currently facing a problem with mericbeat on a Windows Server 2019 with Hyper-V role installed. This server is a pure virtualization server, hosting a lot of VMs.
Using metricbeat 7.13.1 and have the following config in my system.yml module:
module: system period: 10s cpu.metrics: ["normalized_percentages"] metricsets: - cpu - memory
While the memory readings are consistent with what I'm seeing in taskmanager, the cpu metrics are completely off. I figured out, that the performance metrics don't "see" the CPU load created by all the virtual machines running on ther server ... so my CPU load is always around 1%. I tested with some heavy load on the server itself (running prime95) and the load goes up accordingly. Doing the same (running prime95 in a VM) doesn't increase the CPU load seen by metricbeat (but taskmanager shows it).
Is there a way to fix that?