Description of problem:

Version-Release number of selected component (if applicable):

How reproducible:

Steps to Reproduce:

1. set ipv6.disabled=1 on 4.14+ cluster

Actual results:

OVN-kuberetes pods go into crashloop backoff

Expected results:

Not crash, only ipv6 is disabled.

Additional info:

An alternative to fixing the root issue is to add an upgrade blocker in the network operator if any nodes have ipv6 disabled.

Affected Platforms:

Customer Environment during the upgrade from 4.13 > 4.14 Extremely difficult to fix due to the circular dependency between MCP and CNI. After CNI is broken the karg must be manually removed so the CNI can run, But the MCP cannot roll forward because there is an on-disk conflict, adding the karg back causes the cni to fail to start.