Larger packet size than MTU causes packet drops observed. This is reproducible when we have TCP segment size is larger than negotiated MSS which is kicked in TSO. 

Test Result(Page 34) - CASE 2-4 - OVN - Pod to Pod in different nodes(Slide Page 34)
TCP Test - "without Packet length option."
+--------------+----------+
| Packets/sec* | TCP Retr |
+--------------+----------+
| 5,800        | 3        |
| 5,800        | 216      |
| 5,800        | 121      |
+--------------+----------+ 
** https://docs.google.com/presentation/d/1HbSAvuIf4gf0NGZyu59EawUHXSopVr2PKNA-T9bvC40/edit?usp=sharing

Negotiated MSS 8340, but TCP segment size larger than 58Kbytes(TSO kicks in). 

**Noted - this would be a minor contribution for packet drop because cluster MTU is big enough(pMTU 8500) for vCU cluster. 

[root@bastion ~]# oc get clusterversions.config.openshift.io 
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.12.4    True        False         10d     Cluster version is 4.12.4
[root@bastion ~]# 

iperf3 commands

1. bring up iPerf3 Server - iperf3 -s -p 60001
2. bring and client and issue the command - iperf3 -c 2001:feed:beef::301 -p 60001 -P 64 -b 30M -N -t 100 -V

Same as how reproducible.

No TCP transmission even TSO kicks in.