Josh Berkus <josh@agliodbs.com> writes:
> Vik's feature would be useful for detecting an accidental split in a
> replication cluster. That is, it would be another tool for detecting if
> you've made a mistake and created two masters. So +1 from me.
We don't change the ID when promoting a slave to master, do we?
So how would this help for that?
> However, given that the value is the same for all servers in a
> replication set, are we sure we want to call it system_identifier? Is
> there a better name?
I think there are definitely use cases for a system identifier of some
kind, I'm just not sure that what we have got right now is especially
useful to users. I'd rather see some thought go into what's needed
before we expose the existing definition (and consume the valuable
namespace of "pg_system_identifier").
regards, tom lane