> Can I just do a bdr.bdr_group_join <http://bdr-project.org/docs/stable/functions-node-mgmt.html#FUNCTION-BDR-GROUP-JOIN>? Just want to confirm what the best practice is as I haven't seen anything in the documentation about this.
Yes, you can just do a bdr.bdr_group_join ( http://bdr-project.org/docs/stable/functions-node-mgmt.html) on the new node, specifying the dsn of either existing node, just like you did when creating the first node. Alternately, you can use bdr_init_copy ( http://bdr-project.org/docs/stable/command-bdr-init-copy.html) to do a clone of the whole database installation and bring it up as a new peer. Note that this will copy other non-BDR databases too, since it does a pg_basebackup and pg_basebackup doesn't have any facility for including/excluding individual databases. This can be a better option for a big database. In either case if the new node join fails for any reason you must make sure to manually remove the replication slots created on either/both existing nodes to prevent WAL from accumulating until they run out of disk space. On 16 January 2016 at 02:45, Cj B <blackc2...@gmail.com> wrote: > Hi there, I want to add another node to my BDR group and was wondering > what the best practice is. My database is about 4gb. Current servers on > west coast and new server is on east coast. > > Can I just do a bdr.bdr_group_join > <http://bdr-project.org/docs/stable/functions-node-mgmt.html#FUNCTION-BDR-GROUP-JOIN>? > Just want to confirm what the best practice is as I haven't seen anything > in the documentation about this. > -- Craig Ringer http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
