JP4241405B2 - Computer system, I / O retry method and program - Google Patents

Description

  The present invention relates to a computer system in which a central processing unit (CPU) is connected to an input / output device (device) such as a magnetic disk device or a magnetic tape (MT) device via a channel. Furthermore, the present invention relates to an I / O retry method performed in such a computer system and a program for realizing the control.

  In general, in a computer system in which a CPU is connected to a device via a channel (sometimes called an input / output channel), if a path failure occurs for some reason during an I / O operation, an I / O by an OS (operating system) is performed. O retry processing is performed. In this I / O retry processing, the same I / O instruction is executed once again, or if there is an alternative path, the path is changed to the alternative path and the normal I / O instruction is executed. Expecting. Japanese Patent Application Laid-Open No. 2004-133830 proposes that when a path failure occurs and an alternative path exists, the I / O instruction is re-executed using the alternative path.

  As an example of a form in which an I / O instruction is re-executed using an alternative path, for example, a connection form using Fiber Channel (FC) which is one of the high-speed interface standards as shown in FIG. 8 can be considered. . In this example, two cards, a dual port FC card 1 and a single port FC card 2, are used. One port of the dual port FC card 1 is directly connected to the magnetic disk device 4, and the other port is connected to the magnetic disk device 5 via the FC switch (FC-SW) 3. The port of the single port FC card 2 is connected to the magnetic disk device 5 via the FC-SW 3.

In the above connection mode, when the I / O command is abnormally terminated due to a failure of the Fiber Channel cable connected to the other port of the dual port FC card 1, the OS executes the same I / O command through the same route. However, normal termination cannot be expected. In this case, normal termination can be expected by re-executing the I / O command using the connection path (alternate path) to the magnetic disk device 5 via the single port FC card 2.
JP 63-104126 A

  As described above, in the conventional I / O retry method, normal termination can be expected by re-executing the I / O instruction using the alternative path. However, in order to use an alternative path, it is necessary to register information about the alternative path in advance in the OS. For example, in the connection form of FIG. 8, the magnetic disk connected to the other port of the dual port FC card 1 and the magnetic disk device connected to the single port FC card 2 are the same, and these paths It is necessary to register in the OS that can be replaced. This registration has to be performed by the user himself when connecting the magnetic disk using the FC card, which is very troublesome and inconvenient for the user. Note that Patent Document 1 does not describe any specific means for performing path switching.

  An object of the present invention is to provide a system, method, and program capable of solving the above-described problem and reliably performing I / O retry without pre-registration of information regarding an alternative path.

In order to achieve the above object, a computer system of the present invention includes a slot portion in which at least one Fiber Channel card is mounted, and a plurality of ports are provided in each of a plurality of ports provided in the Fiber Channel card mounted in the slot portion. I / O control that is connected to the plurality of input / output devices via the plurality of ports and executes an I / O command for the connected input / output devices. The I / O control device acquires a number for each of the plurality of ports that can identify each input / output device from the plurality of input / output devices to which the I / O control device is connected. When there are a plurality of ports in which all of the acquired plurality of numbers are the same, at least one of these ports is It is set as a port for retrying the I / O instruction. In this case, when the I / O control device detects that the I / O instruction is abnormally terminated at the other port, the I / O controller uses the retry port instead of the other port. The instruction may be re-executed.

  According to the above configuration, at least one of the ports having the same connection destination input / output device is set as a port for retrying an I / O instruction in another port. This retry port setting is a number that can identify each input / output device from all the input / output devices to which it is connected. Specifically, it is unique to the world-wide name (WWN). This is done by obtaining a number that is The WWN is a number that all input / output devices connected via the fiber channel have, and a plurality of input / output devices are not allowed to have the same number. Therefore, it is possible to recognize a path as an alternative path by acquiring and comparing WWNs of all connected input / output devices for each port. For example, if a dual-port Fiber Channel card is used and all the input devices connected to each port are the same, the numbers (WWN) acquired for these ports are the same, and one port is assigned to the I port of the other port. It can be set as a port for retrying the / O instruction. As described above, according to the present invention, the WWN is acquired from the input / output device of the connection destination and the route that becomes the alternative path is recognized, so that information related to the alternative path as in the past is registered in the OS. There is no need.

The I / O retry method of the present invention includes a slot portion in which at least one fiber channel card is mounted, and a plurality of input / output devices for each of a plurality of ports provided in the fiber channel card mounted in the slot portion. In the I / O retry method performed in the computer system to which the I / O is connected, for each of the plurality of ports, a number by which each input / output device can be identified from the plurality of input / output devices to which the port is connected When there are a plurality of ports in which all of the plurality of acquired numbers are the same as the first step to acquire, at least one of those ports is set as a port for retrying an I / O instruction in another port And a second step. In this case, the same operation as the above computer system is achieved.

The program of the present invention includes a slot portion in which at least one Fiber Channel card is mounted, and a plurality of input / output devices are connected to each of a plurality of ports provided in the Fiber Channel card mounted in the slot portion. the computer system, for the plurality of ports, each port from its a connection destination of the plurality of input and output devices, a first processing for acquiring an identifiable number of each of the input and output device, plurality of the acquired If all of the numbers are multiple ports is the same, and characterized in that to execute a second process for setting at least one of them port as port for retry I / O instructions in other ports To do. In this case, the same operation as the above computer system is achieved.

  According to the present invention as described above, there is an effect that the burden on the user is reduced because registration of information on an alternative path, which has been conventionally required, to the OS is unnecessary.

  Next, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a schematic configuration of a computer system according to an embodiment of the present invention. This computer system is connected to an external input / output device (magnetic disk device) via an existing fiber channel (FC) card 120, which is a high-speed interface, and its main part operates under program control. A central processing unit 100, a main storage unit 130, I / O control units 110 to 112 for controlling input / output of data to / from a magnetic disk unit according to a command from the central processing unit 100, and an FC card 120 are attached. And a slot.

  The slot is, for example, a PCI (Peripheral Component Interconnect) slot, and the attached FC card 120 is connected to the I / O control device via the PCI local bus. Normally, a plurality of slots are provided, but one slot may be provided. In the example of FIG. 1, two slots are prepared, and the FC card 120 is connected to the I / O control device 110 via the PCI local bus by attaching the FC card 120 to these slots. Yes.

  The FC card 120 is a dual port FC card having two ports “Port 0” and “Port 1” (hereinafter simply referred to as P0 and P1), and each of these ports P0 and P1 is an FC switch (FC-SW). It is connected to a magnetic disk device via 121. In the example of FIG. 1, the ports P0 and P1 are connected to both of the two magnetic disk devices 122 and 123 via the FC-SW 121, respectively. The ports P0 and P1 and the FC-SW 121 and the FC-SW 121 and the magnetic disk devices 122 and 123 are connected by fiber channel cables such as optical fiber cables and coaxial cables. The connection of the magnetic disk device using the FC card 120 can be arbitrarily performed by the user, and the connection form can be freely changed within the range of the specification of the computer system. For example, the number of magnetic disk devices connected to the FC-SW 121 may be one, or three or more.

  The central processing unit 100 instructs the I / O control devices 110 to 112 to control data input / output. FIG. 2 is a block diagram for explaining the configuration and operation of the main part of the central processing unit 100. Referring to FIG. 2, the central processing unit 100 has an I / O command generation unit 101 that generates an I / O command for interfacing with the I / O control device 110, and an I / O command that transmits the I / O command. An / O command transmission unit 102 and an I / O command execution result acquisition unit 103 that acquires the execution result of the I / O command.

  The I / O command generation unit 101 uses a predetermined format to specify the number of the I / O control device that executes the I / O command (a number that can identify the I / O control device), and the I / O control device The I / O command information 131 is created from data such as the subordinate channel number, the number of the device connected to the channel, the processing contents, the address of the main storage device 130 storing the data to be transferred and its size, and the like. Write to the storage device 130. The I / O command transmission unit 102 specifies the address of the I / O command information 131 written in the main storage device 130 by the I / O control device number included in the I / O command information 131. Notify the O control device. The I / O command execution result acquisition unit 103 refers to the predetermined I / O command execution result notification area 132 for each I / O control device in the main storage device 130, and executes the I / O command. Whether or not the process has been completed is monitored. If the process has been completed, the completion message is taken out and the execution result is determined. Based on this determination, the central processing unit 100 executes control processing such as operation stop at the FC card port corresponding to the failed path and operation start at the FC card port corresponding to the alternative path.

  Each of the I / O control devices 110 to 112 executes an I / O instruction from the central processing unit 100, and performs channel initialization / incorporation processing (including channel duplexing) and channel failure processing at the time of path failure. . Since the I / O control devices 110 to 112 basically have the same configuration, the configuration and operation of the I / O control device 110 will be specifically described here.

  FIG. 3 is a block diagram for explaining the configuration and operation of the main part of the I / O control device 110. Referring to FIG. 3, the I / O control device 110 receives and decodes the I / O command information 131 from the central processing unit 100, and the processing written in the I / O command information 131. The execution control unit 112 executes contents, the ROM 112a and the RAM 112b attached to the execution control unit 112, and a dedicated control program storage RAM 113.

  The execution control unit 112 can change the control depending on the type of the PCI channel card 141 connected to each slot “PCI Slot 0” and “PCI Slot 1” (hereinafter referred to as PS0 and PS1) of the PCI local bus 140. The operation is controlled by a control program stored in advance in the ROM 112a. With this control program, for example, it is determined whether the channel card indicated by the I / O command information 131 is connected to any of the slots PS0 and PS1 and is operable (including failure detection). The operation instructed by the / O instruction information 131 is executed, and the execution result is written in the I / O instruction execution result notification area in the main storage device 130. Furthermore, in response to the initialization / incorporation instruction of the channel card from the central processing unit 100, the information of the channel card is extracted by a method according to the specification of the PCI local bus 140, the card type is determined, and the card is supported. The dedicated control program is read from the dedicated control program storage RAM 113, loaded into the RAM 112b, and the dedicated control program is activated. By starting up this dedicated control program, it becomes possible to execute channel initialization / incorporation processing, which will be described later, and channel failure processing upon path failure. The control program stored in the ROM 112a and the dedicated control program stored in the dedicated control program storage RAM 113 may be provided to the computer system via a storage medium such as an optical disk or the Internet.

  Next, the operation of the computer system of this embodiment will be specifically described. Since the processing related to the execution of the I / O instruction is basically the same as that of the existing system, description of the operation is omitted, and here, initialization, which is a feature of the computer system of the present embodiment, is performed. / Operations of built-in processing (including channel duplication) and channel failure processing (I / O retry processing) will be specifically described.

(1) Initialization / embedding process:
FIG. 4 is a flowchart for explaining channel initialization / incorporation processing performed by the I / O control device. The operation will be described below with reference to FIGS.

  When a channel card is installed in either slot S0 or S1, central processing unit 100 performs I / O control on channel initialization / incorporation instructions in order to put the installed channel card into an operable state. Transmit to device 100. In the I / O control device 100, in accordance with the channel initialization / incorporation instruction, the execution control unit 112 performs channel initialization / incorporation processing according to the following procedure.

  First, “Vender ID” and “Device ID” are extracted from the configuration space of the installed channel card, and the type of the channel card is checked based on the extracted “Vender ID” and “Device ID” (step S10). If it is not an FC card, the process proceeds to a process for determining whether it is another supported channel card, and a necessary process is executed (step S12). In the case of the FC card, the fiber channel control program stored in advance as a dedicated control program is read from the dedicated control program storage RAM 113 to the RAM 112b and activated (step S13). Thereafter, the following processing is executed by the activated fiber channel control program.

  In step S14, initialization of the port P0 (acquisition, setting, etc. of the PCI base address space) is performed to make the FC card operable. Subsequently, in step S15, the link of the port P0 is established and login is performed, information on all devices (magnetic disk devices) connected to the port P0 is collected, and numbers for identifying the connected devices (each A WWN table for port P0, which is a number unique to the device and stores, for example, WWN (Wilde Word Name), is created in the RAM 112b.

  Next, in step S16, it is determined whether or not the FC card is a dual port FC card. If it is not a dual port FC card, in step S17, information related to the connected device of port P0 of the FC card is stored in the main storage device 130 (notification of the connected device information to the OS). Thereafter, the process is terminated. If it is a dual port FC card, in step S18, the port P1 is initially set (PCI base address space acquisition, setting, etc.) to make the FC card operable. After the initial setting of the port P1, in step S19, the link of the port P1 is established and login is performed, and information on the device (magnetic disk device) connected to the port P0 is collected and the WWN of the connected device is stored. The port P1 WWN table is created in the RAM 112b.

  Next, in step S20, it is confirmed whether or not the WWN stored in the port P0 WWN table matches the WWN stored in the port P1 WWN table. If the WWNs stored in both tables do not match, in step S21, the port P0 is logged out and linked down to stop the operation, and information related to the connected device of the port P0 is stored in the main storage device 130 (connection). Notification of device information to the OS). If the WWNs stored in both tables match, a retry possible flag for port P1 is set in step S22, and information relating to the connected device of port P0 is stored in main storage device 130 in step S23 (connected device information). To the OS). In setting the retry possible flag, for example, processing is performed such that a column relating to flag information is provided in the port P1 WWN table, and “1” is set as flag information in this column. In this case, “0” is set as flag information in the default state. After step S21 and step S23, the process ends.

  Through the above processing, the port P0 of the FC card 120 is connected to each of the magnetic disk devices 122 and 123 via the FC-SW 121, and an I / O command can be executed.

(2) Channel failure processing (I / O retry processing):
FIG. 5 is a flowchart for explaining channel failure processing (I / O retry processing) by the execution control unit of the I / O control device. The operation will be described below with reference to FIGS. In the operation described below, a dual port FC card is installed as the FC card 120 in the slot PS0, and the ports P0 and P1 are connected to the magnetic disk devices 122 and 123 via the FC-SW 121, respectively. This is an operation after the above-described channel initialization / incorporation processing of FIG. 4 is executed.

  When a failure is detected during an operation such as DMA transfer by an I / O instruction (during data transfer), first, failure information is read from the FC card and the type of failure is examined (step S30). Then, an end message to be written in the I / O instruction execution result notification area is created according to the type of failure. Abnormal termination and failure information are set in the termination message.

  Next, a retry possible flag is confirmed (step S32). In this flag confirmation process, it is confirmed in steps S20 and S22 in the channel initialization / incorporation process of FIG. 4 whether or not the retryable flag by the port P1 of the dual port FC card is set. If “1” is set in the flag information column of the port P1 WWN table, it is determined that the retry possible flag is set. If “0” is set, the retry possible flag is set. Judge that it is not.

  If it is determined that the retry enable flag is not set, the end message created in step S31 is written in the I / O instruction execution result notification area (step S33), and then the process ends. If it is determined that the retry possible flag is set, the same I / O instruction is executed using the port P1 (step S34).

  After the execution of the I / O instruction, it is confirmed whether or not it has been normally completed (step S35). In the case of abnormal termination, the process ends after executing the process of step S33. In the case of normal termination, the content of the termination message created in step S31 is changed to normal termination and failure information indicating that a failure has been detected in the slot PC0 is set, and this is set in the I / O instruction execution result notification area. (Step S36). Then, after logging out of the slot PS0 and linking it down to stop the operation, the retry possible flag is reset by setting the processing to be executed only at the port P1 (step S37). By resetting the retry enable flag, “0” is set in the flag information column of the port P1 WWN table. After the flag is reset, the process ends.

  According to the above channel failure processing, the port P1 is used when the failure is caused by the port portion of the port P0 of the FC card 120, the fiber channel cable connected to the port P0, or the port portion of the FC-SW 121. By doing so, normal termination of the I / O instruction can be expected. In the fault processing in the existing system, after writing an end message in the I / O command execution result notification area, it is left to the OS to decide whether to retry the I / O command or close the channel. Become.

  In the above description, for example, as shown in FIG. 6, it is assumed that the dual port FC card 1 in which each port is connected to the magnetic disk devices 4 and 5 via the FC-SW 3 is mounted in the slot. However, the present invention is not limited to this connection form. The connection form with the input / output device (magnetic disk device) using the FC card can be arbitrarily set by the user. For example, as shown in FIG. 7, of the two ports P0 and P1 of the FC card 120, the port P0 is connected to the magnetic disk devices 122 and 123 via the FC-SW 121, and the port P1 is connected via the FC-SW 124. It is also possible to connect to the magnetic disk devices 122 and 123. When such a connection form is adopted, the cause of the failure is the port portion of the port P0 of the FC card 120, the port portions of the FC-SWs 121 and 124, and the ports of the magnetic disk devices 122 and 123 due to the failure processing described above. 1 and the fiber channel cable connecting these ports, the normal termination of the I / O retry process can be expected by using the port P1.

  In the initialization / embedding process shown in FIG. 4, the port P1 of the dual port FC card is set as an alternative path for I / O retry, but conversely, the port P0 is set as an alternative path. May be. If a failure is detected at the port P0, the initialization / incorporation process can be continuously executed only at the port P1. Further, in step S21, the operation is stopped by linking down the port P1, but the OS may be notified of information related to the connected devices of the ports P0 and P1 without linking down. Thereby, the port can be used effectively.

  In the failure processing shown in FIG. 5, the port P0 where the failure is detected is closed by the first failure detection, and the processing is continued only by the port P1, but a fixed number of retries are performed at the port P0. However, when the normal termination is not performed, the path switching to the port P1 may be performed.

(Other embodiments)
The type and number of FC cards to be used are not particularly limited as long as the channel can be duplicated. The present invention provides a first connection form using a plurality of dual port FC cards, a second connection form using a combination of a dual port FC card and a single mode FC card (for example, the connection form shown in FIG. 8), The present invention can be applied to various connection forms such as a third connection form using a plurality of single mode FC cards.

  When the configuration can be applied to each connection mode, first, the execution control unit of the I / O control device checks which connection mode is selected from the type and number of FC cards installed in the slot. Is done. When the installed FC card is a dual port FC card, the processing shown in FIGS. 4 and 5 is executed as it is. In the case of any one of the first to third connection forms, the initialization / incorporation process and the failure process are performed as follows.

(1) First connection form:
As an example, the operation when the first dual port FC card is installed in the slot PS0 and the second dual port FC card is installed in the slot PS1 in the configuration of FIG. 1 will be described. In this case, the processes from steps S10 to S19 of the initialization / incorporation process shown in FIG. 4 are executed for each of the slots PS0 and PS1. Then, it is checked whether each port WWN table has the same WWN stored therein. For ports in which the same WWN is stored in the port WWN table, a retry enable flag is set for at least one of these ports, and information on the connected devices of the remaining ports is notified to the OS. For ports for which the same WWN is not stored in the port WWN table, the OS is notified of information related to the connected device of each port.

  According to the above initialization / incorporation process, the retry possible flag is set in at least one of the ports connected to the same disk device among the ports of the first and second dual port FC cards. For example, when one port of the first dual port FC card and one port of the second dual port FC card are connected to the same disk device, a retry enable flag is set to one of these ports. When a failure occurs in the other port, the path is switched to the port for which the retry possible flag is set.

(2) Second connection form:
For each of the dual port FC card and the single port FC card, the processes of steps S10 to S19 of the initialization / incorporation process shown in FIG. 4 are executed. However, if “No” is determined in step S16, the process of step S17 is not performed and the process directly proceeds to step S20. By the processing so far, the port WWN table for each port of the dual port FC card and the single port FC card is created.

  Next, in step S20, it is checked whether or not the same WWN is stored for each port WWN table. For ports in which the same WWN is stored in the port WWN table, a retry enable flag is set for at least one of these ports, and information on the connected devices of the remaining ports is notified to the OS. For ports for which the same WWN is not stored in the port WWN table, the OS is notified of information related to the connected device of each port.

  According to the above initialization / embedding process, the retry possible flag is set in at least one of the ports connected to the same disk device among the ports of the dual port FC card and the single port FC card. For example, when the connection form shown in FIG. 8 is adopted in the configuration of FIG. 1, one of the port 1 of the dual port FC card 1 installed in the slot PS0 and the port of the single port FC card 2 installed in the slot PS1. In addition, a retry possible flag is set. When a failure occurs in the other port, the path is switched to the port for which the retry possible flag is set.

(3) Third connection form:
For each single port FC card, the processes of steps S10 to S19 of the initialization / incorporation process shown in FIG. 4 are executed. However, if “No” is determined in step S16, the process of step S17 is not performed and the process directly proceeds to step S20. By the processing so far, the port WWN table for each port of each single port FC card is created.

  Next, in step S20, it is checked whether or not the same WWN is stored for each port WWN table. For ports in which the same WWN is stored in the port WWN table, a retry enable flag is set for at least one of these ports, and information on the connected devices of the remaining ports is notified to the OS. For ports for which the same WWN is not stored in the port WWN table, the OS is notified of information related to the connected device of each port.

  According to the computer system of each embodiment described above, it is possible to automatically recognize an alternative route that can be retried by checking the matching of connected devices by WWN in the control software without affecting the OS (channel). Duplication).

  In Fiber Channel, optical cables are often used for connection between devices, but failure due to attenuation of the optical level of the optical cable and the optical module that transmits and receives optical signals can be avoided by switching the path. The success of retries can be expected.

  In the computer system of each embodiment, depending on the way of connection, a part of WWNs of a plurality of magnetic disk devices stored in the port WWN table may coincide. For example, in the configuration shown in FIG. 7, when another magnetic disk device is connected to the FC-SW 124 in addition to the magnetic disk devices 122 and 123, the port WWN table for port P1 and the WWN table for port 2 are , A part of the stored WWN matches. In such a case, since the channel cannot be duplicated, the retry enable flag is not set. That is, the retry enable flag is set only when all the WWNs stored in the port WWN table match.

  The present invention can be applied to all computer systems to which a fiber channel card can be connected. For example, the present invention can be applied to a computer system such as a disk array.

It is a block diagram which shows schematic structure of the computer system which is one Embodiment of this invention. It is a block diagram for demonstrating the structure and operation | movement of the principal part of the central processing unit of the computer system shown in FIG. It is a block diagram for demonstrating the structure and operation | movement of the principal part of the I / O control apparatus of the computer system shown in FIG. 4 is a flowchart for explaining channel initialization / incorporation processing performed by an execution control unit of the I / O control device shown in FIG. 3. 4 is a flowchart for explaining channel failure processing performed by an execution control unit of the I / O control device shown in FIG. 3. It is a schematic diagram which shows an example of the connection form applicable to the computer system of this invention. It is a schematic diagram which shows the other example of the connection form applicable to the computer system of this invention. It is a schematic diagram which shows an example of the connection form using a fiber channel card.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Central processing unit 101 I / O command generation part 102 I / O command transmission part 110-112 I / O control apparatus 120 Fiber channel card 121 Fiber channel card switch 122, 123 Magnetic disk unit 130 Main storage unit

Claims (27)

A computer system comprising a slot portion in which at least one Fiber Channel card is mounted, and a plurality of input / output devices connected to each of a plurality of ports provided in the Fiber Channel card mounted in the slot portion,
An I / O control device connected to the plurality of input / output devices via the plurality of ports and executing an I / O instruction for the connected input / output devices;
More the I / O controller, for said plurality of ports, each port from its a connection destination of the plurality of input and output devices, acquires the identifiable number of each of the input and output device, which is the acquired When there are a plurality of ports having the same number , the computer system is characterized in that at least one of the ports is set as a port for retrying the I / O instruction in another port.   When the I / O control unit detects that the I / O instruction is abnormally terminated at the other port, the I / O instruction is re-executed using the retry port instead of the other port. The computer system according to claim 1, wherein the computer system is executed. The I / O control device has storage means, and for each of the plurality of ports, creates a table storing the numbers of the plurality of input / output devices to be connected for each port, and stores the table in the storage means 2. The computer according to claim 1, wherein a retry possible flag indicating that the port can be set as the retry port is set for at least one of the ports in which all of the plurality of numbers in the stored table match. system.   When the I / O control device detects that the I / O instruction is abnormally terminated at the other port, the I / O control device uses the port with the retry enable flag set instead of the other port. The computer system of claim 3, wherein the O instruction is re-executed.   The computer system of claim 1, wherein the fiber channel card is a dual port fiber channel card.   The computer system according to claim 5, wherein the I / O control device sets one of the ports of the dual port Fiber Channel card as the retry port.   The computer system of claim 1, wherein the fiber channel card is a plurality of single mode fiber channel cards.   8. The computer system according to claim 7, wherein the I / O control device sets a port of at least one of the plurality of single mode fiber channel cards as the retry port.   The computer system of claim 1, wherein the fiber channel card comprises a combination of a dual port fiber channel card and a single mode fiber channel card.   8. The computer system according to claim 7, wherein the I / O control device sets a port of one of the dual port Fiber Channel card and the single mode Fiber Channel card as the retry port.   The computer system according to claim 1, wherein the number that can specify the plurality of input / output devices is a world wide name. I is performed in a computer system including a slot portion in which at least one Fiber Channel card is mounted, and a plurality of input / output devices are connected to each of a plurality of ports provided in the Fiber Channel card mounted in the slot portion. In the / O retry method,
For each of the plurality of ports, a first step of obtaining a number that can identify each input / output device from the plurality of input / output devices to which the port is connected;
If the port all of the acquired plurality of numbers are the same there are multiple, comprise a second step of setting at least one of their ports as ports for retry I / O instructions in other ports An I / O retry method characterized by the above.   The method further includes the step of re-executing the I / O instruction using the retry port instead of the other port when detecting that the I / O instruction is abnormally terminated at the other port. Item 13. The I / O retry method according to Item 12. The second step includes
For each of the plurality of ports, creating a table storing the numbers of the plurality of input / output devices to be connected for each port;
And a step of setting a retry possible flag indicating that the port can be set as the retry port for at least one of the ports in which all of the plurality of numbers in the table match. I / O retry method.   When it is detected that the I / O instruction is abnormally terminated in the other port, the step of re-executing the I / O instruction using a port in which the retry possible flag is set instead of the other port The I / O retry method according to claim 14, further comprising: The Fiber Channel card is a dual port Fiber Channel card;
13. The I / O retry method according to claim 12, wherein the second step includes a step of setting one of the ports of the dual port fiber channel card as the retry port. The Fiber Channel card is a plurality of single mode Fiber Channel cards;
The I / O retry method according to claim 12, wherein the second step includes a step of setting a port of at least one of the plurality of single mode Fiber Channel cards as the port for retry. The Fiber Channel card is a combination of a dual port Fiber Channel card and a single mode Fiber Channel card,
The I / O retry method according to claim 12, wherein the second step includes a step of setting a port of one of the dual port Fiber Channel card and the single mode Fiber Channel card as the retry port.   19. The I / O retry method according to claim 12, wherein the number that can specify the plurality of input / output devices is a world wide name. A computer system including a slot portion in which at least one Fiber Channel card is mounted, and a plurality of input / output devices connected to each of a plurality of ports provided in the Fiber Channel card mounted in the slot portion;
For each of the plurality of ports, for each port, from the plurality of input / output devices that are connection destinations, a first process for acquiring a number that can identify each input / output device;
When there are a plurality of ports where all of the plurality of acquired numbers are the same, a second process for setting at least one of the ports as a port for retrying an I / O instruction in another port is executed. program.   When it is detected that the I / O instruction is abnormally terminated at the other port, a process for re-executing the I / O instruction using the retry port instead of the other port is performed on the computer. The program according to claim 20, further causing execution. The second process includes
For the plurality of ports, for each port, a process of creating a table storing the numbers of the plurality of input / output devices to be connected;
21. A process of setting a retry possible flag indicating that the port can be set as the retry port for at least one of the ports in which all of the plurality of numbers in the table match. Program.   When it is detected that the I / O instruction is abnormally terminated in the other port, a process of re-executing the I / O instruction using the port in which the retry possible flag is set instead of the other port The program according to claim 22, further causing the computer to execute. The Fiber Channel card is a dual port Fiber Channel card;
21. The program according to claim 20, wherein the second process includes a process of setting one of the ports of the dual port fiber channel card as the retry port. The Fiber Channel card is a plurality of single mode Fiber Channel cards;
The program according to claim 20, wherein the second process includes a process of setting a port of at least one of the plurality of single mode fiber channel cards as the retry port. The Fiber Channel card is a combination of a dual port Fiber Channel card and a single mode Fiber Channel card,
The program according to claim 20, wherein the second process includes a process of setting a port of one of the dual port fiber channel card and the single mode fiber channel card as the retry port. 27. The program according to claim 20, wherein the number that can specify the plurality of input / output devices is a world wide name.

Images