迭代用户CSV文件 - 它会产生任何问题吗?(Iterating in the users CSV file - will it create any issues?)
我们有一个JMeter脚本,从头到尾运行大约1分钟。 它使用用户CSV文件来获取用户登录ID。 我们的目标是为900个虚拟用户运行脚本20分钟。 我们决定每个用户2s是一个很好的加速时间。
为了避免JMeter上出现任何瓶颈,我们设置了6个JMeter从站,并决定分拆150个用户。 现在,每个从属Jmeter机器都有自己独特的300个用户列表。
但令我们担忧的是:允许脚本在加速后运行20分钟(900 VU * 2秒/用户= 1800秒= 30分钟),我们需要执行脚本50分钟。 所以,我们决定迭代脚本50次。 现在,这将重复使用相同的用户ID,我们认为在某些时候,不同的用户线程将使用相同的用户(用户ID)。
有没有办法避免这种情况? 我知道我们可以为每个从站提供很长的唯一用户列表(例如,每个JMeter从站有4000个唯一用户,而不是300个),并确保没有两个线程使用相同的用户ID,但创建那么多用户是一个挑战。
还有其他建议吗?
两个人会分享会议吗? 实际上,如果我在浏览器的两个不同选项卡中使用同一用户登录,则从一个选项卡注销,会自动将我从另一个选项卡中注销。 但是JMeter therads会发生什么?
我希望我的问题对你有意义,如果没有,我会道歉。
谢谢。
We have a JMeter script that runs around 1 minute from start to end. It uses a users CSV file to get the user login id. Our target is to run the script for 900 virtual users for 20 minutes. We decided 2s per user would be a good ramp-up time.
To avoid any bottlenecks on the JMeter, we have set up 6 JMeter slaves and decided to split 150 users in each. Now, each slave Jmeter machine has it's own unique 300 users' list.
But what is concerning us is: to allow the script to run for 20 minutes after the ramp up (900 VU * 2 seconds/user = 1800 sec = 30 minutes) we need to execute the script for 50 minutes. So, we decided to iterate through the script 50 times. Now, that will use same user id repeatedly and we think at some point, different user thread will be using the same user (user id).
Is there a way to avoid this? I know we could give really long list of unique users to each slave (e.g., 4000 unique users for each JMeter slave, instead of 300) and make sure no two threads use the same user id, but creating that many users is a challenge.
Any other suggestions?
Will two tread share the session? In reality, if I login using the same user in two different tabs in a browser, logging out from one tab, automatically logs me out from the other. But what will happen for JMeter therads?
I hope my question makes sense to you, apologies if it does not.
Thanks.
原文: