The CSV used for JMeter performance test is bullshit

Many people are using jmeter Do interface tests 、 Automated testing and performance testing , All like to use CSV Data file setting function , To read the prepared test data . Although this method is not the best solution , In our performance testing course , Explained a better scheme , however , Testers who haven't taken our class , It is still widely used CSV The way .

today , Let's tell you a story about CSV Prepare a huge pit of test data and solutions .

jmeter Do a test , To read the prepared test data file , If you only think of CSV This is a way , Then use ‘CSV Data Set Config’ This component , Never use csvRead function , Use this function , You've been killed , Please don't say I didn't remind you .

however , Use ‘CSV Data Set Config’ This component , You can only start from... At a time 1 Start reading row data , Or ignore the first line from 2 OK, let's start , There's no way to start from scratch .
Every time from 1 Line start value , In daily low concurrency performance testing , There is no problem , Because you only need to run one jmeter Can , however , If you want to do high concurrency performance testing , There need to be multiple jmeter Running at the same time , Form a distributed to make requests to the server , Now there's a problem .

because , Distributed , Run simultaneously on multiple machines jmeter, Every machine will start from 1 Line start reading the test file you prepared , that , At the same time, multiple threads use the same test data to make requests to the server , This is likely to lead to various exceptions in your tested service .

Have you ever used distributed for performance testing , There are many 4xx Series error , The performance test performed by a single machine does not ; Have you ever appeared , The error rate increases when distributed , And when it's stand-alone , Same concurrent number , The error rate is much lower ; Have you ever appeared , Distributed request , Server log , There are a lot of lock conflicts ？

These problems , If you use... In your script ‘CSV Data Set Config’, So these mistakes , It is probably because this component can only read data from the... Every time 1 The reason why the line starts reading .

The pit , I once met , And it once made me miserable for a few days , however , Then I found 4 A solution , today , Let me tell you a method with the lowest technical difficulty .

️ open jmeter Plug in management , stay ‘available plugins’ Mid search ‘CSV’, Check one called ‘Random CSV Data Set’ Plug in for , Click on the ‘Apply Changes and Restart JMeter’ Download and install this plug-in . After downloading successfully and restarting automatically , stay jmeter In the configuration element of , We will find ‘bzm - Random CSV Data Set Config’ Components .

You can click ** Below ‘Test CSV Reading’ Let's look at the order of the data , You'll find that , The data order is different every time .** such , This solves the problem of the same data retrieval when distributed . The method is very simple .

Next , Let's compare the performance of this component .

First add ‘CSV Data Set Config’ Components , Read a copy greater than 10000 Row data file , Define a variable to receive , then , Use a debug sampler , Apply this variable , Located in 1000 Concurrent users , Keep running 120s, Save script , use CLI mode , View runtime resource usage .

From monitoring to data ,1000 Concurrent ,120 In seconds , In all 59830295 Requests , Average per second 498141.6 Time /s, My own machine CPU The utilization rate rose to about 48.3%, Memory usage is about 26.8%.

Now? , Replace the components with ‘bzm - Random CSV Data Set Config’, Everything else remains the same . use CLI Mode to run , View monitoring data .

From monitoring to data ,1000 Concurrent ,120 In seconds , In all 20860718 Requests , Average per second 172912.8 Time /s, My own machine CPU The utilization rate rose to about 66.5%, Memory usage is about 26.7%.

️ From the test results of the two components , Official CSV Data Set Config Performance is better than bzm - Random CSV Data Set Config, In practice , You can choose according to your actual situation .

Finally, thank everyone who reads my article carefully , The following online link is also a very comprehensive one that I spent a few days sorting out , I hope it can also help you in need ！

These materials , For those who want to change careers 【 software test 】 For our friends, it should be the most comprehensive and complete war preparation warehouse , This warehouse also accompanied me through the most difficult journey , I hope it can help you ！ Everything should be done as soon as possible , Especially in the technology industry , We must improve our technical skills . I hope that's helpful ……

If you don't want to grow up alone , Unable to find the information of the system , The problem is not helped , If you insist on giving up after a few days , You can click the small card below to join our group , We can discuss and exchange , There will be various software testing materials and technical exchanges .

Click the small card at the end of the document to receive it

Typing is not easy , If this article is helpful to you , Click a like, collect a hide and pay attention , Give the author an encouragement . It's also convenient for you to find it quickly next time .

Self study recommendation B Stop video ：

Zero basis transition software testing ：25 Days from zero basis to software testing post , I finished today , Employment tomorrow .【 Include features / Interface / automation /python automated testing / performance / Test Development 】

Advanced automation testing ：2022B The first station is super detailed python Practical course of automated software testing , Prepare for the golden, silver and four job hopping season , After advanced learning, it soared 20K