Can you share a config that shows the "implicit monitor fail" behavior and the haproxy version it happens on?
I've tried the config below on a couple of 1.4 and 1.5 versions and they behave the same for me. All return 200 unless the explicit "monitor fail" directive is present. global defaults timeout client 5s timeout server 5s timeout connect 5s mode http listen one bind 127.0.0.1:5001 monitor-uri /admin?site_check acl dead nbsrv(one) lt 1 # monitor fail if dead server 127.0.0.2 127.0.0.2 check On Thu, Nov 13, 2014 at 2:57 AM, JCM <[email protected]> wrote: > Hi all - > > What's the default behaviour supposed to be when a monitor-uri path is > hit, and an HTTP backend (or listener) knows that it has no healthy > servers? > > I'm positive I've previously seen haproxy doing The Right Thing (i.e. > 503'ing the monitor-uri path if all servers are down), but I'm finding > that this isn't the case with 1.5.8. > > Did the default behaviour change from 1.4->1.5? I'm sure it was using > 1.4 that the behaviour seemed to be as if there were an implicit > "monitor fail if nbsrv($backend) lt 1" present, but I have to > explicitly add that in 1.5(.8) which seems ... ungood. > > What am I missing? > > Cheers, > Jonathan > >
