This content has been marked as final. Show 2 replies
Later in the past i've configured a Wildfly 9 version as a load-balancer with an Undertow based configuration (Stuart Douglas video) and acordingly to documentation I had to configure the worker nodes with unique identifiers in order to get session stickiness. That worked.
Now I'm using the same architecture with Wildfly 10.1 (Undertow 3.1), but the Load Balancer act always in a round robin fashion.
In Load Balancer log nothing goes wrong, since I cant get into undertow algorithm logic.
The main difference I've found is that every time the client request came to the worker, the http session is different. (I assume that this is because of the round robin balancing, because each request points to a diferente worker every time).
What changed from 9 to 10.1 version?