Re: should we enable log_checkpoints out of the box?

On Sun, Oct 31, 2021 at 10:59:19AM -0400, Tom Lane wrote:
> Bharath Rupireddy <bharath(dot)rupireddyforpostgres(at)gmail(dot)com> writes:
> > How about we enable it out of the box?
>
> No.
>
> The general policy at the moment is that a normally-functioning server
> should emit *no* log traffic by default (other than a few messages
> at startup and shutdown). log_checkpoints is a particularly poor
> candidate for an exception to that policy, because it would produce so
> much traffic. No DBA would be likely to consider it as anything but
> log spam.

Huh? As I see it, this is something that everyone should look at, at least
sometimes, and may be particularly important when deploying a new instance. In
order to set shared_buffers/checkpoint_timeout/wal_size, it's important to know
what fraction of the buffers were dirty during a typical/peak checkpoint. And
important to know if fsync is taking unreasonably large amount of time.

In cases that log_checkpoints writes "so much", then checkpoint_warning would
also be complaining, already.

I think writing a low volume of "routine" messages would allow some confidence
that a server is functioning normally. If there's nothing being logged at all,
maybe logging isn't working, maybe I'm looking in the wrong place, or maybe the
server hasn't completed a checkpoint in the last 3 weeks and we're in deep
trouble - the absence of any logs at all does not support high confidence in
the health of one's cluster.

Note my somewhat recent message at
https://www.postgresql.org/message-id/20210615161830.GQ31772@telsasoft.com:

On Tue, Jun 15, 2021 at 11:18:30AM -0500, Justin Pryzby wrote:
> I propose to change some defaults:
>
> log_autovacuum_min_duration = 0
> log_checkpoints = on
> log_lock_waits = on (and log_recovery_conflict_waits too?)
> log_temp_files = 128kB

Robert opined that autovacuum shouldn't be zero, but shouldn't be disabled
either, which is fine. Any value that doesn't completely disable logging these
events would be preferable - same for log_temp_files.