Robert Treat <xzilla@users.sourceforge.net> writes:
> I have to add, given that we already provide the time of last checkpoint
> information via pg_controldata, I don't understand why people are against
> making that information accesible to remote clients.
So, I can expect to see a patch next week that invents a separate SQL
function to expose each field printed by pg_controldata?
You are failing to make the distinction between debug/developer support
(which is what pg_controldata is) and things that are interesting for
DBAs to track. Would you expect someone to install handwritten
monitoring infrastructure to awaken them in the middle of the night when
pg_control_version changes?
I am not saying that there is no use for some sort of facility in this
area. But I think it needs some more-general thinking and design than
has been evidenced so far in this thread.
regards, tom lane
