test(perf): raise performance thresholds to absorb LatencyHarness overhead - #3245
Open
nathangeology wants to merge 2 commits into
Open
test(perf): raise performance thresholds to absorb LatencyHarness overhead#3245nathangeology wants to merge 2 commits into
nathangeology wants to merge 2 commits into
Conversation
Absorbs LatencyHarness (750de8d) scrape overhead. Previous 260 MB was captured before the harness landed and reports > 260 MB post-harness. Bumping basic/consolidation avg CPU base 0.25 -> 0.40 for the same reason. Regression detection retained: a 50-percent regression still trips. Signed-off-by: Nathaniel Jones <jonesflp@amazon.com>
Absorbs LatencyHarness (750de8d) scrape overhead. Observed avg CPU is at the previous 0.30 base with no headroom. 0.45 retains detection of a 50-percent regression. Fixes a comment/constant mismatch on scaleOut TotalTime (5 min constant, message said 10 min). Signed-off-by: Nathaniel Jones <jonesflp@amazon.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: nathangeology The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
kubernetes-prow
Bot
added
cncf-cla: yes
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #N/A
Description
The
basicandwideDeploymentsperformance suites carry inline base thresholds that predate the LatencyHarness metric scraping added in the perf-reporting improvements track. The harness adds observable overhead to the captured window (roughly 0.10 cores average CPU and 30–60 MB P95 memory), which is enough to push three thresholds over their historical caps on a normal run.This change raises three base thresholds and corrects one assertion message. Regression sensitivity is retained: the new values keep any 50-percent-or-greater regression detectable.
test/suites/performance/basic_test.go:basic/scaleOutMemoryThreshold 260 → 320 MBbasic/consolidationMemoryThreshold 260 → 320 MBbasic/consolidationCPUThreshold 0.25 → 0.40 corestest/suites/performance/wide_deployments_test.go:wideDeployments/consolidationCPUThreshold 0.30 → 0.45 coreswideDeployments/scaleOutTotalTime assertion message: "less than 10 minutes" → "less than 5 minutes" to match the 5-minute constantAll other thresholds in these files are left as-is (not observed to breach). Provider-specific overrides continue to work through the
KARPENTER_PERF_THRESHOLDSenv var from #3165.How was this change tested?
make verifypasses locally.go vet ./test/suites/performance/andgo test -c ./test/suites/performance/succeed on the rebased branch. The perf assertions themselves run under kind-perf-e2e and cannot be exercised outside that harness; the values were chosen to leave headroom over recently observed samples rather than from a full re-baseline capture.AI Disclosure
I used an LLM to parallel program with me on this. That included drafting the initial commit messages and PR description, and proposing the specific threshold values from a review of the observed run samples. Each value was accepted by me after checking it against the historical base and the LatencyHarness overhead.
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.