Re: How does BDR replicate changes among nodes in a BDR group

On 9 June 2017 at 02:47, Zhu, Joshua <jzhu(at)vormetric(dot)com> wrote:
> Thanks for the clarification.
>
> A follow up question, then, given *once joined all nodes are equal*, is that:
>
> should the node A dies or taken out of the group, the remaining three node group (with B, C and D) would continue to function properly, correct?

Yes. Though you'll need to tell them node A isn't coming back by
parting it from one of the other nodes at some point, otherwise
they'll accumulate WAL in pg_xlog and run out of disk space eventually
as they hold onto the resources A would need to sync back up.
Monitoring is important.

--
Craig Ringer http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services