Expose information from Linux /proc pseudo filesystem.
At this point, the only thing that is implemented is reading from /proc/stat to get CPU usage statistics.
linux_proc_stat is a gen_server that reads the file periodically (every second by default)
and store the current cpu information.
You can then call linux_proc_stat:cpu_ratios/0 to get the current CPU usage statistics.
It returns a map with the following fields with the ratio of time spent in each state:
user Time spent with normal processing in user mode.
nice Time spent with niced processes in user mode.
system Time spent running in kernel mode.
idle Time spent in vacations twiddling thumbs.
iowait Time spent waiting for I/O to completed. This is considered idle
time too. Since 2.5.41
irq Time spent serving hardware interrupts. See the description of
the intr line for more details. Since 2.6.0
softirq Time spent serving software interrupts. Since 2.6.0
steal Time stolen by other operating systems running in a virtual
environment. Since 2.6.11
guest Time spent for running a virtual CPU or guest OS under the
control of the kernel. Since 2.6.24
guest_nice Time spent running a niced guest (virtual CPU for guest operating
systems under the control of the Linux kernel). Since 2.6.33
This is most useful to get the load on the system for use in rate limiting.
For example, if the system has significant iowait time, then avoid reading and
writing to disks. If the system has significant user time, then avoid taking
on new batch jobs.
Since this module is Linux specific, it is not started by default. You can conditionally add it to your application's supervision tree based on the OS.
Maybe check for the existence of /proc/stat or {unix, linux} = os:type().
%{id: linux_proc_stat, start: {linux_proc_stat, start_link, []}},