0

I ran the following command on a 6 core machines

stress -c 10
dd if=/dev/zero of=/dev/null # 10 times

the loads on the system was as shown below

without Cgroups

I created two cgroups and added their pids to them

for i in $(pidof stress |tr ' ' '\n' ); do echo $i > writelimit/tasks; done
for i in $(pidof dd |tr ' ' '\n' ); do echo $i > lowlimit/tasks; done

jabir@host /sys/fs/cgroup # cat writelimit/cpu.shares 
900
jabir@host /sys/fs/cgroup  # cat lowlimit/cpu.shares 
100

But the CPU% of stress cmd has also dropped significantly with this change

with Cgroups

I expected stress to still use most of the CPU% and dd to drop as shown in the image, whats causing stress also to drop ? any pointers

FYI : i am running on a virtualized environment.

Jabir Ahmed
  • 82
  • 1
  • 1
  • 5

1 Answers1

0

the issue was with cpuset.cpus, had only one cpu in each of the cpuset.cpus

Jabir Ahmed
  • 82
  • 1
  • 1
  • 5