I have HA proxy load balancing across 3 web servers. The session limit is set to 200 and I am seeing the "Curr sessions" hit the limit. Am I losing traffic or are the additional requests queued up and served later?
I know this is not ideal, I am going to bump it up but I am curious to know what is actually happening.
My application behind the load balancer does not depend on sticky sessions.
According to the Documentation at 1wt.eu for version 1.3 it will queue the additional requests.
But only if there are less than
maxqueue
requests in the queue already, fortimeout queue
seconds, if after that timeout no unsaturated server is found, the request will be dropped.More Info see here: HAProxy Doc and here: Sample Config, as well as here: Article about queuing in HAProxy