CNTRLPLANE-3790

Conformance Test Suitability Analysis

Should e2e-azure-aks-ovn-conformance be promoted from informing to blocking in the ARO HCP release gating pipeline?

Analysis 1 of 2 for CNTRLPLANE-3790

CNTRLPLANE-3790 CNTRLPLANE-3434 (parent epic) Analysis 2: External OIDC test

Last updated: 2026-08-05 · Investigator: Alessandro Rossi

Context

Background

CNTRLPLANE-3790 tracks the expansion of the ARO HCP release gating pipeline's blocking test suite. The current pipeline (ITS source) has 9 blocking tests and 1 informing test.

The issue requires two separate investigations, each covered by its own report:

  1. This report (1 of 2): Should e2e-azure-aks-ovn-conformance be promoted from informing to blocking?
  2. Analysis 2 of 2: Why does e2e-azure-aks-external-oidc not exist for 4.22/5.0, and should it be added?
Current blocking tests (9):
  • e2e-aks (5.0)
  • hcm-azure-e2e-aks-upgrade-minor (5.0)
  • e2e-aks-multi-x-ax (5.0, 4.22, 4.21, 4.20)
  • hcm-azure-e2e-aks-upgrade-from-zero (4.22, 4.21, 4.20)
Current informing test (1):
  • e2e-azure-aks-ovn-conformance (5.0) ← subject of this analysis

Two conformance test variants exist:

JobScheduleScope
periodic-ci-openshift-hypershift-release-5.0-periodics-e2e-azure-aks-ovn-conformance Every 2 hours Standard OCP conformance suite, no test skips
periodic-ci-openshift-hypershift-release-5.0-periodics-e2e-azure-aks-ovn-conformance-fips Weekly (Sundays 4am) FIPS-enabled, narrower suite (openshift/conformance/parallel/minimal), skips etcd/azure-disk/azure-file/TestFIPS tests

The non-FIPS variant is the primary candidate for blocking due to its broader coverage and higher run frequency. This analysis focuses on determining whether its pass rate is high enough to gate promotion without causing excessive false-positive blocks.

Methodology

The analysis was performed using Sippy, the OpenShift CI health dashboard. Two views were consulted:

  • Job Health (ALL JOBS) -aggregated pass rate and run count over the current reporting period
  • Job Analysis -overall success/failure donut chart with job variant metadata

Data was collected on 2026-08-05. The reporting period covers the current and previous Sippy windows (exact dates determined by Sippy's rolling window configuration).

Results

Job Health: Pass Rate
Sippy Job Health view showing conformance test pass rates

Figure 1: Sippy Job Health for 5.0, filtered by aks-ovn-conformance. Source: Sippy Job Health

JobCurrent Pass RateRunsPrevious Pass RateRunsTrend
e2e-azure-aks-ovn-conformance 69.9% 249 65.4% 260 Improving
e2e-azure-aks-ovn-conformance-fips 100% 1 100% 1 Insufficient data
Key finding: The conformance test has a 69.9% pass rate over 249 runs. If added as a blocking test, it would block approximately 1 in 3 promotion attempts, regardless of the actual HO image quality.
Job Analysis: Overall Health
Sippy Job Analysis showing 509 runs, 344 successful

Figure 2: Sippy Job Analysis for e2e-azure-aks-ovn-conformance on release 5.0. Source: Sippy Job Analysis

Over a broader window, Sippy reports 509 total runs with 344 successful (67.6%), consistent with the current-period figure of 69.9%.

Job variant metadata: Platform:azure, Architecture:amd64, Network:ovn, NetworkStack:ipv4, Topology:external, Installer:hypershift, Suite:parallel, ContainerRuntime:crun, CGroupMode:v2, OS:rhcos10, JobTier:standard

FIPS Variant

The FIPS variant (e2e-azure-aks-ovn-conformance-fips) shows 100% pass rate, but with only 1 run in both the current and previous reporting periods. This sample size is far too small to draw any conclusions about reliability.

The FIPS variant runs weekly on Sundays, uses the narrower openshift/conformance/parallel/minimal suite, and skips several test categories. It would need significantly more run data before it could be considered as a blocking candidate.

Conclusion

Not suitable for blocking at this time

The e2e-azure-aks-ovn-conformance test is not currently reliable enough to be promoted to a blocking test in the ARO HCP release gating pipeline.

  • Pass rate too low: at 69.9% (249 runs), the test would block roughly 1 in 3 nightly promotions, creating significant noise and eroding trust in the pipeline
  • Trend is improving (up from 65.4%), but not yet at a level suitable for gating
  • FIPS variant has insufficient data (1 run) to evaluate
Recommendation: Keep e2e-azure-aks-ovn-conformance as an informing test. Re-evaluate when the pass rate consistently exceeds 90% over multiple reporting periods. The FIPS variant should also accumulate more run data before being considered.

Sources