14:51:41 GMT <gerl1ng> Hello, is there any option to not only do an automatic failover in a redis cluster but also allow slaves to do an automatic takeover? It seems that this is a manual process / needs to be done by a 3rd party control-unit.
22:08:48 GMT <nDuff> Huh. Trying to start ''redis-server --slaveof "[$ipv6_addr]" 16379'' is leading to a cycle of "* Connecting to MASTER [...]" and "# Unable to connect to MASTER: Success"
22:11:38 GMT <nDuff> ...looks like it's trying to do a DNS lookup, so presumably not recognizing that syntax for an IPv6 address...
22:34:36 GMT <nDuff> ...taking out the square brackets, it looks like we're creating an AF_INET6 socket, but then binding it locally to 127.0.0.1, and getting EINVAL... upgrading to 4.0.1, it binds to ::1 and gets EINPROGRESS instead, but still never actually connects.
22:47:06 GMT <nDuff> ah HAH
22:47:21 GMT <nDuff> bind address for replication needs to be the first one given to the bind directive
22:47:46 GMT <*> nDuff  wonders if this is documented and he missed it.