I have istio currently configured with ConsistentHash load balancing via headerName and the session affinity works for single synchronous calls. The issue I have now is when one long running HTTP call is made, and then 2 additional HTTP calls are made while the first is still processing, the 2 additional calls don’t go to the same backend pod until the long running call completes. Seems like session affinity fails under this scenario. Has anyone ran into this before?
This was an application side issue not related to istio/envoy at all…