The test [sig-network] [Feature:IPv6DualStack] should have ipv4 and ipv6 node podCIDRs [LinuxOnly] incorrectly assumes that all clusters (except Windows ones) use --allocate-node-cidrs, and so fails under OCP where we do not use that. What the test probably should do is, *if* the single-stack podCIDR field is set on a node, then require that podCIDRs is also set, and correct. (And if podCIDR is unset, then podCIDRs should be unset too.) And then the `[LinuxOnly]` can be removed because it should never have been there in the first place; they were just working around the bug incorrectly.
That test should have moved out of the e2e framework since podCIDR is not mandatory. I rather prefer to remove it than to add more conditions to the test ... and move it as an integration or unit test in the controller manager if we don't want to loose coverage.
For the sake of tracking, here's the link to the discussion opened by Antonio in upstream kubernetes on this issue: https://github.com/kubernetes/kubernetes/issues/104544
The test was removed upstream (https://github.com/kubernetes/kubernetes/pull/105079) so the skip rule can be removed once origin is rebased to kube 1.23.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Important: OpenShift Container Platform 4.10.31 security update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2022:6258