"David G. Johnston" <david.g.johnston@gmail.com> writes:
> Admittedly, I'm not sure how you would fix any problems without access to
> the server and its config files - at which point you are back to simply
> reviewing those.
Yeah. Other related problems include being unable to *find* the log file
if you don't know what the server configuration is.
There is already a "postgres -C guc_name" option if you want to find out
from the command line what value a particular GUC is set to by the
cluster's configuration files. I could see some value in a variant of
that that prints all GUCs with non-default sources. But that would go
to stdout in any case. Wanting it to go into a log file sounds to me
a whole lot like wanting to duplicate some Oracle-based DBA habits at a
bug-compatible level.
(Note also that there's already logging of on-the-fly *changes* in
configuration file settings, so I'm not buying the "historical info"
angle at all.)
regards, tom lane