Andres also suggested we try source based balancing rather than roundrobin. This should lead to all the requests from a given node going to the same server, which makes it much easier to debug. Right now, requests from the same client can go to any server, which makes reading logs very hard.
Andres also suggested we try source based balancing rather than roundrobin. This should lead to all the requests from a given node going to the same server, which makes it much easier to debug. Right now, requests from the same client can go to any server, which makes reading logs very hard.
https:/ /www.haproxy. com/blog/ client- ip-persistence- or-source- ip-hash- load-balancing/