at the moment we are running an active/standby reverse proxy installation.
one virtual IP by keepalived.

it works great thanks to NGINX, but has the disantvantage of having an empty cache on the standby proxy, so in case of a failover the cache must first be filled up, which slows the proxy massively down.

i'm now thinking about adding the standby proxy the list of origin servers, so i get something like this

proxy01 -> origin01
-> origin02
-> proxy02 -> origin01
-> origin02

this would make shure that proxy02's cache is filled to

will this make sense? can i configure proxy02 so he would directly answer (direct server return)? or do i really need an loadbalancer in front of the two ngingx proxys?