@selea @maswan HAProxy is taking all the CPUs (4 in a vm) and doing around 1200 requests/sec, the backend can locally do 23k req/sec. If I enable caching in HA, then I can push to max 2700-2800 req/sec. I am wondering if I can push it more.