v0.7.11 x86 w/PCP: kepler_node_platform_joules_total readings very low in Perf CoPilot

[jharriga]

What happened?

Running v0.7.11 on x86 Dell dual socket server (RHEL9.4)
Viewing PCP openmetrics values, specifically 'openmetrics.kepler.kepler_node_platform_joules_total'
Readings for 'kepler_node_platform_joules_total' (both sockets) are very low compared to Intel RAPL.

PACKAGE	RAPL	kepler_node_platform_joules_total	kepler_node_package_joules_total
Sample: 0-package	87.810	0.006	99.134
Sample: 1-package	90.907	0.242	95.421

What did you expect to happen?

Readings for 'kepler_node_platform_joules_total' would align with RAPL and 'kepler_node_package_joules_total'

How can we reproduce it (as minimally and precisely as possible)?

Install v0.7.11 on x86 system and compare power readings for 'kepler_node_platform_joules_total' and 'kepler_node_package_joules_total'

Anything else we need to know?

Product Name: PowerEdge R650
Intel(R) Xeon(R) Platinum 8352V CPU @ 2.10GHz (2 socket)
PRETTY_NAME="Red Hat Enterprise Linux 9.4 (Plow)"

sut# rpm -qa | grep kepler
container-kepler-0.7.11-1.noarch

Kepler image tag

sut# rpm -qa | grep kepler container-kepler-0.7.11-1.noarch

sut# curl localhost:8888/metrics | grep node_info

TYPE kepler_node_info counter

kepler_node_info{components_power_source="rapl-sysfs",cpu_architecture="Sunny Cove",platform_power_source="none",source="os"} 1

Kubernetes version

$ kubectl version
# paste output here

Cloud provider or bare metal

bare metal

OS version

sut# uname -a Linux perf-intel-28.perf.eng.bos2.dc.redhat.com 5.14.0-417.kpq1.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Fri Feb 2 14:05:04 EST 2024 x86_64 x86_64 x86_64 GNU/Linux

Install tools

sut# wget https://github.com/sustainable-computing-io/kepler/releases/download/v0.7.11/kepler.rpm.tar.gz 2024-07-15 09:26:16 (6.09 MB/s) - ‘kepler.rpm.tar.gz’ saved [12975/12975]

sut# tar xzvf kepler.rpm.tar.gz
RPMS/
RPMS/noarch/
RPMS/noarch/container-kepler-0.7.11-1.noarch.rpm
sut# systemctl stop container-kepler --now
sut# rpm -Uvh ./RPMS/noarch/container-kepler-0.7.11-1.noarch.rpm
Updating / installing...
1:container-kepler-0.7.11-1 ######################### [ 50%]
Cleaning up / removing...
2:container-kepler-0.7.10-1 ######################### [100%]
sut# rpm -qa | grep -i kepler
container-kepler-0.7.11-1.noarch

Kepler deployment config

For standalone:

put your Kepler command argument here

sut# systemctl start container-kepler --now
sut# podman ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
c9670d01a92a quay.io/sustainable_computing_io/kepler:latest 8 seconds ago Up 8 seconds kepler

Container runtime (CRI) and version (if applicable)

Related plugins (CNI, CSI, ...) and versions (if applicable)