I am writing a program that is supposed help run foreign code "safely". I also want to get the memory utilization and CPU compute time for the foreign code so I thought I would use cgroups.
I created a cgroup at
/sys/fs/cgroup/cpuacct/<program name> and then add the executing foreign code's PID to the cgroups.procs file.
After the program executes I read
/sys/fs/cgroup/cpuacct/<program name>/cpuacct.usage to try and get the CPU time.
According to this link the number I am getting is in USER_HZ.
After doing some reading and looking through the man pages of sysconf and getconf I ran the following command
getconf CLK_TCK which output 100.
From the sysconf man pages this is "The number of clock ticks per second." So does that mean I can just take the number I got from cpuacct.usage and divide it by 100 to get my CPU compute time in seconds?
Because the numbers I am getting average out to about 270 million USER_HZs for a simple python program that creates a list of all the numbers between 0 and 1000 then prints the length of that list. And 2.7 million seconds doesn't sound right to me.
Can someone explain to me how to convert from USER_HZ to milliseconds? Or maybe I am going about this whole process wrong.
Thanks in advance!