0

So I have below set of CW Agent Metrics Collected from an AWS EC2 Instance for CPU.

time_active, time_guest, time_guest_nice, time_idle, time_iowait, time_irq, time_nice,time_softirq, time_steal, time_system, time_user, usage_active, usage_guest, usage_guest_nice, usage_idle, usage_iowait, usage_irq, usage_nice, usage_softirq, usage_steal, usage_system, and usage_user.

Looking at the metrics definition here I am considering usage_active as CPU Utilization % and usage_idle as the Idle CPU time %. However, for an instance when I am validating the CPU Utilization % with respect to the AWS/EC2 namespace metrics, I am getting a different value. [ The CWAgent namespace has usage_active as 100% where as the AWS/EC2 namespace says its 10.35% ]. Am I considering the wrong CPU Metric from CWAgent for CPU Utilization ?

Similarly, the Total Processes metrics (processes_total) is not adding up with Running Processes and Sleeping Processes. Also I am getting the Running Processes as 0.02. 0.03 etc. Why it is a fraction ?

Santosh Kumar Sahoo
  • 47
  • 6
  • What is your instance type used? Certain types such as m5.12xlarge and m5.24xlarge use intel_idle CPU drivers which cause some discrepancies in reported active and idle usage. See https://stackoverflow.com/questions/12111954/context-switches-much-slower-in-new-linux-kernels for example. – Rolo787 Dec 22 '22 at 19:17
  • @Rolo787 I have 4 t2.micro and 1 c5.xlarge instance running. In all these 5 instances CW Agent is running. – Santosh Kumar Sahoo Dec 23 '22 at 05:34

0 Answers0