I'm tryng to stress a web server with requests, I used vmstat to get the memory usage of the server, but I don't understand few things.

  1. When the requests start, the free memory decreases and the swpd increases, but after a while, even if the requests continue at the same rate, the free memory goes up. I think it's because the system goes out of free memory, or at least to avoid that before it's to late, allocates new memory with swap memory, am I right?
  2. Is there a way to limit the swap memory allocation?

Memory Usage Compared

  • 1) Could you add the graph for mem used by application too ? 2) Why do you want to reduce swap allocation ? Do you have bad performance ? – Soren A Feb 2 '19 at 16:57
  • 1) Unfortunately no 2) I want to test the performances without the swap memory – simone bianco Feb 2 '19 at 17:01
  • As long as you have free memory swap doesn't impact performance in any way. But if you wan't, you can disable swap bu running sudo swapoff -a – Soren A Feb 2 '19 at 17:07
  • Perfect thanks. Also is there a way to set a specific dimension, same for buffer? – simone bianco Feb 2 '19 at 17:52
  1. No: Free memory is unused memory: you're not going out of memory at all.
  2. It's possible, but a bad idea: the kernel knows better than you when to use swap... ;-)

For more information on how the kernel uses memory, have a look here.

Not the answer you're looking for? Browse other questions tagged or ask your own question.