JP5039683B2 - Test support system - Google Patents

Description

  The present invention relates to a test support system, and more particularly to a technique for supporting an online test for verifying the overall operation of an online system in which a plurality of clients and servers are connected by a communication network.

Opportunities to provide various services to clients are increasing through an online system in which a plurality of clients and servers are connected via a communication network.
FIG. 18 is a system configuration diagram showing an example of such an online system. The online system 80 includes a client 12 equipped with a web browser program, a PL (Presentation Logic) server 14, and a plurality of BLs (Business Logic). A server 16 (first BL server 16a to third BL server 16c) is provided. The client 12 and the PL server 14 and the PL server 14 and the BL server 16 are connected via a communication network.

After receiving the request data 30 from the client 12, the PL server 14 sends input data to the responsible BL server 16 to request processing, and sends the processing result data 32 returned from the BL server 16 to the client. Functions as a front-end server.
Therefore, the PL server 14 includes a screen control unit 20, a dispatch engine 21, and a connector 22.

The screen control unit 20 refers to the screen transition definition DB 23 and functions to generate a screen (Html file) to be transmitted to the client 12.
The dispatch engine 21 refers to the dispatch definition DB 24 and performs a function of specifying a transaction corresponding to the request data 30 and a responsible BL server.
The connector 22 also functions to convert the input data into a specific format required by each BL server 16 and to convert the processing result data transmitted from the BL server 16 into the standard format of the PL server 14. For this reason, a corresponding data conversion unit (first BL server data conversion unit 22a to third BL server data conversion unit 22c) is provided for each BL server 16.

  Each BL server 16 functions as a back-end server that executes business processing such as search and update on the database 28 that is directly or indirectly managed through a DB server (not shown) based on the input data received from the PL server 14. .

  In FIG. 18, the data (1) included in the request data 30 transmitted from the client 12 is transmitted to the first BL server 16a to request a predetermined transaction process, and the data (2) which is the process result data is transmitted. The second BL server 16b is transmitted to request a predetermined transaction processing, and the processing result data (3) is transmitted to the third BL server 16c to request a predetermined transaction processing, and the processing result A pattern in which data (4), which is data, is transmitted to the client 12 as final processing result data 32 is illustrated.

When constructing such an online system 80, it is a matter of course that unit tests and integration tests of individual application programs installed in each server are necessary. However, before the service is released, the client 12 and each server must be connected. It is necessary to verify the operation of the entire online system while being connected via a communication network.
In addition, when a new function is added to an existing online system or a bug is corrected, it is necessary to check whether other existing functions are adversely affected (degrade).

  In this so-called online test, an operator performs actual input from the client 12 in accordance with a plurality of scenarios prepared in advance, and it is verified whether various functions provided by the system operate normally. And when some malfunction arises, it returns to a point before a little, and inputs again, and the operation | work which locates a problem location and a cause is repeated.

In the case of a large-scale online system, processing is executed by cooperation of a large number of servers, so that test items are inevitably diversified and the possibility of occurrence of defects increases. As a result, many operators need to repeat the same input many times in order to complete confirmation of all items.
For this reason, as shown in Non-Patent Documents 1 and 2, several tools for reducing the input work by the operator have been proposed.
QuickTest Professional Internet URL: http://www.ashisuto.co.jp/prod/qtp/sum/index.html Search date: June 4, 2008 What is Selenium? Internet URL: http://www.thinkit.co.jp/free/article/0705/2/1/ Search date: June 4, 2008

This type of input support tool has the function of recording the input operation from the input device 31 by the operator and the display screen on the display device 33 in the input support DB 82 one by one, so that the operator does not re-input the same. The request data 30 can be re-entered into the server as many times as possible, and a new display screen can be compared with the previous display screen to determine whether or not they match.
Also, if the database 28 managed by each BL server 16 is backed up only once at the beginning, the state of the database 28 at that point is restored by reproducing the input up to any point based on this backup. It becomes possible to do.

  However, this conventional input support tool is an application program introduced for each client 12, and the data to be monitored is also included in the request data 30 transmitted from the client 12 and the processing result data 32 received by the client 12. There was a problem of being limited.

  In fact, even when one request data is received from the client 12, a plurality of BL servers 16 operate in cooperation with each other via the PL server 14 and often result in final processing result data. Regardless of the above, only the request data 30 and the processing result data 32 in FIG. 18 can be captured by the above input support tool, and at the level of data (1) to (4) generated in units of transactions in the PL server 14. The data could not be saved.

  For this reason, in the conventional input support tool, the rerun of the test is limited to the request unit, and the test cannot be reexecuted in units of transactions by each BL server 16. As a result, as shown in FIG. 19, even if the state of a certain database 28 is actually updated in the order of (A) → (B) → (C) → (D) by transactions α to γ. The conventional input support tool reproduces only (i) → (d) update, and it is impossible to restore the states (b) and (c).

  In order to solve this problem, the applicant stores the input data for each BL server 16 and the output data from each BL server in the PL server 14 one by one. Has previously proposed a technology that reproduces the behavior of the machine (Japanese Patent Application No. 2008-252054).

Specifically, as shown in FIG. 1, a test DB 26 is provided in the PL server 14 and a test support unit 25 is provided in the dispatch engine 21 so that data passing through the dispatch engine 21 during the first run of the test ( The test support unit 25 supplements all of 1) to (4) and stores them in the test DB 26.
When the test is rerun, as shown in FIG. 4, the data (1) to (3) stored in the test DB 26 are retrieved by the test support unit 25, and the data is transferred to the BL via the dispatch engine 21 and the connector 22. Since it is passed to the server 16, input from the client 12 can be made unnecessary.

In this way, since input data and processing result data are stored in units of transactions, it is possible to execute test reruns in detail in units of transactions, and the database is updated in units of transactions as shown in FIG. It will be possible to confirm the appearance of the incident.
Further, since the test result data (2) ′ to (4) ′ at the time of re-execution of the test is also stored in the test DB 26 by the test support unit 25, the result data (2) to (4) and It is also possible to compare by transaction.

  By the way, in an online system, a function called “optimistic exclusive lock” is generally provided for the purpose of preventing a plurality of clients from accessing the same database at the same time and performing update processing contradictory to each other. It has been.

FIG. 9 is a diagram for explaining the mechanism of this optimistic exclusive lock, and each record of the database 28 is provided with a data item 42 of “last update date”.
Here, as shown in FIG. 9 (a), when the search request data 30 specifying “employee ID = 001” is transmitted from the client 12, the BL server 16 that has received this via the PL server 14 The record that matches the search condition is extracted from 28, and the search processing result data 32 is transmitted to the client 12 via the PL server 14. The search processing result data 32 includes a data item 43 of “update date”. It is provided and filled with “05/24/10: 00”, which is the value of “last update date / time” in the database 28.

When the update request data for the record of “employee ID = 001” is transmitted from the client 12, the above “update date and time” is also taken over for this data.
For example, as shown in FIG. 9 (b), at 14:00 on May 24, update request data 30 is sent from the client 12 to change the employee's affiliation from “sales department” to “general affairs department”. In this case, data “05/24/10: 00” is added.

  The BL server 16 that has received the update request data 30 compares the update date and time with the last update date and time of the corresponding record stored in the database 28, and confirms that they match. Perform the update. At this time, the last update date and time of the record is updated to “05/24/14: 00”. The update processing result data 32 is transmitted to the client 12, and the latest “05/24/14: 00” is recorded as the update date and time of this data.

  On the other hand, if the update date / time of the update request data 30 does not match the last update date / time on the database 28 side, it means that the record has been updated by another client in the meantime. An error message is transmitted from the client to the client 12, and the data update is rejected.

Instead of the above time stamp, a unique management code may be given to the processing result data for each processing. In any case, by providing such an optimistic exclusive lock mechanism, it is possible to effectively suppress the occurrence of contradiction in the database included in the online system.
However, if this test support system is applied as it is to an online system equipped with this optimistic exclusive lock mechanism, a problem occurs during rerun.

  First, FIG. 10 shows the procedure at the time of the first run. As shown in FIG. 10 (a), the update request data (update date and time :) from the client 12 at the time of 05/24/14: 00. 05/24/10: 00) is transmitted, the BL server 16 receiving this via the PL server 14 compares the update date / time of the update request data (A) with the last update date / time of the database 28 side, After confirming that both match, an update process is executed to change the affiliation from “sales department” to “general affairs department”. At this time, the last update date / time of the record is also updated to “05/24/14: 00”.

At the same time, update processing result data (B) is transmitted from the BL server 16 to the client 12 via the PL server 14. “05/24/14: 00” that is the last update date / time is also recorded in the update date / time of the update processing result data (B).
During this time, update request data (A) and update processing result data (B) are stored in the test DB 26 by the test support unit 25 (not shown).

  Next, as shown in FIG. 10 (b), when the update request data (update date and time: 05/24/14: 00) of (C) is transmitted from the client 12 at 05/24/16: 00. The BL server 16 compares the update date and time of the update request data (C) with the last update date and time on the database 28 side, confirms that they match, and then changes the affiliation from “General Affairs Department” to “Legal Affairs Department”. Update processing to be changed to At this time, the last update date and time of the record is also updated to “05/24/16: 00”.

At the same time, update processing result data (D) is transmitted from the BL server 16 to the client 12 via the PL server 14. “05/24/16: 00” which is the last update date / time is also recorded in the update date / time of the update result data (D).
During this time, update request data (C) and update result data (D) are stored in the test DB 26 by the test support unit 25.

  Next, FIG. 11 shows the procedure at the time of rerun. As shown in FIG. 11A, the update request data (update) of (A) is received from the PL server 14 at 05/24/21: 00. Date / time: 05/24/10: 00), the BL server 16 compares the update date / time of the update request data (A) with the last update date / time of the database 28 side. Since the data backed up immediately before the first run is used at the time of rerun, both dates and times naturally match, so the BL server 16 executes an update process to change the affiliation from “Sales Department” to “General Affairs Department” To do. At this time, the last update date and time of the record is updated to “05/24/21: 00” of the current time.

At the same time, update processing result data (B) ′ is transmitted from the BL server 16 to the PL server 14. “05/24/21: 00” that is the last update date / time is also recorded in the update date / time of the update processing result data (B) ′.
During this time, update result data (B) ′ is stored in the test DB 26 by the test support unit 25.

Next, as shown in FIG. 11 (b), update request data (update date and time: 05/24/14: 00) of (C) is transmitted from the PL server 14 at 05/24/23: 00. Then, the BL server 16 compares the update date and time of the update request data (C) with the last update date and time on the database 28 side. In this case, since the update date / time does not match, an error message is transmitted from the BL server 16 to the PL server 14, and the update process of the database 28 is naturally rejected.
During this time, the error message (D) ′ is recorded in the test DB 26 as rerun result data by the test support unit 25.

In short, if this test support system is applied as it is under the optimistic exclusive lock mechanism, the following problems arise.
(1) Since the update date and time of the input data stored in the test DB 26 at the first run and the last update date and time of the database updated by the BL server 16 at the time of rerun do not match, an optimistic exclusive lock is activated and Run result data is an error.
(2) Since the current date and time at the time of rerun is given to the update date and time of rerun result data, when comparing the process result data of each transaction with the rerun result data, as shown in FIG. It is determined that “comparison result = mismatch” because of the mismatch of the update date and time.

  The present invention provides an optimistic exclusive lock for a test support system that enables re-execution of a test and restoration of a database in units of transactions by storing input data and processing result data in units of transactions of a back-end server. An object of the present invention is to solve the above-described problems that occur when applied to an online system having a mechanism.

In order to achieve the above object, a test support system according to claim 1 is necessary for a client, a back-end server that executes business processing for a database, and request data from the client, and is required for a responsible back-end server. Input to the back-end server during the first run of the test in order to test the online system with a front-end server that sends the input data and requests the processing, and sends the processing result data to the client The data and the processing result data sent from the backend server are accumulated in the test database in the frontend server, and the input data at the first run stored in this test database is stored in the backend when the test is rerun. When sending to the server And a test support system having a function of registering the rerun processing result data transmitted from the back-end server in the test database, and the back-end server returns the process result data to the front-end server. At the same time, a unique management code assigned to the corresponding record in the database is assigned to the processing result data, and the front-end server sends input data based on the processing result data to the back-end server to perform subsequent processing on the record When the request is made, the above management code is carried over to the input data, and the back-end server is requested only when the management code included in the input data matches the management code of the corresponding record in the database. Optimistic exclusive lock mechanism In this case, the management of subsequent processing input data accumulated in the test database at the first run is performed by a new management code added to the preceding processing result data transmitted from the back-end server at the time of the rerun of the test. It is characterized by the function of transmitting this input data to the backend server after overwriting the code.
The above-mentioned “first run” means a first online test performed based on request data input from a client. On the other hand, “rerun” means a second or later online test replayed based on the data of the first run accumulated in the database.

  The test support system according to claim 2 is the system according to claim 1, wherein the test support system further includes processing result data at the first run and processing result data at the time of rerun stored in the test database. And comparing the management code given to the processing result data at the first run and the management code given to the processing result data at the rerun. It is characterized by processing.

  The test support system according to claim 3 is the system according to claim 1 or 2, wherein the management code is a time stamp given to a corresponding code of a database managed by a back-end server. .

  In the test support system according to claim 1, when the input data accumulated in the test database at the first run is transmitted to the back-end server at the time of the rerun, it is given to the processing result data for the preceding input data. Since the management code at the time of rerun will be taken over, it will match the management code at the time of rerun on the database side, and it will be possible to effectively avoid an error determined by the optimistic exclusive lock mechanism. .

  In the test support system according to claim 2, when comparing the processing result data at the time of the first run and the processing result data at the time of the rerun, each management code portion is masked, so that it is optimistic. Due to the mismatch of the management codes given for the purpose of exclusive lock, it is possible to effectively avoid the inconvenience determined as a mismatch as a whole despite the fact that the other substantial data portions match.

  FIG. 1 shows an online system 10 including a test support system according to the present invention, which includes a client 12 loaded with a Web browser program, a PL server 14, and a plurality of BL servers 16 (first BL server 16a, first A second BL server 16b, a third BL server 16c), and a test terminal 18. The client 12 and the PL server 14, the PL server 14 and the BL servers 16a to 16c, and the PL server 14 and the test terminal 18 are connected via a communication network.

The PL server 14 receives the request data 30 from the client 12, identifies the transaction necessary for realizing the request and the responsible BL server 16, and then sends input data (argument) to the responsible BL server 16. Function as a front-end server that sends a screen including the final processing result data 32 returned from the BL server 16 to the client 12.
Therefore, the PL server 14 includes a screen control unit 20, a dispatch engine 21, and a connector 22.

  The screen control unit 20 mainly functions to generate a screen (Html file) to be transmitted to the client 12 with reference to the screen transition definition DB 23.

The dispatch engine 21 performs a function of referring to the dispatch definition DB 24 and specifying a transaction corresponding to a request from the client 12 and a responsible BL server 16.
A test support unit 25 is generated in the dispatch engine 21 and manages the test DB 26 provided in the disk of the PL server 14. The function of the test support unit 25 and the generation method thereof will be described later.

The connector 22 functions to convert input data into a unique format required by each BL server 16 and to convert processing result data transmitted from the BL server 16 into a standard format of the PL server 14. Therefore, a plurality of data converters (first BL server data converter 22a to third BL server data converter 22c) are provided for each of the BL servers 16a to 16c.
For example, when the first BL server 16a is equipped with a TP monitor corresponding to the C language, the first BL server data conversion unit 22a has a data conversion function for the TP monitor. When the second BL server is equipped with an EJB (Enterprise Java Beans) container, the second BL server data conversion unit 22b has a data conversion function for the EJB container. When the third BL server 16c is equipped with a DB / BC system compatible with the COBOL language, the third BL server data conversion unit 22c has a data conversion function for the host OS.

  The BL servers 16a to 16c function as back-end servers that execute business processes such as search and update for the database 28 that is directly or indirectly managed through a DB server (not shown) based on input data received from the PL server 14. To do.

Next, the processing procedure at the time of the first run for the online system 10 will be described according to the flowchart of FIG.
In the first run, request data 30 is first transmitted from the client 12. For example, when the operator inputs a search character string via the input device 31 in the input field of the search screen according to the test scenario, the request data including the search request ID and the search character string is sent from the client 12 to the PL server 14. 30 is sent.

  The screen control unit 20 that has received the request data 30 (S10) acquires the request ID included in the request data 30 (S11). Next, the screen control unit 20 refers to the screen transition definition DB 23 and identifies the PL transaction ID corresponding to the request ID (S12). This PL transaction ID is passed to the dispatch engine 21.

  Upon receiving the PL transaction ID from the screen control unit 20, the dispatch engine 21 searches the dispatch definition DB 24 using this PL transaction ID as a key, and specifies the BE transaction ID associated with the PL transaction ID (S13).

  The BE transaction ID defines a specific process (transaction) for a specific BL server, and one or more BE transaction IDs are associated with the PL transaction ID in the order of processing. For this reason, when the dispatch engine 21 identifies a plurality of BE transaction IDs based on the PL transaction ID, the dispatch engine 21 recognizes the transactions necessary for responding to the current request, each responsible BL server 16, the order of the transactions, etc. Can do.

  Here, data (1), which is a search character string transmitted from the client 12, is transmitted to the first BL server 16a to request a predetermined search process, and data (2), which is the process result data, is stored in the second data. To the BL server 16b and request a predetermined search process, and the data (3) which is the process result data is transmitted to the third BL server 16c to request the predetermined search process, and the process result data It is assumed that a processing procedure for transmitting certain data (4) to the client 12 as final processing result data 32 is defined in the dispatch definition DB 24.

  In this case, the dispatch engine 21 first outputs the data (1) to the first BL server data converter 22 a of the connector 22. The first BL server data converter 22a converts this data (1) into a data format for the first BL server 16a, and then transmits it to the first BL server 16a (S14).

In parallel with this, the test support unit 25 captures the data (1) and stores it in the test DB 26 in association with the search transaction and the information specifying the first BL server 16a (S15). Specifically, as shown in FIG. 3, data (1) is registered in the IN data (= input data) field of BE transaction ID “111”.
In this case, the BE transaction ID “111” corresponds to information specifying the search transaction and the first BL server 16a.
As illustrated, the test DB 26 stores control information including a request ID, a PL transaction ID, and a BE transaction ID for each transaction.

  Next, when processing result data is returned from the first BL server 16a (S16), the data is converted into the standard format for the PL server 14 in the first BL server data conversion unit 22a (S17), and the data ( It is output to the dispatch engine 21 as 2).

  At this time, the test support unit 25 captures the data (2) and stores it as the processing result data of the search transaction from the first BL server 16a in the test DB 26 (S18). Specifically, data (2) is registered in the item of OUT data (= processing result data) of BE transaction ID “111”.

  Next, the dispatch engine 21 outputs the data (2) to the second BL server data converter 22b of the connector 22 in accordance with the above dispatch definition. The second BL server data converter 22b converts this data (2) into a data format for the second BL server 16b, and then transmits it to the second BL server 16b (S19).

At the same time, the test support unit 25 captures the data (2), and stores it in the test DB 26 in association with the search transaction and information specifying the second BL server 16b (S20). Specifically, data (2) is registered in the IN data item of BE transaction ID “222”.
In this case, the BE transaction ID “222” corresponds to information specifying the search transaction and the second BL server 16b.

  When processing result data is returned from the second BL server 16b (S21), the data is converted into a standard format for the PL server 14 in the second BL server data conversion unit 22b (S22), and then data (3) Output to the dispatch engine 21.

  At this time, the test support unit 25 captures the data (3) and stores it in the test DB 26 as search transaction processing result data from the second BL server 16b (S23). Specifically, data (3) is registered in the OUT data item of BE transaction ID “222”.

  Next, the dispatch engine 21 outputs the data (3) to the third BL server data converter 22c of the connector 22 in accordance with the above dispatch definition. The third BL server data converter 22c converts this data (3) into a data format for the third BL server 16c, and then transmits it to the third BL server 16c (S24).

At the same time, the test support unit 25 captures the data (3), and stores it in the test DB 26 in association with the search transaction and the information specifying the third BL server 16c (S25). Specifically, data (3) is registered in the IN data item of BE transaction ID “333”.
In this case, the BE transaction ID “333” corresponds to information specifying the search transaction and the third BL server 16c.

  When processing result data is returned from the third BL server 16c (S26), the data is converted into the standard format for the PL server 14 in the third BL server data conversion unit 22c (S27), and then the data (4). Output to the dispatch engine 21.

  At this time, the test support unit 25 captures the data (4) and stores it as test result data from the third BL server 16c in the test DB 26 (S28). Specifically, data (4) is registered in the OUT data item of BE transaction ID “333”.

This data (4) is embedded in the processing result display screen as the final request processing result data 32 by the screen control unit 20 and transmitted to the client 12 (S29).
As a result, the processing result data 32 is displayed on the display device 33 of the client 12.

  As described above, when one request data 30 is transmitted from the client 12, the test support unit 25 causes the test DB 26 to store the first BL server 16a, the second BL server 16b, and the third BL. Input data and processing result data related to the server 16c are accumulated in units of transactions.

  Of course, it is not limited to the case where a plurality of BL servers 16 are sequentially requested for a predetermined process, or when a plurality of processes are continuously requested to one BL server 16, or to a plurality of BL servers 16. Even when a plurality of processes are requested simultaneously in parallel, the test support unit 25 records the input data and the processing result data in the test DB 26 for each transaction.

Next, a case where a similar test is rerun based on input data and processing result data for each transaction collected by the first run will be described.
FIG. 4 is a system configuration diagram of the online system 10 at the time of rerun, and expresses that rerun is executed without involvement of the client 12 or the screen control unit 20.

Hereinafter, the processing procedure at the time of rerun will be described according to the flowchart of FIG.
First, when the rerun instruction data specifying the PL transaction ID “1111” is transmitted from the test terminal 18, the test support unit 25 receiving this (S40) specifies the rerun execution range (S41). . Here, since the PL transaction ID “1111” is a superordinate concept including the BE transaction IDs “111” to “333”, the test support unit 25 determines that rerun should be performed within the same range as the initial run.

  Therefore, the test support unit 25 takes out the data (1) which is the input data of the BE transaction ID “111” from the test DB 26 (S42) and passes it to the dispatch engine 21.

  The dispatch engine 21 outputs the data (1) to the first BL server data converter 22 a of the connector 22. The first BL server data conversion unit 22a converts this data (1) into a data format for the first BL server 16a, and then transmits it to the first BL server (S43).

  When the processing result data is returned from the first BL server 16a (S44), the data is converted into the standard format for the PL server 14 in the first BL server data conversion unit 22a (S45), and the data (2 ) 'Is output to the dispatch engine 21.

  This data (2) ′ is captured by the test support unit 25 and stored in the test DB 26 (S46). Specifically, as shown in FIG. 6, data (2) ′ is registered in the rerun result data item of BE transaction ID “111”.

  Next, the test support unit 25 takes out the data (2) which is the input data of the BE transaction ID “222” from the test DB 26 (S47) and passes it to the dispatch engine 21.

  The dispatch engine 21 outputs the data (2) to the second BL server data converter 22b of the connector 22. The second BL server data converter 22b converts this data (2) into a data format for the second BL server 16b, and then transmits it to the second BL server 16b (S48).

  When the processing result data is returned from the second BL server 16b (S49), the second BL server data conversion unit 22b converts the data into the standard format for the PL server 14 (S50), and the data (3 ) 'Is output to the dispatch engine 21.

  This data (3) ′ is captured by the test support unit 25 and stored in the test DB 26 (S51). Specifically, data (3) ′ is registered in the rerun result data item of BE transaction ID “222”.

  Next, the test support unit 25 takes out the data (3) which is the IN data of the BE transaction ID “333” from the test DB 26 (S52) and passes it to the dispatch engine 21.

  The dispatch engine 21 outputs the data (3) to the third BL server data converter 22c of the connector 22. The second BL server data conversion unit 22c converts this data (3) into a data format for the third BL server 16c, and then transmits it to the third BL server 16c (S53).

  When the processing result data is returned from the third BL server 16c (S54), the data is converted into the standard format for the PL server 14 in the third BL server data conversion unit 22c (S55), and the data (4 ) 'Is output to the dispatch engine 21.

  This data (4) ′ is captured by the test support unit 25 and stored in the test DB 26 (S56). Specifically, data (4) ′ is registered in the rerun result data item of BE transaction ID “333”.

  Next, the test support unit 25 compares the processing result data (OUT data) of each BE transaction ID stored in the test DB 26 with the rerun result data at the time of the first run (S57), and displays a list of the comparison results. The data is transmitted to the test terminal 18 (S58).

  For example, if a first run is performed before a bug is corrected and a rerun is performed after the bug is corrected, it can be confirmed whether or not the bug correction has caused an adverse effect. That is, there is no problem if all the comparison result lists are “match”, but there is a possibility that degradation has occurred in the case of “mismatch”, and therefore it is necessary to perform rerun at a finer level.

In this test support system, rerun can be performed individually for each transaction.
For example, when rerun instruction data specifying the BE transaction ID “222” is transmitted from the test terminal 18, the test support unit 25 is input data “222” from the test DB 26 as shown in FIG. Data (2) is extracted and passed to the dispatch engine 21.

  The dispatch engine 21 outputs the data (2) to the second BL server data converter 22b of the connector 22. The second BL server data converter 22b converts the data (2) into a data format for the second BL server 16b, and then transmits the data format to the second BL server 16b.

When the processing result data is returned from the second BL server 16b, the second BL server data conversion unit 22b converts the processing result data into a standard format for the PL server 14, and then dispatches the dispatch engine 21 as data (3) '. Is output.
This data (3) ′ is captured by the test support unit 25 and registered in the rerun result data item of the BE transaction ID “222” in the test DB.

  In this way, a rerun is performed for each transaction, and the processing result data at the first run and the rerun result data are compared, thereby comparing the input data with the processing result data for each request. In comparison, the operation of the BL server 16 can be verified at a finer level.

In addition, since the BL server 16 can be rerun in units of transactions as described above, according to this test support system, the state of the database 28 can be restored in units of transactions.
In other words, since the database 28 managed by each BL server 16 is backed up immediately before the first run, the necessary transactions are sequentially updated based on the backed up data to update each database 28. It is possible to check the condition each time.

  FIG. 8 exemplifies the update status for a certain address management database 28, in which “TEL” is updated by transaction α, “FAX” is updated by transaction β, and “address” is updated by transaction γ. In some cases, by rerunning transactions α to γ one step at a time, it is possible to individually confirm how the database 28 has been updated as (b) → (b) → (c) → (d). .

  On the other hand, when the conventional input support tool is used, rerun can be executed only in units of requests, and rerun cannot be executed in units of transactions of the BL server 16, as shown in FIG. In addition, only the states (a) and (c) in the database 28 were compared, and the states at points (b) and (c) could not be confirmed.

By the way, as described above, in the online system, there is a function called “optimistic exclusive lock” for the purpose of preventing a plurality of clients from accessing the same database at the same time and executing mutually contradictory update processes. Generally provided, a problem occurs when the test support unit 25 is operated as it is.
The following is a brief description of the optimistic exclusive locking mechanism and problems, and then how to solve them.

First, as shown in FIG. 9, under the optimistic exclusive lock mechanism, a data item 42 of “last update date” is provided in each record of the database 28.
Here, as shown in FIG. 9 (a), when the search request data 30 specifying “employee ID = 001” is transmitted from the client 12, the BL server 16 that has received this via the PL server 14 The record that matches the search condition is extracted from 28, and the search processing result data 32 is transmitted to the client 12 via the PL server 14. The search processing result data 32 includes a data item 43 of “update date”. It is provided and filled with “05/24/10: 00”, which is the value of “last update date / time” in the database 28.

When the update request data for the record of “employee ID = 001” is transmitted from the client 12, the above “update date and time” is also taken over for this data.
For example, as shown in FIG. 9 (b), at 14:00 on May 24, update request data 30 is sent from the client 12 to change the employee's affiliation from “sales department” to “general affairs department”. In this case, data “05/24/10: 00” is added.

  The BL server 16 that has received the update request data 30 compares the update date and time with the last update date and time of the corresponding record stored in the database 28, and confirms that they match. Perform the update. At this time, the last update date and time of the record is updated to “05/24/14: 00”. The update processing result data 32 is transmitted to the client 12, and the latest “05/24/14: 00” is recorded as the update date and time of this data.

  On the other hand, if the update date / time of the update request data 30 does not match the last update date / time on the database 28 side, it means that the record has been updated by another client in the meantime. An error message is transmitted from the client to the client 12, and the data update is rejected.

  By providing such an optimistic exclusive lock mechanism, it is possible to effectively suppress the occurrence of inconsistencies in the database included in the online system, but this test support is provided to an online system equipped with this optimistic exclusive lock mechanism. If the system is applied as it is, there will be a problem during rerun.

  First, as shown in Fig. 10 (a), the update request data (update date and time: 05/24/10: 00) of (A) is sent from the client 12 at 05/24/14: 00 at the time of the first run. Then, the BL server 16 that has received this via the PL server 14 compares the update date and time of the update request data (A) with the last update date and time on the database 28 side and confirms that they match. , Update processing to change affiliation from “Sales Department” to “General Affairs Department” is executed. At this time, the last update date / time of the record is also updated to “05/24/14: 00”.

At the same time, update processing result data (B) is transmitted from the BL server 16 to the client 12 via the PL server 14. “05/24/14: 00” that is the last update date / time is also recorded in the update date / time of the update processing result data (B).
During this time, update request data (A) and update processing result data (B) are stored in the test DB 26 by the test support unit 25 (not shown).

  Next, as shown in FIG. 10 (b), when the update request data (update date and time: 05/24/14: 00) of (C) is transmitted from the client 12 at 05/24/16: 00. The BL server 16 compares the update date and time of the update request data (C) with the last update date and time on the database 28 side, confirms that they match, and then changes the affiliation from “General Affairs Department” to “Legal Affairs Department”. Update processing to be changed to At this time, the last update date and time of the record is also updated to “05/24/16: 00”.

At the same time, update processing result data (D) is transmitted from the BL server 16 to the client 12 via the PL server 14. “05/24/16: 00” which is the last update date / time is also recorded in the update date / time of the update result data (D).
During this time, update request data (C) and update result data (D) are stored in the test DB 26 by the test support unit 25.

  Next, at the time of rerun, as shown in FIG. 11 (a), the update request data (update date and time: 05/24/10: 00) of (A) is received from the PL server 14 at 05/24/21: 00. When transmitted, the BL server 16 compares the update date and time of the update request data (A) with the last update date and time on the database 28 side. Since the data backed up immediately before the first run is used at the time of the rerun, both dates and times naturally match, so the BL server 16 executes an update process to change the affiliation from “Sales Department” to “General Affairs Department” . At this time, the last update date and time of the record is updated to “05/24/21: 00” of the current time.

At the same time, update processing result data (B) ′ is transmitted from the BL server 16 to the PL server 14. “05/24/21: 00” that is the last update date / time is also recorded in the update date / time of the update processing result data (B) ′.
During this time, update result data (B) ′ is stored in the test DB 26 by the test support unit 25.

Next, as shown in FIG. 11 (b), update request data (update date and time: 05/24/14: 00) of (C) is transmitted from the PL server 14 at 05/24/23: 00. Then, the BL server 16 compares the update date and time of the update request data (C) with the last update date and time on the database 28 side. In this case, since the update date / time does not match, an error message is transmitted from the BL server 16 to the PL server 14, and the update process of the database 28 is naturally rejected.
During this time, the error message (D) ′ is recorded in the test DB 26 as rerun result data by the test support unit 25.

From the above, if this test support system is applied as it is under the optimistic exclusive lock mechanism, the following problems arise.
(1) Since the update date and time of the input data stored in the test DB 26 at the first run and the last update date and time of the database updated by the BL server 16 at the time of rerun do not match, an optimistic exclusive lock is activated and Run result data is an error.
(2) Since the current date and time at the time of rerun is given to the update date and time of rerun result data, when comparing the process result data of each transaction with the rerun result data, as shown in FIG. It is determined that “comparison result = mismatch” because of the mismatch of the update date and time.

  On the other hand, this test support system solves the above-mentioned problem by applying the methods of “takeover” and “masking” with respect to the update date and time.

  First, “takeover” means that, as shown in FIG. 12A, using the update date and time “05/24/21: 00” assigned to the rerun result data (B) ′ by the BL server 16, This is a method of overwriting the update date and time of “05/24/14: 00” given at the time of the first run to the input data (C), and then transmitting to the BL server 16. The takeover of the update date and time is executed by the test support unit 25.

As a result, the update date and time of the update request data (C) and the last update date and time on the database 28 side match, and the update process for changing the affiliation of employee ID = 001 from “General Affairs Department” to “Legal Affairs Department” It will be executed correctly in the BL server 16.
Further, update result data (D) ′ is transmitted from the BL server 16 to the PL server 14. In the update date / time of the update result data (D) ′, “05/24/23: 00” which is the last update date / time is recorded. This update result data (D) ′ is registered in the rerun result data item of the test DB 26 by the test support unit 25.

Next, as shown in FIG. 13 (a), “masking” means that the update date and time are different when the processing result data (B) stored in the test DB 26 and the rerun result data (B) ′ are compared as they are. Means that a comparison is made after performing a mask process on the update date and time of both data, as shown in FIG. 13 (b). Specifically, it corresponds to replacing the update date of both data with a common character string “***” or the like.
This mask processing is also executed by the test support unit 25.

As a result, due to inconsistency in the values of the update date and time items that are given for the sake of optimistic exclusive lock, the entire data is determined to be inconsistent even when there are no changes in the values of the remaining substantial data items. It is possible to avoid the inconvenience.
Although illustration is omitted, masking processing is performed on each update date and time when comparing the processing result data (D) and the rerun result data (D) ′.

As described above, the test support unit 25 is incorporated in the dispatch engine 21, captures data passing through the dispatch engine 21 and stores it in the test DB 26, and retrieves necessary data from the test DB 26 during rerun. A function to send to the dispatch engine 21 and request transmission to the specified BL server 16, a function to capture the data transmitted from the BL server 16 and store it in the test DB 26, and compare it with the data at the first run However, it is not necessary at all after the test.
For this reason, it is not preferable to incorporate the code for the test support unit 25 in the application for the dispatch engine 21 in advance.
Also, because the environment such as machine configuration and line capacity is often different between the production and test, the application settings are usually rewritten for production after the test is completed. There was a risk of contamination.

  Therefore, in this system 10, the test support unit 25 is incorporated from the outside into the dispatch engine 21 only at the time of testing, without changing the application program itself to be tested, and setting information for testing arranged outside is also provided. Adopting a mechanism to reflect.

  FIG. 14 is a schematic diagram for explaining this mechanism. In the disk area 50 of the PL server 14, a dispatch engine execution file 51, a test support section library 52, and a test setting file 53 are stored. Yes. The dispatch engine execution file 51 includes a production setting file 54.

Further, an upper class loader 57 is provided in the memory space 56 of the PL server 14, and a switch module 58 is arranged in advance in the upper class loader 57.
This upper class loader 57 is realized based on the specification of J2EE (Java 2 Enterprise Edition) which is an operating environment of Java (registered trademark), and the modules arranged in the upper class loader 57 are memory It has a feature that it can be referred to from all applications developed in the space 56.
The switch module 58 is automatically placed in the upper class loader 57 when a JVM (Java Virtual Machine) is started by storing a library for the switch module in a predetermined directory of the PL server 14 in advance.

Next, a procedure for incorporating the test support unit 25 and reflecting the setting information for testing will be described with reference to the flowchart of FIG.
First, the dispatch engine execution file 51 is activated, and the framework 60, the specific function unit 61, and the production setting information 62 are arranged in the WAR class loader 59 in the memory space 56 (S51). Under the J2EE environment, each Web application is managed in units of WAR class loaders.

Next, the framework 60 checks whether or not the switch module 58 exists in the upper class loader 57 (S52).
If the switch module 58 exists (S53 / YES), the framework 60 recognizes that it is in the test mode, and loads the test support unit library 52 into its own WAR class loader 59 (S54). Then, the test support unit 25 is generated.

  The test support unit 25 reads the corresponding test setting file 53 from the disk area 50 of the PL server 14 into the memory space 56 (S55), and overwrites the actual setting information expanded in the WAR class loader 59 with the test setting information. To do.

  As a result of the above, as shown in FIG. 16B, in the WAR class loader 59 in the memory space, in addition to the specific function unit 61 and the framework 60 in charge of functions specific to the dispatch engine, the test support unit 25 At the same time, the setting information 62 for production is switched to the setting information 63 for testing.

After the online test is completed, the switch module library is deleted, the JVM is restarted, and the switch module 58 in the upper class loader 57 is removed, and the processing of S54 to S56 is skipped. As shown in FIG. 16A, a framework 60, a specific function unit 61, and production setting information 62 are arranged in the WAR class loader 59.
As a result, the dispatch engine 21 can be switched to the production specification without modifying the dispatch engine execution file 51 (Web application).

  Since online systems generally have a mechanism for inter-model cooperation that links multiple servers (including host systems) with different implementation languages and frameworks, as shown in FIG. Has connected dedicated testers 70a to 70c to each of the BL servers 16a to 16c, and has confirmed the IN / OUT data for each BL server. For this reason, there was a problem that a plurality of testers had to be prepared and each operation method had to be mastered.

  On the other hand, in the case of this test support system, input data is transmitted after being converted into a format for each BL server 16a to 16c in the connector 22 as described above, and from each BL server 16a to 16c. The test support unit 25 can perform the tests of various BL servers 16 based on the standard format in a unified manner because the processing result data is converted into the standard format at the connector 22. Only the operation method of the test support unit 25 needs to be learned.

It is a system configuration diagram showing the configuration at the time of the first run of the test support system according to the present invention. It is a flowchart which shows the process sequence at the time of the first run of an online test. It is explanatory drawing which shows the data item example of test DB. It is a system configuration figure showing the composition at the time of rerun of a test support system. It is a flowchart which shows the process sequence at the time of the rerun of an online test. It is explanatory drawing which shows the data item example of test DB. It is a conceptual diagram which shows the point of the rerun in a test assistance system. It is explanatory drawing which shows a mode that the database of BL server is updated per transaction at the time of a test run system rerun. It is a schematic diagram explaining the general mechanism of an optimistic exclusive lock. It is a schematic diagram which shows a problem at the time of operating the optimistic exclusive lock in this test support system. It is a schematic diagram which shows a problem at the time of operating the optimistic exclusive lock in this test support system. It is a schematic diagram which shows the method of avoiding the problem at the time of operating optimistic exclusive lock in a test support system. It is a schematic diagram which shows the method of avoiding the problem at the time of operating optimistic exclusive lock in a test support system. It is a schematic diagram which shows the point at the time of incorporating an online support part in a dispatch engine and performing the setting for a test. It is a flowchart which shows the procedure at the time of giving an online support part to a dispatch engine and setting for a test. FIG. 10 is an explanatory diagram showing that the dispatch engine can be switched between a test mode and a production mode. It is a schematic diagram which shows the conventional test method which tests using a dedicated tester for every BL server from which language use and a framework differ. It is a system configuration diagram showing an example of a conventional online system. It is explanatory drawing which shows the reproducibility of the database in the conventional online test.

Explanation of symbols

10 Online system
12 clients
14 PL server
16 BL server
16a First BL server
16b Second BL server
16c Third BL server
18 Test terminal
20 Screen controller
21 dispatch engine
22 Connector
22a Data converter for the first BL server
22b Second BL server data converter
22c Third BL server data converter
25 Test Support Department
26 Test database
28 Database
30 Request data
31 Input device
32 Processing result data
33 Display device
42 Data item of "Last modified"
43 "Update Date / Time" data item
50 PL server disk space
51 Executable file for dispatch engine
52 Library for test support
53 Test configuration file
54 Production configuration file
56 PL server memory space
57 Upper class loader
58 Switch module
59 WAR class loader
60 Framework
61 Unique functions
62 Production setting information
63 Test setup information

Claims (3)

Receives request data from the client, the back-end server that performs business processing for the database, and the above client, sends the input data required to the back-end server in charge, requests processing, and sends the processing result data to the client When performing a test on an online system with a front-end server to send, the input data sent to the back-end server at the first run of the test and the processing result data sent from the back-end server are stored in the front-end server. Accumulated in the test database, when rerunning the test, the input data at the time of the first run stored in this test database is sent to the backend server and at the time of the rerun sent from the backend server Process result data up A test support system having a function to be registered in the test database,
When the backend server returns the processing result data to the frontend server, the unique management code assigned to the corresponding record in the database is added to the processing result data, and the frontend server inputs based on the processing result data. When sending the data to the back-end server and requesting the subsequent processing for the same record, the above management code is carried over to the input data, and the management code included in this input data is stored in the database at the back-end server. If it has an optimistic exclusive lock mechanism that executes the requested process only when it matches the management code of the corresponding record,
The management code of the input data for subsequent processing accumulated in the test database at the first run is overwritten by the new management code given to the previous processing result data sent from the backend server at the time of the rerun of the test. Then, a test support system having a function of transmitting the input data to the back-end server. The test support system has a function that compares the processing result data at the first run stored in the test database with the processing result data at the rerun and outputs the comparison result.
The test support according to claim 1, wherein a mask process is performed on the management code assigned to the processing result data at the time of the first run and the management code assigned to the processing result data at the time of the rerun for the comparison. system.   The test support system according to claim 1 or 2, wherein the management code is a time stamp given to a corresponding code of a database managed by a back-end server.

Images