On Sunday, January 10, 2016, Rodrick Brown <rodr...@orchard-app.com> wrote:
> We run 100% on AWS and have been running Mesos in production since version > 0.19 > Our cluster consists of 3 dedicated zookeeper nodes (M3.2lx), 3 dedicated > masters (M3.2lx), 8 dedicated slaves (M4.4xl) and 2 haproxy (M4.Medium) > instances used in conjunction with marathon-lb for routing requests into > backend services running on Mesos. > Question. You have 2 haproxy nodes also running marathon-lb, or 2 haproxy nodes with 2 more for marathon-lb? Do you use something like keepalived to provide some sort of a vip, or just cheat and use ELB to balance to your haproxy nodes? Thanks! -- Text by Jeff, typos by iPhone