JP5556226B2 - Interface test apparatus and interface test method - Google Patents

Description

  The present invention relates to an interface test apparatus and an interface test method.

  The interface test apparatus is inserted between the transmission apparatus and the reception apparatus, generates a pseudo interface failure, and tests the transmission device or the reception device when the interface failure occurs. Interfaces, in other words, communication paths include, for example, SATA (Serial Advanced Technology Attachment), SAS (Serial Attached SCSI), PCI (Peripheral Component Interconnect) -express, Fiber-Channel, and LAN (Local Area Network). . An interface failure is a failure that has occurred in an interface, such as a temporary disconnection of a communication path. For example, the interface test apparatus temporarily disconnects the communication path by inserting an electrical switch into a cable or connector that connects the transmission apparatus and the reception apparatus. The test in the case where an interface failure occurs is, for example, a test as to whether communication retry and retry out are normally executed in the transmission device that is the device under test.

  In a computer equipped with a simulated fault generating circuit for generating a simulated fault, a mechanism for measuring a predetermined time is provided in the simulated fault generating circuit, and when the simulated fault generating circuit is activated, a predetermined time is determined based on the output of the timing mechanism. A pseudo-fault generating circuit that generates a pseudo-fault signal only has been proposed.

JP 58-39351 A

  For example, the interface test apparatus tests whether the retry of the communication in the transmission apparatus is normally executed, for example, by disconnecting the communication between the transmission apparatus and the reception apparatus. The retry is executed when a timeout occurs while monitoring communication. Therefore, it is necessary to disconnect communication so that a timeout occurs due to an interface failure and a retry is executed.

  Further, for example, the interface test apparatus repeatedly disconnects communication between the transmission apparatus and the reception apparatus, thereby testing whether, for example, a retry-out of the communication in the transmission apparatus occurs normally. The retry-out occurs when the transmission device cannot return to normal operation even if the retry is executed a predetermined number of times. Accordingly, it is necessary to repeatedly disconnect communication so that a retry is executed by the number of retries due to an interface failure and a retry out occurs.

  An object of the present invention is to provide an interface test apparatus that can set a communication path disconnection time to a timing at which a retry-out occurs.

The disclosed interface test apparatus includes a switch unit, a parameter setting unit, and a switch control unit. The switch unit is inserted between the first transmission / reception device and the second transmission / reception device, and enables or disables communication between the first transmission / reception device and the second transmission / reception device. One of the states. The parameter setting unit is a predetermined initial value for the disconnection time, which is a time during which the state in which communication is impossible is continued, and the first transmission / reception device is in a normal state due to the state in which communication is not possible. The initial value, which is the time required for one retry execution, and a predetermined increase amount for the disconnection time, which is an operation for performing communication again when the communication is not performed, and the time required for executing the retry The increase amount which is is stored. Each time the switch control unit switches between a communication enabling state and a communication disabling state, the switch control unit sets the disconnection time to the increase amount based on the initial value and the increase amount stored in the parameter setting unit. Increase control is performed on the switch unit.

  According to the disclosed interface test apparatus, a retry process and a retry-out process that have a high probability of occurrence of a bug without investigating the number of retries specific to the vendor of the transmission / reception apparatus without the intervention of an evaluator of the interface test. The cuts associated with can be reproduced.

It is a figure which shows an example of a structure of an interface test apparatus. It is explanatory drawing of an interface test. It is explanatory drawing of an interface test. It is explanatory drawing of an interface test. It is explanatory drawing of an interface test. It is explanatory drawing of an interface test. It is explanatory drawing of an interface test. It is an interface test processing flow.

  A defect related to the interface in the device under test is empirically detected in many cases when the communication retry process fails and when the recovery process fails after retry-out. For this reason, in the interface test of the device under test, a pseudo interface failure is generated to evaluate the communication retry process and the recovery process after retry-out.

  The retry process and the retry-out process for communication will be described using an example in which a disconnection failure has occurred in the interface, for example.

  For example, when the transmission device transmits a write command to the reception device, the reception device transmits a command response to the transmission device. The transmitting device that has received the command response transmits a plurality of write data to the receiving device. The receiving device that has received the plurality of write data transmits a status to the transmitting device. The transmitting device that has received the status transmits a status response to the receiving device. As described above, the communication between the transmission device and the reception device ends normally.

  However, for example, the transmission apparatus may not be able to receive a command response due to an interface failure, for example, a temporary signal disconnection.

  In this case, the transmitting device that has transmitted the write command waits for a command response from the receiving device until a predetermined timeout time. When the specified timeout time elapses, the transmission device transmits a write command again. In other words, when the transmission device detects a timeout, it performs a retry for the write command. The retry is to execute failed communication, for example, transmission of a write command again.

  If the transmission device does not receive a command response from the reception device even if the above operation is repeated for the number of retries, the transmission device performs a retry-out process on the assumption that the retry has failed. The number of retries is the number of times a retry for the communication is repeated for one communication failure.

  Here, the specified timeout time is defined in the interface standard, in other words, standardized. Therefore, the evaluator of the interface test can know in advance by investigating the interface standard.

  However, since the number of retries to be retried is not standardized, it cannot be known in advance in almost all cases. In other words, the number of retries is a value unique to the vendor of each transmitting apparatus and is not disclosed. The same applies to the receiving apparatus. Furthermore, empirically, the number of retries may be different for each transmitting / receiving device due to various circumstances.

  Accordingly, the evaluator of the interface test executes the interface test after setting the number of retries by estimating the number of retries that cause a retry-out for each device under test. When the retry-out fails, the interface test evaluator needs to manually set the retry count in the interface test apparatus so that the retry count is changed little by little. For this reason, estimating and setting the number of repeated retries for each device under test places a great burden on the evaluator of the interface test.

  If the number of retries is known, the time for retry-out can be calculated by calculating (timeout specified time) × (number of retries). However, as described above, since the number of retries is unknown, the time for retry-out cannot be calculated.

  The disclosed interface test apparatus and interface test method provide an interface test apparatus and an interface test method that can set the disconnection time of a communication path to a timing at which an interface failure occurs.

  FIG. 1 is a diagram illustrating an example of the configuration of an interface test apparatus.

  The interface test apparatus 1 is inserted, for example, between a server 2 that is a first transmission / reception device and a disk device 3 that is a second transmission / reception device. The first transmission / reception device is not limited to the server 2. The first transmission / reception device may be a device that transmits / receives a signal to / from another transmission / reception device, and may be a host computer, for example. The second transmission / reception device is not limited to the disk device 3. The second transmission / reception device may be a device that transmits / receives a signal to / from another transmission / reception device, and may be a computer, for example.

  In the example of FIG. 1, the server 2 is, for example, a transmission device that transmits a write command to the disk device 3 and a reception device that receives a command response from the disk device 3. The disk device 3 is a receiving device that receives a write command and is a transmitting device that transmits a command response to the server 2.

  The server 2 and the disk device 3 are connected by communication paths 51 and 52. The communication path 51 is a path for transmitting a signal from the server 2 to the disk device 3. The communication path 52 is a path having a communication direction different from that of the communication path 51 and is a path for transmitting a signal from the disk device 3 to the server 2. The communication paths 51 and 52 are transmission paths that transmit high-speed serial signals, for example.

  The communication paths 51 and 52 include an external part of the interface test apparatus 1 and an internal part of the interface test apparatus 1. The communication paths 51 and 52 outside the interface test apparatus 1 are, for example, connection cables or connection connectors. The communication paths 51 and 52 inside the interface test apparatus 1 are internal wirings of the interface test apparatus 1, for example.

  The interface test apparatus 1 includes a switch unit 11, a switch control unit 12, and a parameter setting unit 13. The switch unit 11 includes electrical switches 111 and 112. The switches 111 and 112 are connected to the internal wiring of the interface test apparatus 1 that constitutes the communication paths 51 and 52 inside the interface test apparatus 1.

  The switch unit 11 is inserted between the server 2 and the disk device 3 under the control of the switch control unit 12. In other words, as shown in FIG. 1, the switch 111 is inserted into the communication path 51 and the switch 112 is inserted into the communication path 52. The switch 111 and the switch 112 are bus switches configured by MOSFETs, for example. ON / OFF of the switch 111 and the switch 112 is controlled by the switch control unit 12.

  The switch unit 11 is either in a state that enables communication between the server 2 and the disk device 3 or a state that disables communication. The state where communication is possible is a state where there is no interface failure, and the server 2 and the disk device 3 operate normally. The state where communication is impossible is a state where there is an interface failure, and the server 2 and the disk device 3 do not operate normally.

  Therefore, when the switch unit 11 is in a state in which communication is impossible, the server 2 may execute a retry process and a retry-out process, for example. The time during which the state in which the switch unit 11 cannot communicate is continued is the disconnection time Toff. On the other hand, the time during which the switch unit 11 is allowed to communicate is the communication time Ton.

  On the other hand, the parameter setting unit 13 includes an initial value storage unit 131, an increase amount storage unit 132, a maximum value storage unit 133, a communication time storage unit 134, a test number storage unit 135, and a test interval storage unit 136. Including. The initial value storage unit 131 stores a predetermined initial value t0 for the cutting time Toff. The increase amount storage unit 132 stores a predetermined increase amount Δt for the cutting time Toff. The maximum value storage unit 133 stores a predetermined maximum value N · Δt for the cutting time Toff. The communication time storage unit 134 stores a predetermined communication time Ton. The test count storage unit 135 stores the test count T. The test interval storage unit 136 stores the test interval T0.

  The initial value t0 is the time required to execute one retry of communication in which the server 2 as the device under test has failed, in other words, the retry time. For example, the initial value t0 is the time required for execution of one retry of the write command transmission when the server 2 cannot transmit normally due to the state in which communication is impossible. The retry is an operation for executing the failed communication again. The relationship between the retry time and the retry timeout prescribed time will be described later with reference to FIG.

  The increase amount Δt is a time required to execute one retry of the communication in which the server 2 as the device under test has failed, in other words, a retry time. Therefore, in the interface test apparatus 1 of FIG. 1, the increase amount Δt is equal to the initial value t0.

  The number of tests T is the number of times the test is repeated when the cutting time Toff is one test until the cutting time Toff reaches the maximum value N · Δt from the initial value t0. The number of tests T is predetermined. The test interval T0 is the time from the end of the previous one test to the start of the next test. The test interval T0 is predetermined.

  For example, the initial value t0 is 1 millisecond (ms), the increment Δt is 1 millisecond, and the maximum value N · Δt is 256 milliseconds. The number of tests T is, for example, 100 to 1000 times. The test interval T0 is, for example, 5 milliseconds to 10 milliseconds. These numerical values are merely examples, and are set for each device under test.

  The initial value t0, the increment Δt, the maximum value N · Δt, the communication time Ton, the number of tests T, and the test interval T0 are parameters of the interface test. Prior to the start of the interface test, the parameters of the interface test are stored in the parameter setting unit 13 from the control computer 4 according to the input of the evaluator of the interface test. As a result, the evaluator of the interface test can execute the interface test only by first setting the parameter of the interface test in the parameter setting unit 13.

  Further, by making the increase amount Δt equal to the initial value t0, the cutting time Toff can be gradually increased with the normalized retry time as an increment unit. In other words, by dividing the cutting time Toff at the time when the retry-out occurs by the retry time, the number of retries executed until the time when the retry-out occurs, in other words, the number of retries can be known. Thus, the interface test evaluator can execute the interface test without manual intervention after setting the interface test parameters.

  Based on the parameters stored in the parameter setting unit 13, the switch control unit 12 sets the disconnection time Toff each time the switch unit 11 switches between the communication enabling state and the communication disabling state. The switch unit 11 is controlled so as to increase by a predetermined increase amount. In other words, the cutting time Toff is increased each time the switch unit 11 is switched on and off. Further, the switch control unit 12 switches the switch unit 11 so as to alternately switch between a state enabling communication and a state disabling communication based on the parameters stored in the parameter setting unit 13. To control. In other words, the switch control unit 12 controls on and off of the switch unit 11.

  Thereby, in the interface test apparatus 1 of FIG. 1, the communication time Ton and the disconnection time Toff are alternately switched. At this time, the cutting time Toff is gradually increased with the increase amount Δt as a unit of increase. On the other hand, the communication time Ton is fixed, for example.

  For this purpose, the switch control unit 12 includes a timer 121 and a time increase control unit 122. The timer 121 is a local timer in the interface test apparatus 1. The timer is referred to by the switch control unit 12.

  The time increase control unit 122 determines the cutting time Toff based on the initial value t0 stored in the parameter setting unit 13 and the increase amount Δt stored in the parameter setting unit 13. When the cutting time Toff reaches a predetermined maximum value N · Δt, the time increase control unit 122 switches from the state in which the first communication is possible to the state in which the communication is not possible thereafter. The cutting time Toff is set to the initial value t0.

  As a result, the switch control unit 12 increases the disconnection time Toff by the increase amount Δt each time switching between the state enabling communication and the state disabling communication, and then disabling communication. The switch unit 11 is controlled so as to be switched to. Thereafter, the switch control unit 12 controls the switch unit 11 to switch to a state in which communication is possible when the increased disconnection time Toff has elapsed.

  In one test, the switch control unit 12 switches alternately between a state in which communication is impossible and a state in which communication is possible according to the communication time according to the disconnection time Toff and the communication time Ton. To control. The switch control unit 12 ends one test when the cutting time Toff reaches a predetermined maximum value N · Δt. After the end of one test, the switch control unit 12 performs the next test with the test interval T0 interposed therebetween. At this time, the cutting time Toff is set to an initial value t0. The switch control unit 12 executes tests for the number of tests, with one test as a unit of repetition.

  Here, in the interface test apparatus 1 of FIG. 1, a state in which communication between the server 2 and the disk device 3 is impossible is realized by turning off the switch 112. By turning off the switch 112, the communication path 52 through which the command response passes is disconnected. This is shown in FIG. 2 with a cross mark attached to the communication path 52. The signal cut by the switch 112 is not limited to a command response of a write command or a command response to various commands.

  In other words, the switch unit 11 disables communication between the server 2 and the disk device 3 by disconnecting the response from the disk device 3 that is the receiving device to the server 2 that is the transmitting device. Is done. The signal cut by the switch 112 is not limited to a response, and may be a signal transmitted from the disk device 3 to the server 2.

  For this purpose, the switch control information is actually stored in the parameter setting unit 13 by the control computer 4 together with the parameters. The switch control information indicates whether the parameter of the interface test is a control parameter for the switch 111 or the switch 112. In the interface test apparatus 1 of FIG. 1, switch control information indicating that it is a parameter for controlling the switch 112 is stored in the parameter setting unit 13.

  Note that the state in which communication between the server 2 and the disk device 3 is impossible may be realized by turning off the switch 111. In other words, the switch unit 11 may be in a state in which communication between the server 2 and the disk device 3 is disabled by disconnecting communication from the server 2 that is a transmission device. The signal cut by the switch 111 is not limited to a write command or various commands.

  Further, the state in which communication between the server 2 and the disk device 3 is impossible may be realized by means other than physically disconnecting the switch 111 or 112. For example, the communication path 51 or 52 may be logically disconnected. Alternatively, normal signal transmission / reception may not be performed between the server 2 and the disk device 3 through the communication path 51 or 52.

  As described above, the state in which communication is impossible, in other words, the disconnection of communication may be a state that causes a state in which the device under test cannot normally terminate communication, and the cause thereof is questioned. Absent. The device under test is not aware of the cause of the state that makes communication impossible. As a result of the device under test being unable to end communication normally, the device under test executes the disconnected communication again. In other words, a retry for the communication is executed.

  FIG. 2 is an explanatory diagram of the interface test, and mainly shows an example of the configuration of the server 2 which is a device under test.

  The server 2 includes a transmission / reception unit 21, an error occurrence detection unit 22, an error processing unit 23, and an error output storage unit 24. The error processing unit 23 includes a retry processing unit 231, a normal operation return unit 232, a retry-out processing unit 233, and an error report processing unit 234. The disk device 3 includes a transmission / reception unit 31.

  The transmission / reception unit 21 of the server 2 transmits a predetermined signal to the transmission / reception unit 31 of the disk device 3 via the communication path 51. As described above, this signal is, for example, a write command. The transmission / reception unit 21 receives a predetermined signal from the transmission / reception unit 31 via the communication path 52. As described above, this signal is, for example, a command response.

  The error occurrence detection unit 22 monitors the transmission / reception unit 21 and detects an error in signal transmission / reception. For example, when the transmission / reception unit 21 transmits a signal, for example, a write command, the error occurrence detection unit 22 waits until a timeout specified time until the transmission / reception unit 21 receives a response signal, for example, a command response from the transmission / reception unit 31. When the specified timeout time elapses, the error occurrence detection unit 22 notifies the error processing unit 23 of the detection of the error occurrence. As described above, the timeout specified time is standardized.

  When the detection of the error occurrence is notified, in the error processing unit 23, the retry processing unit 231 instructs the transmission / reception unit 21 to retry the write command to be monitored. The error occurrence detection unit 22 monitors the retry, in other words, the re-transmission of the write command in the same manner as described above.

  When the transmission / reception unit 21 receives a command response within the specified timeout time by this retry, for example, the error occurrence detection unit 22 notifies the error processing unit 23 of reception of the command response. When the reception of the command response is notified, in the error processing unit 23, the retry processing unit 231 instructs the normal operation return unit 232 to execute a process for returning to normal operation, in other words, a recovery process. In response to this, the normal operation return unit 232 executes a recovery process.

  On the other hand, if the transmission / reception unit 21 does not receive a command response within the specified timeout time even after the retry, the retry processing unit 231 repeatedly performs the retry for the number of times of retry. When the transmission / reception unit 21 receives a command response within the time-out specified time while repeating the retry, as described above, the normal operation return unit 232 executes the recovery process. As described above, the number of retries cannot be known in advance.

  If the transmission / reception unit 21 does not receive a command response within the time-out specified time even if the retry is repeated for the number of retries, the retry processing unit 231 determines that the retry is out, and sends a retry-out process to the retry-out processing unit 233. Is instructed to execute. In response to this, the retry-out processing unit 233 executes a retry-out process. Accordingly, for example, an output indicating that the retry is out is displayed on the display output device of the server 2. Further, the error report processing unit 234 stores error information in the error output storage unit 24 in accordance with an instruction from the retry-out processing unit 233. The error information includes, for example, the occurrence of a retry-out, the number of retries executed until the retry-out occurs, a log at the time of the retry-out, and the like. The evaluator of the interface test can obtain the number of retries and a log when a retry-out occurs by reading the error information in the error output storage unit 24.

  The interface test for the server 2 in FIG. 2 by the interface test apparatus 1 in FIG. 1 will be described below with reference to FIGS.

  FIG. 3 is an explanatory diagram of the interface test, and particularly shows the switching of the communication state between the server 2 and the disk device 3 by the interface test apparatus 1. In the case of FIG. 3, it is assumed that the switch 112 configuring the communication path 52 is turned off in the switch unit 11.

  FIG. 3A schematically shows an input from the disk device 3 to the switch unit 11. In other words, FIG. 3A shows a state in which the disk device 3 normally outputs a signal to the server 2. Therefore, the signal output from the disk device 3 normally reaches the interface test device 1.

  However, as described above, the switch unit 11 alternately switches between a state where communication between the server 2 and the disk device 3 is possible and a state where communication is impossible. In response to this, the output of the switch unit 11 changes as shown in FIG.

  FIG. 3B schematically shows an output from the switch unit 11 to the server 2. In other words, FIG. 3B shows a state in which the server 2 receives a signal from the disk device 3 normally and does not normally receive the signal.

  For example, when the switch unit 11 is in a state where communication is possible, in other words, when the switch unit 11 is ON, the output from the switch unit 11 to the server 2 is sent to the switch unit 11 from the disk device 3 shown in FIG. Is equal to the input. As a result, the signal output from the disk device 3 reaches the server 2 normally.

Thereafter, the switch unit 11 is switched by the switch control unit 12 to a state in which the switch unit 11 cannot communicate, in other words, OFF. In this case, the output from the switch unit 11 to the server 2 is in spite of the presence of the input to the switch unit 11 from the disk device 3.
not exist. As a result, the signal output from the disk device 3 does not reach the server 2 normally. In other words, the communication between the server 2 and the disk device 3 is disconnected. The period during which communication is disconnected is the disconnection time.

  FIG. 4 is an explanatory diagram of the interface test, and particularly shows the disconnection of communication in the interface test.

  As described above, the switch control unit 12 turns the switch unit 11 on or off according to the parameters set in the parameter setting unit 13. In the case of FIG. 4, as described above, in the switch unit 11, the switch 112 configuring the communication path 52 is turned off.

  Specifically, the switch control unit 12 repeats the same test for the test count T at the test interval T0. Therefore, test # 1 to test #T are executed. At the test interval T0, the switch unit 11 is turned on, for example. Note that the switch unit 11 may be turned off at the test interval T0.

  In one test, the communication time Ton and the disconnection time Toff are alternately repeated. The communication time Ton is a time during which communication between the server 2 and the disk device 3 is enabled when the switch unit 11 is turned on. The disconnection time Toff is a time during which communication between the server 2 and the disk device 3 is disabled when the switch unit 11 is turned off.

  The communication time Ton is, for example, constant. In other words, Ton (1) = Ton (2) =... Ton (N + 1). Here, for example, Ton (1) is the first communication time.

  On the other hand, the cutting time Toff is increased by an increase amount Δt until the maximum value N · Δt is reached each time the switch unit 11 is switched on and off in one test. The initial value of the cutting time Toff is an initial value t0. Therefore, the first cutting time Toff (1) is t0, and the second cutting time Toff (2) is t0 + Δt.

  Actually, the increase amount Δt and the initial value t0 are equal. Therefore, the maximum value N · Δt of the cutting time Toff is t0 + (N−1) Δt. For example, N is 256 and Δt = t0 is 1 millisecond.

  The increase amount Δt and the initial value t0 are times determined based on the retry time of the server 2. Generally speaking, the increase amount Δt and the initial value t0 are approximately equal to the retry time of the server 2. However, it is required that one retry can be completed within the time. Therefore, strictly speaking, the increase amount Δt and the initial value t0 are slightly longer than the retry time of the server 2.

  On the other hand, when N is 256, for example, the maximum value 256 · Δt is required to be a value that can be retried 256 times within that time, but cannot be executed 257 times. Therefore, strictly speaking, the difference S between the increase amount Δt or the initial value t0 and the retry time of the server 2 is a time smaller than a value obtained by dividing the retry time by N = 256. The value of the difference S changes according to the value of N when the value of N is a value other than 256.

  FIG. 5 is an explanatory diagram of the interface test, and particularly shows a case where the communication between the server 2 and the disk device 3 is normal as a result of the normal communication paths 51 and 52.

  The server 2 transmits a write command to the disk device 3 via the communication path 51. The disk device 3 that has received the write command transmits a command response to the server 2 via the communication path 52.

  The server 2 that has received the command response transmits a plurality of write data to the disk device 3 via the communication path 51. The disk device 3 that has received the plurality of write data transmits a status to the server 2 via the communication path 52.

  The server 2 that has received the status transmits a status response to the disk device 3 via the communication path 51. As described above, the communication between the server 2 and the disk device 3 ends normally.

  FIG. 6 is an explanatory diagram of the interface test, and particularly shows a case where communication between the server 2 and the disk device 3 is not normal as a result of the communication paths 51 and 52 being not normal. In this case, as described above, in the switch unit 11, the switch 112 configuring the communication path 52 is turned off.

  The server 2 transmits the first write command # 1 to the disk device 3 via the communication path 51. The disk device 3 that has received the write command # 1 transmits the first command response # 1 to the server 2 via the communication path 52. After the transmission of the write command # 1, the server 2 executes a time monitoring # 1 process for receiving the command response # 1 with respect to the write command # 1.

  However, as described above, since the switch 112 configuring the communication path 52 is off, the server 2 cannot receive the command response # 1 for the write command # 1. For this reason, the server 2 detects the elapse of the time-out prescribed time for the command response # 1, in other words, the time-out of the process of the time monitoring # 1. Note that the specified timeout time is the same for any time monitoring.

  Here, as can be seen from FIG. 6, the retry time is, for example, a retry timeout prescribed time, in other words, a time for detecting the timeout of the process of time monitoring # 1, and is normalized as described above.

  The retry time may be, for example, a time obtained by adding a predetermined margin, for example, a time for transmitting a write command by retry, to a retry timeout specified time. As described above, when the retry time itself is a time obtained by adding a margin to the retry timeout time, the increment Δt and the initial value t0 may be equal to the retry time of the server 2.

  The server 2 that detects the timeout retries transmission of the write command. As a result, the number of retries, in other words, the number of write command transmissions is two. Specifically, the server 2 transmits a second write command # 2 to the disk device 3, and executes a time monitoring # 2 process for receiving the second command response # 2. However, since the switch 112 configuring the communication path 52 is off, the server 2 cannot receive the command response # 2 for the write command # 2. Therefore, the server 2 detects the time-out of the time monitoring # 2 process for the command response # 2.

  Similarly, the server 2 retries transmission of the write command. As a result, the number of retries is three. Specifically, the server 2 transmits a third write command # 3 to the disk device 3 and executes a time monitoring # 3 process for receiving the third command response # 3. However, as described above, the server 2 cannot receive the command response # 3. For this reason, the server 2 detects a time-out of the process of the time monitoring # 3 for the command response # 3.

  Thereafter, for example, as shown in FIG. 7 described later, when the number of retries is determined to be five, the time monitoring process for the command response is performed even though the server 2 transmits the fifth write command. Timeout is detected. In this case, the server 2 executes a retry-out process.

  On the other hand, for example, when the number of retries is set to 5 as shown in FIG. 7 described later, the server 2 receives the command response after transmitting the fifth write command. In this case, the server 2 executes a recovery process.

  FIG. 7 is an explanatory diagram of the interface test, and particularly shows communication retry and retry out.

  As described above, the initial value t0 is 1 millisecond (ms), the increment Δt is 1 millisecond, the maximum value N · Δt is 256 milliseconds, the test number T is 100 times, and the test interval T0 is 10 milliseconds. Suppose that

  In FIG. 7A, first, the switch control unit 12 of the interface test apparatus 1 sets the switch unit 11 in a state in which communication between the server 2 and the disk device 3 is possible. In other words, the switch 112 is turned on.

  In this state, the server 2 and the disk device 3 execute communication as shown in FIG. In other words, transmission / reception of a write command, command response, a plurality of write data, status, and status response is executed. The server 2 that has received the status response transmits the write command again. As a result, the server 2 and the disk device 3 repeat the communication shown in FIG.

  In the repetition of the communication shown in FIG. 5, it is assumed that the communication time Ton has passed at a timing after receiving a certain write command and before transmitting a command response. In response to this, the switch control unit 12 puts the switch unit 11 into a state in which communication between the server 2 and the disk device 3 is impossible. In other words, the switch 112 is turned off. Since it is the first communication disconnection, the disconnection time is 1 millisecond (ms) which is the initial value t0.

  As a result, the server 2 cannot receive the command response despite the transmission of the write command. Therefore, the server 2 retries transmission of the write command.

  As described above, the cutting time is slightly longer than the retry time. In other words, the switch 112 is off when the server 2 executes the first retry, but the switch 112 is turned on immediately after that, and the switch 112 is turned on when the server 2 executes the second retry. It is. Therefore, the server 2 cannot return to normal operation by the first retry, but returns to normal operation by the second retry.

  Thus, as a test result, one retry is executed at the first disconnection time OFF # 1, but the server 2 does not return to the normal operation, and the recovery process is performed by the second retry at the subsequent communication time Ton. Is executed, and the server 2 returns to normal operation. Although this test result is not recorded, a normal output storage unit may be provided separately from the error output storage unit 24 and stored therein.

  Next, the server 2 that has returned to normal operation repeats the communication shown in FIG. 5 with the disk device 3. Furthermore, in the repetition of this communication, it is assumed that the communication time Ton has passed at a timing after the disk device 3 receives a certain write command and before transmitting a command response. In response to this, the switch 112 is turned off. Since the communication is disconnected for the second time, the disconnection time is 2 milliseconds (ms) obtained by adding Δt to the initial value t0.

  Since the server 2 cannot receive the command response, the server 2 retries transmission of the write command. In this case, the switch 112 is off when the server 2 executes the second retry, but immediately after that, the switch 112 is turned on, and when the server 2 executes the third retry, the switch 112 is on. is there. Therefore, the server 2 returns to normal operation by the third retry. As a test result, two retries are executed at the second disconnection time OFF # 2, but the server 2 does not return to normal operation, and a recovery process is executed by the third retry at the subsequent communication time Ton. Thus, the server 2 returns to normal operation.

  Next, in the process in which the server 2 that has returned to normal operation repeats the communication shown in FIG. 5, the communication time Ton elapses and the switch 112 is turned off as in the second disconnection of communication. Since this is the third communication disconnection, the disconnection time is 3 milliseconds (ms) obtained by adding 2Δt to the initial value t0. In this case, the switch 112 is OFF when the server 2 executes the third retry, but the switch 112 is ON when the server 2 executes the fourth retry. Therefore, the server 2 returns to normal operation by the fourth retry. As a test result, three retries are executed at the third disconnection time OFF # 3, but the server 2 does not return to normal operation, and the recovery process is executed by the fourth retry at the subsequent communication time Ton. Thus, the server 2 returns to normal operation.

  Next, in the process in which the server 2 that has returned to normal operation repeats the communication shown in FIG. 5, the communication time Ton elapses and the switch 112 is turned off as in the second disconnection of communication. Since this is the fourth communication disconnection, the disconnection time is 4 milliseconds (ms) obtained by adding 3Δt to the initial value t0. In this case, the switch 112 is OFF when the server 2 executes the fourth retry, but the switch 112 is ON when the server 2 executes the fifth retry. Therefore, the server 2 returns to normal operation by the fifth retry. As a test result, four retries are executed in the fourth disconnection time OFF # 4, but the server 2 does not return to normal operation, and the recovery process is executed by the fifth retry in the subsequent communication time Ton. Thus, the server 2 returns to normal operation.

  Next, in the process in which the server 2 that has returned to normal operation repeats the communication shown in FIG. 5, the communication time Ton elapses and the switch 112 is turned off as in the second disconnection of communication. Since this is the fifth communication disconnection, the disconnection time is 5 milliseconds (ms) obtained by adding 4Δt to the initial value t0. In this case, the switch 112 is off when the server 2 executes the fifth retry. Therefore, five retries are executed in the fifth disconnection time OFF # 5, but the server 2 does not return to normal operation.

  Here, it is assumed that the retry count of the server 2 is set to 5 times. As described above, the number of retries is not standardized and cannot be known.

  In accordance with the stipulation of the number of retries, the server 2 does not execute the sixth retry and does not execute the recovery process. Therefore, the server 2 does not return to normal operation. Instead, the server 2 executes a retry-out process.

  Thereby, as a test result, five retries are executed in the fifth disconnection time OFF # 5, but the server 2 does not return to the normal operation, and then the retry-out process is executed. The test result is output to the error output storage unit 24. Further, in the sixth and subsequent disconnection times OFF # 6..., Similarly, five retries are performed, but the server 2 does not return to normal operation, and the test result indicates that the retry-out process has been performed. Is obtained.

  As described above, for example, for the test in which the number of tests T = 100, there is no continuous record of the retry-out process for the cutting time OFF # 1 to the cutting time OFF # 4, but the cutting time OFF # 5 to the cutting time OFF #. A record that there is a continuous record of retry-out processing for 100 is obtained in the error output storage unit 24. This is the correct error information.

  As a result, it can be known that the number of retries of the server 2 as the device under test is 5 times. Further, the server 2 as the device under test can normally know that five retries have been executed normally and four recovery processes have been executed normally.

  Here, the operations of the server 2 and the disk device 3 are asynchronous with the operations of the interface test apparatus 1. In other words, the timing for transmitting the write command and its command response and the timing for disconnecting the communication paths 51 and 52 by the switch unit 11 are independent of each other. Thereby, the evaluator of the interface test can set the parameter of the interface test in the parameter setting unit 13 first, and thereafter can execute the interface test without manual intervention.

  However, actually, as described with reference to FIG. 7A, the communication time Ton elapses and the switch 112 is turned on at a timing after the disk device 3 receives a certain write command and before transmitting a command response. Not necessarily off. Further, the cutting time at that time is not always the initial value t0 = 1 millisecond (ms).

  Therefore, as shown in FIG. 7B, the switch control unit 12 performs tests # 1 to #T shown in FIG. 4 by repeating the same test after the test interval T0 has elapsed. Thereby, in any of the tests # 1 to #T, it can be expected with a high probability that the state shown in FIG.

  For example, it is assumed that the state shown in FIG. In this case, the error information obtained in the test #A and stored in the error output storage unit 24 has no continuous record of the retry-out process for the cutting time OFF # 1 to the cutting time OFF # 4, and In the fifth cutting time OFF # 5, five retries are performed, and then the retry-out process is performed, and in the sixth and subsequent cutting times OFF # 6. Indicates that it has been executed. As described above, this is correct error information, in other words, error information when the state shown in FIG. As a result, the evaluator of the interface test can know that the number of retries is five in the A-th test #A as shown in FIG. 7A.

  FIG. 8 is an interface test processing flow executed by the interface test apparatus.

  Prior to the start of the interface test, the control computer 4 sets the initial value t0, the increment Δt, the maximum value N · Δt, the communication time Ton, the number of tests T, and the test interval T0 in the parameter setting unit 13 of the interface test apparatus 1. Store.

  In response to this, the switch control unit 12 reads the initial value t0, the increase amount Δt, the maximum value N · Δt, the communication time Ton, the number of tests T, and the test interval T0 from the parameter setting unit 13 (step S1) and holds them. To do.

  Thereafter, the switch control unit 12 turns on the switch 111 configuring the communication path 51 and the switch 112 configuring the communication path 52 in the switch unit 11 (step S2). As a result, communication via the communication paths 51 and 52 is possible. Further, the switch control unit 12 monitors the passage of the communication time Ton.

  On the other hand, the server 2 starts communication with the disk device 3 as shown in FIGS. Thereby, the communication between the server 2 and the disk device 3 is executed as shown in FIG.

  Thereafter, the switch control unit 12 sets “1” as the number of tests, and starts monitoring the progress of the test interval T0 (step S3).

  When the test interval T0 has elapsed, the switch control unit 12 sets an initial value t0 for the disconnection time Toff, and starts monitoring the passage of the communication time Ton (step S4).

  When the communication time Ton has elapsed, the switch control unit 12 disconnects the communication between the server 2 and the disk device 3 and starts monitoring the elapse of the disconnection time Toff (step S5). Specifically, the switch control unit 12 turns off the switch 112 configuring the communication path 52 in order to disconnect communication between the server 2 and the disk device 3.

  As a result, the command response from the disk device 3 to the server 2 is disconnected between the server 2 and the disk device 3 as shown in FIG. As a result, since the server 2 cannot receive the command response, the server 2 retries transmission of the write command.

  Furthermore, in this case, when the disconnection time Toff is longer than (retry time) × (retry count), the server 2 cannot receive a command response even if the retry is repeated by the number of retries. Therefore, the server determines that it is a retry-out, executes a retry-out process, and stores error information in the error output storage unit 24.

  When the disconnection time Toff has elapsed, the switch control unit 12 restores the communication path 52 disconnected in step S5 and starts monitoring the passage of the communication time Ton (step S6). Specifically, the switch control unit 12 turns on the switch 112 configuring the communication path 52 in order to restore communication between the server 2 and the disk device 3.

  As a result, the server 2 receives a command response from the disk device 3 as shown in FIG. As a result, the server 2 returns to normal operation.

  Thereafter, the switch control unit 12 determines whether or not the current cutting time Toff is the maximum value N · Δt (step S7). When the cutting time Toff is not the maximum value N · Δt (No in step S7), the switch control unit 12 sets a value obtained by adding the increase Δt to the current cutting time Toff as a new cutting time Toff (step S8). ), Step S5 is repeated. As a result, the disconnection time Toff increases by an increment Δt each time the switch unit 11 switches between a state in which communication is possible and a state in which communication is not possible. Note that the timing for increasing the cutting time Toff can be changed.

  When the cutting time Toff is the maximum value N · Δt (step S7 Yes), the switch control unit 12 determines whether or not the test number is the test number T (step S9). When the test count is not the test count T (No in step S9), the switch control unit 12 sets a value obtained by adding “+1” to the current test count as a new test count, and sets the test interval T0 after the communication time Ton has elapsed. Is started (step S10), and step S4 is repeated.

  As described above, even if the test is repeated for the number of times T, correct error information is not always obtained. In this case, for example, the value of N for the maximum value is changed from 256 to 1024, the test count T is changed from 100 to 1000, and the interface test is repeated. Even in this case, the burden on the evaluator of the interface test is only to set the parameters of the interface test first and then change the parameters.

DESCRIPTION OF SYMBOLS 1 Interface test apparatus 2 Server 3 Disk apparatus 11 Switch part 12 Switch control part 13 Parameter setting part 121 Timer 122 Time increase control part 131 Initial value storage part 132 Increase amount storage part 133 Maximum value storage part 134 Communication time storage part 135 Test frequency Storage unit 136 Test interval storage unit

Claims (5)

A state inserted between the first transmission / reception device and the second transmission / reception device, enabling communication between the first transmission / reception device and the second transmission / reception device, or disabling communication. A switch part that is in any of the states of
It is a predetermined initial value for a disconnection time that is a time during which the state in which communication is disabled is continued, and the first transmission / reception device causes the communication to be normally performed due to the state in which communication is disabled. The initial value, which is the time required for one retry execution, which is an operation to execute the communication again when not performed, and a predetermined increase amount for the disconnection time, and the execution of the retry A parameter setting unit for storing an increase amount which is a required time;
Each time of switching of the state to not the communication with the state that allows the communication, based on said increment and said initial value stored in the parameter setting unit, a pre-Symbol disconnect time, interface test apparatus characterized by comprising a switch control unit that performs control for increasing by the increment in the switch unit. The switch control unit, when the disconnection time has reached a predetermined maximum value, when switching from a state that enables the first communication after that to the state that disables the communication, The interface test apparatus according to claim 1 , wherein a cutting time is set to the initial value. The switch unit is in a state in which the communication between the first transmission / reception device and the second transmission / reception device is disabled by disconnecting communication from the first transmission / reception device. The interface test apparatus according to claim 1, wherein: The switch unit disables the communication between the first transmission / reception device and the second transmission / reception device by disconnecting a response from the second transmission / reception device to the first transmission / reception device. The interface test apparatus according to claim 1, wherein the interface test apparatus is in a state to perform. A state inserted between the first transmission / reception device and the second transmission / reception device, enabling communication between the first transmission / reception device and the second transmission / reception device, or disabling communication. An interface test method using an interface test apparatus including a switch unit that is in any one of the following states, and a switch control unit that controls the switch unit,
The switch control unit is stored in the parameter setting unit each time switching between the state enabling the communication and the state disabling the communication, and is the time during which the state disabling the communication continues. A predetermined initial value for the disconnection time, which is an operation in which the first transmission / reception device re-executes the communication when the communication is not normally performed due to the state in which the communication is impossible. , An initial value that is the time required for one retry execution, and a predetermined increase amount that is stored in the parameter setting unit and that is related to the cutting time and that is the time required for the retry execution a step of bets based on a prior SL disconnect time is increased by the increase amount,
Controlling the switch unit so that the switch control unit switches to a state in which the communication is disabled;
The switch control unit includes a step of controlling the switch unit so as to switch to a state in which the communication is possible when the increased disconnection time has elapsed.

Images