It was seen in downstream and upstream that ovn-controller was constantly restarting. This was due to ovnkube-node telling it to exit after it thought that the encap IP (the primary node IP) had changed.

This has been mitigated by:
https://github.com/ovn-org/ovn-kubernetes/pull/3711

But we still need to know why the c.nodePrimaryAddrChanged() function is returning true when nothing is really changing on the node. Example after the fix above:

ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 22:37:02.020612    1670 node_ip_handler_linux.go:212] Node primary address changed to 172.18.0.3. Updating OVN encap IP.
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 22:37:02.037852    1670 node_ip_handler_linux.go:343] Will not update encap IP, value: 172.18.0.3 is the already configured
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:03:03.115881   16698 node_ip_handler_linux.go:212] Node primary address changed to 172.18.0.3. Updating OVN encap IP.
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:03:03.122365   16698 node_ip_handler_linux.go:343] Will not update encap IP, value: 172.18.0.3 is the already configured
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:18:08.381694   27220 node_ip_handler_linux.go:212] Node primary address changed to 172.18.0.3. Updating OVN encap IP.
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:18:08.389655   27220 node_ip_handler_linux.go:343] Will not update encap IP, value: 172.18.0.3 is the already configured
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:19:26.638221   28746 node_ip_handler_linux.go:212] Node primary address changed to 172.18.0.3. Updating OVN encap IP.
ovn-control-plane/ovn-kubernetes/ovnkube.log:I0627 23:19:26.644217   28746 node_ip_handler_linux.go:343] Will not update encap IP, value: 172.18.0.3 is the already configured

This can be observed in kind deployments as well.

Could affect versions earlier than 4.14