Use metersphere to keep your testing work efficient

Editor's note ： This paper is about CSDN Blogger weixin_41974278 The original article of .

Link to the original text ：

https://blog.csdn.net/weixin_41974278/article/details/124174009

Introduce yourself , I am a testing team of a medical industry company Leader, Leading a 5 Human test team . This team is mainly responsible for the testing of the company's products , Include App test 、 Black box test, etc , It will also cooperate with the R & D department to conduct some white box tests .

Because the product iteration cycle is short , This leads to heavy testing tasks before the launch , Busy with work, there is no time to calm down and think well . It happened that this epidemic was forced to be quarantined at home , The work task is suddenly much less , Just take advantage of this time to calm down and review the problems encountered at work , Also calculate “ A blessing in disguise ” Well .

Our team are all quasi senior test engineers who have worked for more than three years , I usually use it Postman、JMeter Interface 、 Performance testing , I think the efficiency of our testing department is good . Our classmates in the R & D department also have some technical heroes , Code ability is very strong , R & D efficiency is also very high . From the perspective of internal resources , Functional development 、 A functional test 、 Performance testing 、 The efficiency of each link such as requirement review is very high . But from a business perspective ？ That's another thing .

Practical problems in testing

First , The problem we face is that efficiency is concentrated locally , And difficult to sustain .

From the perspective of users , Users all hope to realize their requirements or needs as soon as possible , Solve their most urgent problems . But when we switch to this perspective , Although we think the local efficiency is good internally , But users' perceptions are different . in other words , Partial efficiency does not necessarily lead to efficient product delivery .

As an organization , Can the efficiency of each part be transformed into efficient delivery ？ Among them, the cooperation between departments 、 R & D quality 、 Test quality 、 Whether rework is required , Will result in “ Local efficiency is not equal to efficient delivery ”. Of course , In order to solve the problem of efficient delivery , Usually before going online , Or set up a project delivery team temporarily when doing the project , The station is also close , In order to solve the problem of communication . This really achieves temporary efficiency , But this efficiency is unsustainable .

second , The tests we have “ assets ” It is scattered and old .

From the perspective of our testing team , The test cases we maintain can be reused in the future , And there will be no quality problems when reusing . We used to call these things “ Test assets ”.

Since it is called as an asset , Then we certainly hope that they can generate positive interest . But in reality , They are often not assets to us , It's debt . It took us a long time to maintain , But it doesn't accumulate assets well . Our team's test cases are scattered everywhere , Some are TAPD On , Some use it by themselves Excel To maintain , Often too lazy to upload synchronization . These are all factors that hinder me from conducting continuous and efficient testing .

Based on the above two problems , I started using MeterSphere One stop open source continuous testing platform . I have communicated with their product team online before , But it hasn't been used properly , I've tried it for a few days and found it more appropriate to my needs , So write this article for your mutual encouragement and reference .

Use MeterSphere The actual process of

MeterSphere Open source continuous testing platform （metersphere.io） It mainly covers test tracking 、 The interface test 、UI test 、 Performance testing 、 Teamwork and other functions . Because it is compatible JMeter As the underlying platform , So there is no technical barrier in use . Don't talk much , Tell you how I use it .

install MeterSphere I won't repeat the process of , There are... On the official website , If not clear, you can refer to MeterSphere Online installation documentation （
https://metersphere.io/docs/installation/online_installation/）. Find one 4 nucleus 8GB Of Linux The machine can be installed with one key , Particularly convenient .

After installation , Enter the initial user name and password admin/metersphere, You can log in normally , The login interface is shown in the following figure ：

 

You can see from the navigation box above ,MeterSphere It is mainly divided into the following work modules ：

My workbench ： It belongs to the enterprise version function , No open source version , The entry to apply for enterprise trial is ：
https://metersphere.io/enterprise.html.

Test track ： Including test case creation and management 、 Use case review 、 Test plan tracking and defect management . This module allows users to create / Import function use cases , Including review function . In this way, you can get through the test 、 Research and development 、 Department walls of product managers and project managers , Let multiple people participate , Unreasonable demand can be eliminated in the early stage . The reviewed use cases can be sorted into a test plan for implementation , The test plan contains functional use cases 、 Interface and performance test cases . The test plan can also be executed regularly and Jenkins Execute by triggering , It can also be established in the test Bug Correspondence with use cases , In this way, the current test assets are well connected .

The interface test ： What surprises me most about this module is the establishment API And Case The relationship between , It greatly alleviates our dilemma of testing asset accumulation . Put all test case assets in API And Case The relationship between , Click on API after , Naturally, you can see the corresponding Case. in addition , This module also includes the implementation of interface testing and scenario testing , Achieve the integration of management and implementation .

Performance testing ：MeterSphere The base is JMeter, The user can turn the interface test into performance test with one click , Of course you upload your own JMX And CSV Files can also create performance tests , And it can automatically generate reports .

Report statistics ： Show the project use case trend and use case status statistics .

Project settings ： Show the information and documents used in the current project , Thus, it is convenient to call .

System settings ： Including user and workspace management , You can also configure messages and nails 、 Get through with enterprise wechat , It also supports configuring external requirements , docking Bug platform , for example TAPD、JIRA、 Zen way, etc .

■ Create functional use cases

Get into “ Test track ” modular , Get into “ Function use case ” You can import and create test cases . Fill in the basic information according to the prompts 、 Step details , You can also edit the dependencies with other functional use cases , Associate with other interfaces 、 Corresponding relationship of performance test , You can also upload attachments .

 

■ Use case review

Then we enter the function review stage . This is also a function I like very much , Let R & D 、 The test and even the father of Party A can participate in the project review stage . Everyone can leave a comment , Unreasonable requirements are eliminated in the review stage , It reduces the possibility of subsequent rework . This can effectively solve the problem I mentioned earlier “ Local efficiency is not equal to business efficiency ” The problem of .

 

After the use case review , The use case passed the review can be considered as executable . Then you can carry out the test plan . This process is more in line with the collation logic .

in addition ,“ The interface test ” The module shows API And Case The hierarchy of , This function really helped me a lot .

I take /userlogin Such an interface will explain to you . Obvious , This is a user login POST The interface of the way , We type in {username:$username,Password:$Password} Of data Group test , After passing the test API The hierarchy is over .

But the little buddy who is familiar with the test must know , The greater workload lies in the subsequent different situations Case Level test . For example, take the one just /userlogin Interface examples , We use the correct user name 、 Is the test over after the password test succeeds ？ Obviously not , We will also try to login with the wrong user name and password , Whether the error message can be returned normally , Whether the blank password login will prompt “Please enter password”. If only key login is allowed , Enter the correct password to check whether the login will take effect .

It can be said that a simple user login API, Dozens of methods will be derived , Over time, this Case The accumulation of will cost a lot of manpower to manage . You say no management , Do you want to rewrite it if you want to use it in the future ？ You say management , Who will bear the high management cost ？

MeterSphere That's a good way to solve the problem . It is the API And Case It becomes a parent-child relationship , First we pass Swagger UI Import MeterSphere Of API, find /userlogin Interface , You can see API And Case Two dimensions .

 

After clicking in, we create two Case, respectively “ Enter the correct user name and password , Whether the login success is displayed normally ” as well as “ Whether the wrong user name and password are displayed normally, login failure ”. You can see the correct user name and password ,Case It was a successful execution , Enter the wrong username and password , Expected to return wrong Reponse Code, It also shows that the execution passed .

■ Use secret correctly Case

 

■ Wrong secret Case

 

You can also set assertion rules , Designate this Case Expected output Response Code The value is 500, In this way, even if the execution fails , But in Case The execution level is also judged to be successful .

 

■ /userlogin Corresponding Case

 

So we start from the previous API and Case The level of parallelism has become today API and Case One to many relationship , It simplifies our management of use cases , Enhanced reusability . also ,MeterSphere The characteristics of platform make the test assets of all our users unified 、 Persist , Achieve what I said earlier “ Testing continues to be efficient ” Idea .

 

MeterSphere It also supports the scenario arrangement of interface use cases . When it passes API and Case Interface after two-layer test , You can think connectivity is ok , And it can return the expected value under different limit conditions . We can test the scenario of joint debugging and parameter transmission between interfaces .

 

in addition , What surprises me more is MeterSphere The interface test of can be converted into performance test with one click . So you don't have to go from Postman After the test , Put the script 、 Copy the interface information into the performance test tool , There is no need to consider the compatibility of script functions between the two tools , It greatly reduces the learning cost and the difficulty of getting started .

■ Interface test is transformed into performance test

 

■ Automatic generation JMX file

 

summary

After this trial , In my submission MeterSphere It is a testing platform tool suitable for medium and large-scale teams . It can manage the test assets hierarchically , Covers performance testing 、 The interface test 、 Basic test capability and use case management of function test . also , be based on MeterSphere The ability to platform itself , You can use user data in a workspace 、 Categorize project dimensions , Ensure data sharing and isolation .

however ,MeterSphere At present, there are few out of the box protocols for the project , Only common HTTP、TCP、Dubbo And SQL form , The performance report is also relatively simple , Not included JMeter Advanced configuration chart for . Looking forward to the future MeterSphere Our product team can optimize in these places .