Magnus Hagander <magnus@hagander.net> writes:
> On Thu, Aug 31, 2017 at 2:34 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Right, because the decision whether to redirect stdout/stderr can't
>> be changed on the fly.
> Right.
> We could of course also say we only care about things generated by our
> ereport framework, in which case we don't need to redirect stderr and can
> just use a "regular pipe". But IIRC that's functionality we also
> specifically wanted (though I don't think I've ever needed it myself,
> presumably others have).
Yes, it's pretty important, because of assorted stuff not-under-our-
control that doesn't know about ereport and will just print to stderr
anyway. Some examples: dynamic linker can't-resolve-symbol failure
messages, glibc malloc corruption error messages, just about any external
module in plperl or plpython. I don't find this to be negotiable.
> Are you actually asking for a benchmark of if logging gets slower?
Yes.
> If so,
> could you suggest a workload to make an actual benchmark of it (where
> logging would be high enough that it could be come a bottleneck -- and not
> writing the log data to disk, but the actual logging). I'm not sure what a
> good one would be.
pgbench with log_statement = all would be a pretty easy test case.
regards, tom lane