Cpu metrics values higher than expected

Hi there

I’m using Prometheus to scrape my nomad clients and I’ve reviewing my jobs.

In particular I was trying to graph the cpu its used, I used ‘nomad.client.allocs.cpu.total_percent’ and ‘nomad.client.allocs.cpu.total_ticks’ The problem that I’m see is that the raw values look to exceed the physical specs of the machine.

Does anyone know how this can be?

for example total_ticks should be MHz and the host client machine has 4 cpus totaling 10396MHz yet I have data points in the 13k.

Any ideas?

Gary