英文:
if any specific request fails in the jmeter performance test with lots of users then what should we do to improve its performance?
问题
我正在运行一个脚本，在脚本中，当我增加线程数时，有两个特定请求失败了，为了使它们通过，我必须增加我的逐渐增加周期。
目前我正在使用线程数：300
逐渐增加周期为：1500秒
一切正常，所有请求都通过了。
但如果我减少逐渐增加周期，那么它就会失败。
是否应该采用这么长的逐渐增加周期，或者如何改进这些请求以使其通过？
在这里，应该选择什么样的逐渐增加时间合适？
英文:
i am running a script, in the script 2 specific requests are failing when i am increasing number of threads and in that case, make them pass i have to increase my ramp up period.
for now i am taking no. of threads: 300
and ramp up period is: 1500sec
and its all good, passing it all requests.
but if i take less no. of ramp up period then its failing.
is it good to take this much of ramp up period or how can i improve those requests so it can pass?
what should be the good ramp up time here?
答案1
得分: 0
增加ramp-up期的整个目的是逐渐增加负载。没有所谓的“好”或“坏”的ramp-up，一旦存在，任何时间都是“好”的。
在理想应用的情况下，当您增加负载时，吞吐量（每秒请求数）将成比例增加，响应时间将保持不变。
实际情况是，在某个时候，响应时间将开始增加，吞吐量将停止增长，甚至可能下降。这被称为饱和点。
通常应该有非功能性需求，比如系统应该支持X个用户，响应时间不超过Y。或者系统应该支持X个每秒的请求。
如果您面临高并发的问题，这表明被测试系统存在瓶颈，从JMeter方面增加ramp-up期将减少并发并掩盖问题。
英文:
The whole point of having ramp-up period is increasing the load gradually. There is no "good" or "bad" ramp-up, any period is "good" once it's there.
In case of ideal application when you increase the load the throughput (number of requests per second) will increase proportionally and response time will remain the same.
In reality at some point response time will start increasing and the throughput will stop growing or even will go down. It's known as the saturation point.
Normally there should be non-functional requirements like the system should support X users with response time not more than Y. Or the system should support X requests per second.
If you're facing the issue when concurrency is high it indicates the system under test bottleneck and there is nothing you can "fix" from JMeter side as increasing ramp-up period will decrease concurrency and mask the issue.
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论