The analysis process in which the performance pressure can not respond to a certain degree.

[phenomenon]: in today's LVB system performance test, it is found that the system performance pressure will continue to increase when the virtual user reaches 40, and the response will not go up when the response is raised to 230 seconds.

[analysis process]: judging from experience, this situation is not very normal. In general, you think of resource constraints, but in terms of application machine memory, load, cpu usage, and loadrunner client related resources (memory, cpu), it is still sufficient. Rule out resource reasons.

On the other hand, it is suspected that there is an anomaly in the application machine, moving the application back to another machine and re-stress testing, the situation is the same. This eliminates the problem with the machine.

Continue to suspect that some places are restricted. The first thing that comes to mind is whether or not the thread is a problem.

When the tool for monitoring threads is configured, the number of virtual users increases gradually, and when it increases to 40, the threads of time-waiting gradually appear, and continue to increase the number of virtual users, the greater the proportion found. It should be possible to navigate to thread-related here. Checked the thread configuration of resin

two hundred and fifty six

At the same time, use netstat-na | grep 192.xxx.xxx.xxx | wc-l to see that the number of links to the application has reached 231, and this problem has been located.