The OCP cluster is running with IPsec for east west traffic, it works until cluster is rebooted. After reboot is complete pod to pod traffic is completely broken. 

We checked all the IPsec config, It is as similar as before cluster reboot, didn't notice any changes to xfrm states and policies.

When we reboot the cluster by setting esp-tx-csum-hw-offload to 'off' on all nodes, then cluster is restored.

Cluster Profile: 1_UPI on vSphere 8.0& FIPS ON & OVN IPSEC & Static Network & Bonding & HW19 & Secureboot (IPSEC E-W only)

For more information refer to discussions below:
https://redhat-internal.slack.com/archives/C04L7QWC9CZ/p1705663798680819
https://redhat-internal.slack.com/archives/C05NV4G1W9Y/p1705604505159779
https://issues.redhat.com/browse/OCPBUGS-25312