Customer asking for assistance in understanding how Istio implements tcp load balancing - in particular, they are looking to use an Istio DestinationalRule with the load balance setting of "LEAST_CONN" (deprecated) or "LEAST_REQUEST" to ensure that TCP connections are evenly distributed across a large number of pods. They are observing that only 4-5 pods are receiving connections, and thus asserting that there may be a bug. These are long lived TCP connections.

They were also experiencing load balancing challenges before adding Istio, so there may be other factors impacting how traffic is routed outside of Istio, but we should review the case to see if we can explain the behaviour seen.

See the case for details, including must gather.

It sounds like they are still on OSSM 2.3, which is out of support and have been asked to update.