Hi Zack, The peers in HAProxy are available only for synchronizing data content from stick tables. If you want a failover to occur you must use a tool to force an IP failover, like keepalived's vrrp ability.
Baptiste On Thu, May 9, 2013 at 9:56 PM, Connelly, Zachary (CGI Federal) <zachary.conne...@cgifederal.com> wrote: > HAProxy list, > > > > I wanted to confirm what I’ve read in the documentation and on other posts > concerning HAProxy and failover using the Peer functionality. HAProxy does > not automatically failover for you - if two HAProxy instances are linked as > peers, if an instance go down (or a frontend and/or backend is purposefully > brought down via command line), HAProxy will not automatically start routing > traffic to the other functioning HAProxy instance. It is up to a DNS or some > other layer to send traffic to the second HAProxy instance, but the second > instance will at least have the updated stick-table info from the first > instance to use when determining routing of traffic. Is that how I should > understand the peer functionality or am I just not configuring it correctly? > > > > Thanks (as usual), > > Zack Connelly