feat: add RAPL zone filtering support through configuration #2031

sthaha · 2025-04-28T04:33:55Z

Changes include:

Add RAPL zone filtering option in config (rapl/zones).
Add zone filtering in raplPowerMeter and log excluded and included zones
Update createCPUMeter to use RAPL config and filter zones.
Test for RAPL config and zone filtering.

codecov · 2025-04-28T04:34:32Z

Codecov Report

Attention: Patch coverage is 91.11111% with 4 lines in your changes missing coverage. Please review.

Project coverage is 93.35%. Comparing base (d6e67ac) to head (6d73db7).
Report is 4 commits behind head on reboot.

Files with missing lines	Patch %	Lines
internal/device/rapl_sysfs_power_meter.go	88.88%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           reboot    #2031      +/-   ##
==========================================
+ Coverage   93.26%   93.35%   +0.08%     
==========================================
  Files          21       21              
  Lines        1099     1143      +44     
==========================================
+ Hits         1025     1067      +42     
- Misses         58       60       +2     
  Partials       16       16

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

vprashar2929 · 2025-04-28T06:15:54Z

config/config_rapl_test.go

Shouldn't we keep all config related tests inside config.go ?

no, there is no hard requirement like that. It needs to be under the same package. Thats all

vprashar2929 · 2025-04-28T06:27:44Z

cmd/kepler/main.go

@@ -256,5 +256,18 @@ func createCPUMeter(logger *slog.Logger, cfg *config.Config) (device.CPUPowerMet
 	if fake := cfg.Dev.FakeCpuMeter; fake.Enabled {
 		return device.NewFakeCPUMeter(fake.Zones, device.WithFakeLogger(logger))
 	}
-	return device.NewCPUPowerMeter(cfg.Host.SysFS)
+
+	if !cfg.Rapl.Enabled {


What should be the behaviour when rapl is disabled?

Kepler should continue to monitor power of other meters. E.g. say there is platform power and gpu ...

When I disable rapl I still can see Kepler producing node metrics

Kepler should continue to monitor power of other meters. E.g. say there is platform power and GPU ...

For now it should only produce metrics like CPU info and build info right?

@vprashar2929, on second thought, I am removing enable.rapl flag for now since it requires other changes to monitor (more than the scope of this work). Lets revisit this when we have platform power as well.

vprashar2929 · 2025-04-28T09:38:04Z

hack/config.yaml

@@ -6,6 +6,10 @@ host:
  sysfs: /sys # Path to sysfs filesystem (default: /sys)
  procfs: /proc # Path to procfs filesystem (default: /proc)

+rapl:
+  enabled: false # Enable RAPL (default: true)


can we remove this?

we should set it true

config/config.go

vprashar2929 · 2025-04-28T12:19:21Z

hack/config.yaml

 # WARN DO NOT ENABLE THIS IN PRODUCTION - for development / testing only
 dev:
  fake-cpu-meter:
-    enabled: false
+    enabled: true


This should stay false

vprashar2929 · 2025-04-28T12:20:11Z

hack/config.yaml

@@ -6,8 +6,12 @@ host:
  sysfs: /sys # Path to sysfs filesystem (default: /sys)
  procfs: /proc # Path to procfs filesystem (default: /proc)

+rapl:
+  enabled: true # Enable RAPL (default: true)


We removed enable from rapl so no need to expose here until we add platform etc right?

vimalk78 · 2025-04-28T13:07:12Z

cmd/kepler/main.go

-	return device.NewCPUPowerMeter(cfg.Host.SysFS)
+
+	if len(cfg.Rapl.Zones) > 0 {
+		logger.Info("Filtering RAPL zones", "zones", cfg.Rapl.Zones)


cfg.Rapl.Zones are the zones we want to keep, not the ones we want to filter out.
better say it as enabled zones, like mentioned in comment below

# zones to be enabled, empty enables all default zones

comment was not just about t he log but about replacing the word filter with enable or select.
instead of device.WithZoneFilter we have device.WithSelectedZones or device.WithEnabledZones
instead of filterZones we say selectZones or enabledZones

but this is a minor comment. we can change it in a follow-up PR if needed.

vimalk78 · 2025-04-28T13:13:06Z

internal/device/rapl_sysfs_power_meter.go

+		return nil, fmt.Errorf("no RAPL zones found after filtering")
+	}
+
+	// filter out non-standard zones


what if a non-standard zone is configured.

non-standard zones only differ in path and not names. At least for the purpose of current implementation, non-standard only differ in its path like kepler_node_rapl_zone{index="0",name="package",path="/sys/class/powercap/intel-rapl-mmio:0"} 1

By that, it will be preserved / filtered in

Changes include: * Add RAPL configuration for filtering only required zones * Add zone filtering in raplPowerMeter; log excluded and included zones * Update createCPUMeter to use RAPL config and filter zones. * Test for RAPL config and zone filtering. Signed-off-by: Sunil Thaha <[email protected]>

sthaha requested a review from vimalk78 April 28, 2025 04:35

vprashar2929 added the feat A new feature or enhancement label Apr 28, 2025

vprashar2929 reviewed Apr 28, 2025

View reviewed changes

sthaha force-pushed the feat-filter-rapl-zones branch from 77fad70 to c337cdd Compare April 28, 2025 08:34

vprashar2929 reviewed Apr 28, 2025

View reviewed changes

config/config.go Show resolved Hide resolved

sthaha force-pushed the feat-filter-rapl-zones branch 2 times, most recently from 677bb3a to 9295f52 Compare April 28, 2025 11:26

vprashar2929 reviewed Apr 28, 2025

View reviewed changes

sthaha force-pushed the feat-filter-rapl-zones branch from 9295f52 to 072c902 Compare April 28, 2025 13:03

vimalk78 reviewed Apr 28, 2025

View reviewed changes

sthaha force-pushed the feat-filter-rapl-zones branch from 072c902 to 6d73db7 Compare April 28, 2025 14:19

vimalk78 approved these changes Apr 28, 2025

View reviewed changes

vimalk78 merged commit 34894b2 into sustainable-computing-io:reboot Apr 28, 2025
9 of 10 checks passed

feat: add RAPL zone filtering support through configuration #2031

feat: add RAPL zone filtering support through configuration #2031

Uh oh!

Conversation

sthaha commented Apr 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Apr 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

sthaha commented Apr 28, 2025 •

edited

Loading

codecov bot commented Apr 28, 2025 •

edited

Loading