Summary

Since Feb 12, the capi-controllers and capi-operator pods crash-loop on every run of periodic-ci-openshift-release-main-nightly-4.22-e2e-metal-ipi-ovn-ipv6-techpreview. This causes a cascade of failures:

• ClusterOperatorDegraded alert fires for ~1.5h
• Pathological Back-off restarting events detected
• Cluster stability checks fail
• Operators reported as not on the cluster version

18 consecutive failures since Feb 12 with 0 passes. The last passing run was Feb 10.

This is NOT a recurrence of the same code change as OCPBUGS-74161 (cluster-api-provider-metal3#57 was reverted Jan 21 and no new PRs have merged in that repo since). The root cause is different.

Error Output

All failures show the same pattern:

32 events happened too frequently
event [namespace/openshift-cluster-api pod/capi-controllers-* - Back-off restarting failed container]
event [namespace/openshift-cluster-api-operator pod/capi-operator-* - Back-off restarting failed container]

ClusterOperatorDegraded alert:

ClusterOperatorDegraded was at or above info for at least 1h29m on metal/amd64/ovn/ha (maxAllowed=1s): firing for 1h29m

Affected Variant

• Platform: metal
• Network: ovn (IPv6)
• FeatureSet: techpreview
• Topology: ha
• Architecture: amd64
• Installer: ipi
• Upgrade: none

Regression Details

Timeline

• Last passing run: 2026-02-10T20:21:33Z
• First failing run: 2026-02-12T10:58:08Z
• First failing payload: 4.22.0-0.nightly-2026-02-12-105448
• Regression opened in Component Readiness: 2026-02-14 to 2026-02-15

Payload Analysis

Payload 4.22.0-0.nightly-2026-02-12-105448 contained 9 new PRs. None directly modify cluster-api-provider-metal3. No new PRs have been merged in openshift/cluster-api-provider-metal3 since the Jan 21 revert (PR #60).

Related

• OCPBUGS-74161 - Previous occurrence caused by cluster-api-provider-metal3#57 (reverted). This is a different root cause.
• Sample failed run with 32 pathological events