts, incomplete data reception, etc., and when the integrated value exceeds a certain number, the input / output of a storage device located upstream of the storage device and whose input / output port is not blocked A disk subsystem having a function of shutting off a port using the switch circuit and reporting the fact to the disk controller via the bidirectional communication interface.

Although the invention made by the present inventor has been specifically described based on the embodiment, the present invention is not limited to the above embodiment, and various modifications can be made without departing from the gist of the invention. Needless to say, there is.

For example, the information processing system is not limited to the disk subsystem exemplified in the above-described embodiment, but is widely applied to a system in which a plurality of devices are connected using a unidirectional circulation interface such as a fiber channel. Can be applied.

[0077]

According to the present invention, a plurality of devices are connected to FC-A
In an information processing system configured to be connected using a unidirectional circulation interface such as an L interface, an effect is obtained in that quick and accurate identification of a failed device can realize quick failure recovery.

According to the present invention, a plurality of devices are connected to the FC-AL
In an information processing system configured to be connected using a unidirectional circulation interface such as an interface, an effect is obtained that preventive maintenance can be realized by predicting the occurrence of a failure in each device.

According to the present invention, a plurality of devices are connected to the FC-AL
In an information processing system configured to be connected using a unidirectional circulation interface such as an interface, an effect is obtained that operation can be continued by localizing a failure portion.

According to the present invention, in a disk subsystem in which a plurality of disk drives are connected and controlled by the FC-AL interface, when one of the disk drives fails and a loop circuit disconnection failure occurs, the disk drive is disconnected. By specifying and eliminating the position on the loop circuit, it is possible to recover the loop circuit failure.

According to the present invention, in a disk subsystem in which a plurality of disk drives are connected and controlled by an FC-AL interface, a disk drive which is likely to fail in the near future and cause a loop circuit disconnection failure is identified in advance. The advantage is that it can be eliminated before getting up.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram illustrating an example of a configuration of a disk subsystem as an example of an information processing system according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram illustrating an example of a configuration under a disk controller of a disk subsystem as an example of an information processing system according to an embodiment of the present invention.

FIG. 3 is a conceptual diagram showing an example of a configuration under a disk controller of a disk subsystem which is a modification of the information processing system according to one embodiment of the present invention.

FIG. 4 is a flowchart illustrating an example of a storage device diagnosis process in a disk subsystem as an example of an information processing system according to an embodiment of the present invention.

FIG. 5 is a flowchart illustrating an example of a process of removing a failed part when an interface disconnection failure occurs during operation in a disk subsystem that is an example of an information processing system according to an embodiment of the present invention.

FIG. 6 is a flowchart illustrating an example of processing for performing preventive maintenance using an RLS command in a disk subsystem as an example of an information processing system according to an embodiment of the present invention.

FIG. 7 is a flowchart illustrating an example of a process of removing a failed part when an interface disconnection failure occurs during operation in a disk subsystem that is a modification of the information processing system according to an embodiment of the present invention.

FIG. 8 is a flowchart illustrating an example of processing for performing preventive maintenance using an RLS function in a disk subsystem that is a modification of the information processing system according to an embodiment of the present invention.

FIG. 9 is a conceptual diagram showing an operation model when a link circuit failure occurs in a device connected to the FC-AL interface.

FIG. 10 is a conceptual diagram showing an example of a Link Error Status Block format corresponding to an RLS (Read Link Error Status Block) command.

[Explanation of symbols]

1001: Host computer, 1002: Disk subsystem (information processing system), 1101 to 110p
... channel controllers A, 1111 ... cache memories, 1112 ... shared memories, 1211, 1122 ... cache switches A, 1131-113m ... disk controllers A (first device), 1141-114m ... fiber switches A, 1151-115m ... FC AL interface (first interface), 1201 to
120p: Channel controller B, 1211: Cache memory, 1212: Shared memory, 1221, 122
2. Cache switch B, 1231 to 123m Disk controller B (first device), 1241 to 124
m: Fiber switch B, 1251 to 125m: FC
-AL interface (first interface), 1
301 to 130 m storage device (second device), 21
01 ... Disk controller A (first device), 210
2 switching bus interface (second interface), 2103 FC-AL interface (first interface), 2104 fiber switch, 2105 FC-AL interface connector, 2
106 switching bus (second interface), 2107 internal memory, 2201 disk controller B (first device), 2202 switching bus interface (second interface), 220
3 FC-AL interface (first interface) 2204 Fiber switch 2205 FC
-AL interface connector, 2206: switching bus (second interface), 2207: internal memory, 2301 to 230n: bypass switch, 240
1 to 240n ... bypass switch, 2501 to 250n
... Magnetic disk (second device), 2601 to 260n
FC-AL interface (first interface), 2701 to 270n FC-AL interface (first interface), 3101 disk controller A (first device), 3102 switching bus interface (second interface), 310
3 FC-AL interface (first interface) 3104 Fiber switch 3105 FC
-AL interface connector, 3106 switching bus (second interface), 3107 internal memory, 3201 disk controller B (first device), 3202 switching bus interface (second interface), 3203 FC-AL Interface (first interface), 3204: Fiber switch, 3205: FC-AL interface connector, 3206: Switching bus (second interface), 3207: Internal memory, 3301 to 33
0n: Bypass switch, 3401-340n: Bypass switch, 3501-350n: Magnetic disk (second
3601-360n... FC-AL interface (first interface), 3701-370n
... FC-AL interface (first interface), 3801-380n ... switching bus interface (second interface), 3901-390
n: switching bus interface (second interface), 9000: disk controller (first device), 9001 to 9005: FC-AL interface device (second device), 9010: bypass line.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06F 13/00 301 G06F 13/00 301M 13/14 310 13/14 310D (72) Inventor Takao Sato Kanagawa 2880 Kozu, Odawara-shi F-term in Hitachi, Ltd. Storage Systems Division (Reference)

Claims (12)

[Claims] An information processing system comprising: at least one first device; and a plurality of second devices that exchange information between the first device and a plurality of second devices. A method of controlling an information processing system configured to be interconnected by one interface, wherein the first interface of the second device is provided between each of the second devices and the first interface. Switch means capable of selectively connecting and disconnecting the first device, and selectively connecting any one of the second devices to the first interface with the switch device to diagnose the first device. A method for controlling an information processing system, comprising: identifying a second device that has failed due to another device. 2. A method according to claim 1, wherein the at least one first device and a plurality of second devices for transmitting and receiving information between the first device and the second device are arranged in a loop-like shape in which the information circulates in a single direction. A method of controlling an information processing system configured to be interconnected by one interface, wherein the first interface of the second device is provided between each of the second devices and the first interface. A switch means capable of selectively connecting and disconnecting the first device from the first device, and detecting an error in the first interface in any of the second devices at the upstream of the second device in the first interface. A method for controlling an information processing system, characterized in that the second device located on the side is separated from the first interface by the switch means. 3. A loop-shaped loop in which the information circulates in a single direction, comprising at least one first device and a plurality of second devices for exchanging information with the first device. A method of controlling an information processing system configured to be interconnected by one interface, wherein the first interface of the second device is provided between each of the second devices and the first interface. A switch means capable of selectively connecting and disconnecting the first device from the first device, storing an error occurrence state related to the first interface in each of the second devices, and storing the error stored in any of the second devices. When the occurrence state deviates from a predetermined allowable range, the second device located upstream of the second device on the first interface is switched to the switch means. Control method for an information processing system, characterized in that disconnecting from said first interface Te. 4. A loop-like loop in which said information circulates in a single direction, comprising at least one first device and a plurality of second devices for exchanging information with said first device. A control method of an information processing system having a configuration in which the information processing systems are interconnected by one interface, wherein the first device is configured to detect the first interface in the first interface detected by each of the second devices. A method for controlling an information processing system, comprising: collecting an error before a disconnection failure of an interface to predict the second device that causes the disconnection failure. 5. The control method for an information processing system according to claim 1, wherein each of the second devices is initialized by the first device on the first interface based on success or failure. A method for controlling an information processing system, comprising: diagnosing whether the second device is normal. 6. The control method for an information processing system according to claim 1, 2 or 3, wherein said switch means and at least one of said first and second devices are connected to a second interface separate from said first interface.
A control method of the information processing system, wherein the first or second device connects and disconnects the second device by the switch means. 7. The information processing system control method according to claim 1, 2, 3, 4, 5, or 6, wherein each of said second devices is connected to a second interface different from said first interface. And connecting the first device to the first device, and notifying the first device of error information regarding the first interface obtained by the second device via the second interface. Control method for an information processing system. 8. A connection between at least one first device, a plurality of second devices for exchanging information with the first device, and the first device and the second device. A first interface in the form of a loop in which the information circulates in a single direction; switch means for selectively connecting and disconnecting each of the second devices to and from the first interface; Control logic for selectively connecting any one of the second devices to the first interface to perform a diagnostic operation of each of the second devices by the first device. An information processing system, comprising: 9. A connection between at least one first device, a plurality of second devices for exchanging information with the first device, and the first device and the second device. A first interface in the form of a loop in which the information circulates in a single direction; switch means for selectively connecting and disconnecting each of the second devices to and from the first interface; The first of the two devices
Triggered by the error detection at the interface of
An information processing system comprising control logic for disconnecting the second device located upstream of the second device in the interface from the first interface by the switch means. 10. A connection between at least one first device, a plurality of second devices for exchanging information with the first device, and the first device and the second device. A first interface in the form of a loop in which the information circulates in a single direction; switch means for selectively connecting and disconnecting each of the second devices to and from the first interface; When the error occurrence state regarding the first interface stored in the second device deviates from a predetermined allowable range, the second interface located upstream of the second device in the first interface is triggered. And a control logic for disconnecting the device from the first interface by the switch means. 11. The information processing system according to claim 8, 9 or 10, wherein said switch means and said first and second devices are connected via a second interface different from said first interface. An information processing system, wherein the information processing system is connected to at least one of the first and second devices, and the first and second devices are connected and disconnected by the switch means via the second interface. 12. The information processing system according to claim 8, 9, 10, or 11, wherein the first device and the second device are connected via a second interface different from the first interface. An information processing system, wherein the first device collects error information on the first interface obtained by the second device via the second interface.