Need Advice on below issue

Issue :-

From the Source OCP cluster, they connect to S3 storage server which is behind LB [5nodes in pool]. [OCP --> Firewall --> LB --> S3 storage. All are on premises]

They create/read/write/delete files in large volumes from OCP to S3 in one go. When this cycle increases they see timeout issue.

Troubleshooting:-

They bypassed the LB in Non-Prod and tried to test the flows with one S3 node with high volume and it was successful. Same test were performed in PROD too and that too were successful.

Meanwhile from TCP dump analysis output we found “tcp zero window” is getting timed out. Initially it was set to 20k Milliseconds, they increased to 60k and later increased to 300k millisecond, still test were failing in Non-Prod when LB was introduced.

x.x.33.11 is VIP IP and x.x.239.226 is OCP IP

As of now, LB is bypassed in PROD and all cycles are running smoothly. Everyone is suspecting that it's an F5 issues. We don't have any evidence to prove ourself. Is there any tuning required on F5 VIP or anything else which could solve this ?

Thankyou