When running the Universal Forwarder on a physical Windows server with multiple CPUs, it looks like the agent selects a random CPU that the Universal Forwarder collects performance metrics from. Sometimes it collects data from CPU #1, sometimes from CPU #2.
If I have two CPUs and a client process that is also not NUMA aware that uses all of the CPU usage on ONE of the CPUs, the Universal Forwarder might be "connected" to the other CPU and only collecting CPU metrics from a potentially idle CPU.
Is this a problem others observe or is it just me?
