* Andres Freund (andres@2ndquadrant.com) wrote:
> On 2014-10-09 09:44:09 -0400, Tom Lane wrote:
> > Stephen Frost <sfrost@snowman.net> writes:
> > > Yeah, I agree with this- it's extremely useful information and it's
> > > really not that verbose in general..
> >
> > -1. Every time we've turned on default logging of routine events,
> > there's been pushback and it was eventually turned off again as log spam.
> > I guarantee you that logging checkpoints will be seen as log spam, except
> > by people who are actually having trouble with that behavior, which is
> > a small minority.
>
> We're talking about 2 log message per checkpoint_timeout interval
> here. That's pretty darn far away from log spam. Was there really any
> case of such low frequency message causing ire?
For embedded devices and similar small-scale systems, I can see Tom's
point. At the same time, I would expect those to require sufficient
configuration that also setting log_checkpoints to 'off' wouldn't be a
huge deal.
> And if it's more frequent you can be happy that you see the log message
> - because your config isn't appropriate for your load. The number of
> times I've seen people being baffled at the bad performance because of a
> inadequate checkpoint configuration, whose effect they couldn't see, is
> baffling.
I certainly agree with this, but then, look at our default
log_line_prefix and other default log settings.. In general, our
defaults are horrible and fixing this one is really just the tip of the
iceburg..
Thanks,
Stephen
